Divided Agencies

Clashes between presidential appointees and civil servants are front-page news. Whether styled as a “deep state” hostile to its democratically selected political principals or as bold “resisters” countering those principals’ ultra vires proposals, accounts of civil servant opposition are legion. Move beyond headlines, however, and little is known about the impact of political divisions within agencies on their workaday functioning.

This Article presents the first comprehensive, empirical examination of the effects of intra-agency political dynamics on policymaking. Leveraging data on political preferences based on campaign donations, we identify “ideological scores” for both appointees and civil servants in dozens of agencies over thirty-four years—the first measure of the political gap between these two groups across agencies and time. We use these scores to examine how ideological divergence between appointees and civil servants affects regulatory activity.

We find that agencies with greater distance between these two groups—which we term “divided agencies”—may adopt a more cautious posture. They tend to extend the rulemaking process and allow consideration of late-filed comments. These features provide appointees with extra time to gather and digest comments from politically aligned outside experts. Divided agencies’ caution may extend to the completion of final rules, which—in some but not all models—tend to be less numerous. Remarkably, we find no evidence that divided agencies are any less successful in shepherding proposed rules to final status. That finding casts doubt on the claim that the longer rulemaking timeframes in these agencies are attributable to civil servants’ attempts to derail oppositional appointees’ initiatives. Instead, one possible interpretation is that divided agencies’ caution pays off.

These findings imply that, with agency heads oscillating between left and right based on the party in power, the generally more moderate civil service can serve as a ballast. Specifically, faced with appointees that may be responsive only to a bare electoral majority, the presence of oppositional civil servants may encourage regulatory caution and push decision-making away from the extremes—thus, paradoxically, moving policy toward the median voter.

Our findings also spotlight the critical role that the notice-and-comment process—which is often maligned as pretextual—can play in divided agencies. Generalist appointees face a principal-agent problem when crafting rules: their key source of necessary in-house expertise, civil servants, may be misaligned. In this circumstance, comments from outside allies can provide a check on civil servants’ work. That civil servants can play a promajoritarian, moderating role in divided agencies highlights the importance of preserving civil service protections—especially in today’s polarized political climate.


Secretary of the Interior Ryan Zinke, who served during the Trump Administration, and John Morton, who helmed Immigration and Customs Enforcement (“ICE”) under President Obama, may not have much in common politically, but they do share one experience: they managed agencies in which approximately one-third of their workforce was estranged. A proponent of increasing industry access to public lands, Secretary Zinke believed he had “thirty [percent] of the crew that’s not loyal to the flag” concerning that goal.[1] He compared his situation to capturing “a prized ship at sea and only the captain”—that would be Secretary Zinke, incidentally, a former Navy SEAL—“and the first mate row over” to manage the captured crew.[2] In response, some Interior Department civil servants styled themselves “the disloyals,” printing T-shirts with that epithet.[3]

Director Morton faced a similar mutiny. After issuing a directive prioritizing deportations of people convicted of crimes and urging prosecutorial discretion in other cases,[4] the union representing nearly thirty-nine percent of ICE employees passed a no-confidence vote against Morton’s leadership.[5] That move was unprecedented.[6]

That other apostates can be found across the executive branch is unsurprising;[7] the conditions are ripe for such conflicts. Civil servants often hold differing views from appointees.[8] With only four thousand appointees atop a federal workforce of over two million[9]—many of whom hold job protections—the former group’s ability to supervise the latter will, by practical necessity, be incomplete. As political polarization grows and hardball tactics typically associated with electoral politics enter administrative agencies,[10] we expect that conflicts between appointees and civil servants will only increase.

In recent years, legal scholars have turned their attention to examining these inner workings of administrative agencies. For instance, some scholars posit that competing centers of power within agencies—civil servants and appointees, along with public participants—serve a checking function on each other’s power and thus mimic the more familiar constitutional separation of powers.[11] Others theorize about the policies produced by agencies that contain competing powers, some of which pull in majoritarian and others in countermajoritarian directions.[12]

Yet while the legislative consequences of political divisions among the branches of government are well studied,[13] relatively little empirical work analyzes the impact on policy of political divisions within agencies.[14] Empirically, political dynamics inside administrative agencies remain terra incognita in some important respects. How do agencies in which key subgroups are at loggerheads differ from agencies that are more politically cohesive? Do deeply divided agencies take longer to regulate, perhaps because of distrust or civil servant foot-dragging? Is White House review more exacting for these agencies, on the theory that White House officials are less likely to trust proposed rules emanating from ideologically divided entities? And do these agencies ultimately produce fewer rules?

This Article seeks answers to these questions. It examines how ideological differences between political appointees and civil servants affect the rulemaking process. These two groups share power within agencies, with generalist appointees relying on expert civil servants to implement the former group’s preferred policies. That division gives rise to a well-studied principal-agent problem: appointees must rely on civil servants who may have very different policy preferences and over whom appointees have limited ability to monitor or control.[15]

Faced with agents they may distrust, appointees may seek out and spend more time considering informed “second opinions” from other sources. These alternative sources of information include comments received during the notice-and-comment process, informal feedback from allies in Congress, and recommendations from advisory committees of outside experts occupying a privileged position within agencies. Indeed, public choice theorists posit that administrative structures and processes can serve just this purpose.[16]

We put this theory to the test, examining how appointees respond when their agents in the civil service hold differing views. To do so, we first develop a measure of ideological distance over time and within agencies so that we can identify divided agencies.

Existing measures are inadequate for that purpose,[17] so we create our own. We leverage a dataset on ideological preferences based on campaign donations to do so. We use these data to generate dynamic “ideal point” estimates for agency heads and civil servants in forty-seven agencies over thirty-four years—and thus, a new measure of the ideological gap between these two groups across agencies and time.[18] We then connect this measure to data concerning the rulemaking process.

Our results show that divided agencies—that is, those with ideologically opposed agency heads and civil servants—adopt a slower rulemaking posture than agencies that are more unified. Several of our findings suggest that greater caution may be at play. Once civil servants generate a proposed rule, appointees take their time. While we cannot rule out all alternative explanations, we observe that one feature of the delay is consideration of late-filed comments. Considering late-filed comments allows appointees to hear from a greater number of ideologically aligned outside groups as a check on civil servants’ work. Delay may also result from appointees spending additional time assessing those comments. In either case, slower rulemaking at divided agencies suggests that appointees may be utilizing rulemaking procedures to blunt civil servants’ informational advantages. Additionally, divided agencies may tend to issue fewer rules. That their rules are no less likely to become final, however, is perhaps evidence that their caution pays off.

This claimed cautious approach means that, whatever policy changes one desires in a first-best world, the reality of policymaking in divided agencies likely will leave one disappointed. Indeed, divided agencies are likely status quo-preserving. Whether this feature is normatively desirable turns, in part, on one’s risk aversion and the extent to which one values policy certainty.

Given that partisan polarization—and thus divided agencies—likely will persist into the foreseeable future, our findings provide a set of best practices for agencies to function as well as possible under these conditions. The policy implication that most closely follows from our findings is that officials must preserve the independence of the civil service. At a time when that independence is challenged, our findings about rulemaking suggest that civil servants comprise a moderating counterweight against more ideologically extreme appointees; thus, they serve as a bulwark against wild changes in regulatory policy. With agency leadership swinging between liberal and conservative poles, as we find, civil servants—who tend to be more moderate, albeit left of center—can pull agency policies toward the median voter. This moderation serves to improve democratic representation in agency policymaking: appointees are aligned with the Presidents who appoint them, and Presidents tend to be more ideologically extreme than the median voter. Allowing policy to swing all the way to their appointees’ preferences would therefore not reflect the public’s preferences. In contrast to common laments of employment-protected civil servants serving as a countermajoritarian force in policymaking, we show that they can serve a democratizing function in divided agencies.[19]

Further, to prevent divided agencies from descending into the gridlock and paralysis that plague other polarized institutions, appointees must have access to high-quality information from ideological allies, which we infer from divided agencies’ greater willingness to consider late-filed comments. We argue that the notice-and-comment process is well suited to transferring high-quality information to distrustful appointees. Notice-and-comment also may discourage civil servants, aware that their work will be “checked” by outsiders, from straying too far from their principals’ goals. Additional measures to inject diverse outside sources of information into agency decision-making could further enhance agencies’ ability to function, even in a challenging partisan climate within their walls—though they would increase resource costs associated with rulemaking.

This Article proceeds in four parts. Part I situates our study in twin literatures: empirical scholarship examining extra-agency influences on regulatory dynamics and descriptive and positive work concerning intra-agency dynamics. Part II presents our theory and expectations concerning the effects of appointee-civil servant preference divergence on regulatory processes and outputs. In Part III, we describe our research design, including our creation of an original dataset identifying appointees’ and civil servants’ political ideologies across agencies and time, and we present our analysis. Part IV discusses normative implications and offers policy prescriptions.

          [1].      Evan Osnos, Trump vs. the “Deep State, New Yorker (May 14, 2018), https://
http://www.newyorker.com/magazine/2018/05/21/trump-vs-the-deep-state [https://perma.cc/9862-ZBGM].

          [2].      Matthew Daly, Interior Chief’s Loyalty Comments Draw Widespread Criticism, Associated Press (Sept. 26, 2017), https://apnews.com/article/8c3ae77664f44159823903b3add31e65 [https://

          [3].      Osnos, supra note 1.

          [4].      Memorandum from John Morton, Dir., U.S. Immigr. & Customs Enf’t, to All Field Off. Dirs., All Special Agents in Charge, & All Chief Couns., U.S. Immigr. & Customs Enf’t (June 17, 2011), https://
http://www.ice.gov/doclib/foia/prosecutorial-discretion/certain-victims-witnesses-plaintiffs.pdf [https://perma.

          [5].      Ted Hesson, 7 Numbers that Tell the Story of an Immigration Boss’s Tenure, ABC News (June 17, 2013, 12:34 PM), https://abcnews.go.com/ABC_Univision/Politics/ice-director-john-mortons-
tenure-numbers/story?id=19422159 [https://perma.cc/W37D-R4QM]; see also Julia Preston, Single-Minded Mission to Block an Immigration Bill, N.Y. Times (June 1, 2013), https://www.nytimes
.com/2013/06/02/us/for-chris-crane-a-quest-to-block-an-immigration-bill.html [https://perma.cc/2ZLK-
TXUC] (providing figures used to calculate the union’s share of ICE’s workforce).

          [6].      Preston, supra note 5.

          [7].      See Osnos, supra note 1 (providing other examples).

          [8].      See infra Part III.

          [9].      Fiona Hill, Public Service and the Federal Government, Brookings (May 27, 2020), https://
http://www.brookings.edu/policy2020/votervital/public-service-and-the-federal-government [https://perma.cc/
JRK2-QYRM] (reporting the size of the federal nonmilitary, nonpostal workforce and the approximate number of political appointees).

        [10].      See Brian D. Feinstein & M. Todd Henderson, Congress’s Commissioners: Former Hill Staffers at the S.E.C. and Other Independent Regulatory Commissions, 38 Yale J. on Regul. 175, 223, 226 (2021) (documenting these developments).

        [11].      See Jon D. Michaels, Of Constitutional Custodians and Regulatory Rivals: An Account of the Old and New Separation of Powers, 91 N.Y.U. L. Rev. 227, 238–39 (2016); Gillian E. Metzger, The Interdependent Relationship Between Internal and External Separation of Powers, 59 Emory L.J. 423, 425 (2009); Neal Kumar Katyal, Internal Separation of Powers: Checking Today’s Most Dangerous Branch from Within, 115 Yale L.J. 2314, 2346 (2006).

        [12].      See Matthew C. Stephenson, Optimal Political Control of the Bureaucracy, 107 Mich. L. Rev. 53, 72 (2008).

        [13].      See generally Daryl J. Levinson & Richard H. Pildes, Separation of Parties, Not Powers, 119 Harv. L. Rev. 2311 (2006); Gary W. Cox & Mathew D. McCubbins, Setting the Agenda: Responsible Party Government in the U.S. House of Representatives (2005); John J. Coleman, Unified Government, Divided Government, and Party Responsiveness, 93 Am. Pol. Sci. Rev. 821 (1999); David R. Mayhew, Divided We Govern: Party Control, Lawmaking, and Investigations, 1946–2002 (2d ed. 2005).

        [14].      But see generally Rachel Augustine Potter, Bending the Rules: Procedural Politicking in the Bureaucracy (2019); Rachel Augustine Potter, Slow-Rolling, Fast-Tracking, and the Pace of Bureaucratic Decisions in Rulemaking, 79 J. Pol. 841 (2017) [hereinafter Potter, Slow-Rolling, Fast-Tracking]; Anne Joseph O’Connell, Political Cycles of Rulemaking: An Empirical Portrait of the Modern Administrative State, 94 Va. L. Rev. 889 (2008); George A. Krause, A Two-Way Street: The Institutional Dynamics of the Modern Administrative State (1999).

        [15].      See Mathew D. McCubbins, Roger G. Noll & Barry R. Weingast, Administrative Procedures as Instruments of Political Control, 3 J.L. Econ. & Org. 243, 243–44 (1987) (outlining this principal-agent problem).

        [16].      See, e.g., id. at 255 (“[P]olitical principals in both branches of government suffer an informational disadvantage with respect to the bureaucracy. . . . [M]any of the provisions of the Administrative Procedures [sic] Act solve this asymmetric information problem.”).

        [17].      For instance, measures based solely on the ideology of the appointing President fail to capture ideological differences in consecutive agency heads appointed by the same President. In other words, they do not capture enough variation over time. Other measures only occur sporadically in time.

        [18].      The included executive agencies are the Departments of Agriculture, Commerce, Defense, Education, Energy, Health and Human Services, Homeland Security, Housing and Urban Development, Interior, Justice, Labor, State, Transportation, Treasury, and Veterans Affairs (operating as the Veterans Administration until 1989); Environmental Protection Agency; and Small Business Administration. The included independent agencies are the Agency for International Development, Civil Aeronautics Board (until its dissolution in 1985), Commodity Futures Trading Commission, Equal Employment Opportunity Commission, Farm Credit Administration, Federal Communications Commission, Federal Deposit Insurance Corporation, Federal Emergency Management Agency (until its subordination to the Department of Homeland Security in 2003), Federal Energy Regulatory Commission, Federal Housing Finance Agency, Federal Housing Finance Board (until its dissolution in 2009), Federal Labor Relations Authority, Federal Maritime Commission, Federal Reserve Board, Federal Trade Commission, General Services Administration, Interstate Commerce Commission (until its dissolution in 1996), National Aeronautics and Space Administration, National Archives and Records Administration, National Credit Union Administration, National Transportation Safety Board, Nuclear Regulatory Commission, Office of Federal Housing Enterprise Oversight (until its dissolution in 2009), Office of Personnel Management, Pension Benefit Guaranty Corporation, Securities and Exchange Commission, Social Security Administration, Surface Transportation Board, and U.S. Postal Service. Also, the Internal Revenue Service, although part of the Treasury Department, is included as a separate agency.

        [19].      See Stephenson, supra note 12, at 72 (presenting a positive theory of this dynamic).

           *      Assistant Professor of Legal Studies and Business Ethics, the Wharton School of the University of Pennsylvania.

           †      Professor of Law, Political Science and Public Policy, University of Southern California Gould School of Law. We thank Adam Bonica, Devin Judge-Lord, and Rachel Potter for data, and Ming Hsu Chen, John Harrison, Erin Hartman, Kathryn Kovacs, Jeff Lubbers, Neysun Mahboubi, Jennifer Mascott, John McGinnis, Jon Michaels, David Noll, Anne Joseph O’Connell, Richard Pierce, Zach Price, Michael Rappaport, Noah Rosenblum, Amy Semet, Bijal Shah, Kevin Stack, Matthew Stephenson, Chris Walker, Dan Walters, Adam White, and participants at the Presidential Administration in a Polarized Era conference at the C. Boyden Gray Center for the Study of the Administrative State for helpful comments. The authors also gratefully acknowledge the Gray Center’s financial support of this research. 


Like this article?

Share on Facebook
Share on Twitter
Share on Linkdin
Share on Pinterest