Proponents,Caretakers,and the Dynamics of Administrative Leadership Turnover in U.S. Executive Agencies

On the Importance of U.S. Federal Executive Appointees’ Tenure: Continuity and Change in Administrative Leadership Positions

Personnel continuity is of the utmost importance to the president, and ultimately for coherent executive branch governance, especially for those individuals serving in administrative leadership positions within government agencies. Appointing loyalists is requisite for ensuring presidents’ policy goals are to be faithfully executed by the federal bureaucracy (Gallo and Lewis 2012; Hollibaugh, et al. 2014; Moe 1985; Nathan 1983). Longer tenured administrative leaders enhance the probability of policy success while reducing the potential of bureaucratic drift that a president may incur when high-level administrative officials must be routinely replaced (Bertelli and Lewis 2013). It is beneficial for individuals serving in administrative leadership positions to serve for a substantial amount of time in their appointed leadership positions within U.S. federal agencies so that they can cultivate trust with careerist civil servants (Aberbach and Rockman 2000; Heclo 1977; Resh 2015). Administrative leadership continuity also mitigates turnover problems resulting in agency inactivity, loss of agency expertise, and disruption to implementing either new or nascent policies (Bertelli and Lewis 2013; Dull, et al. 2012; O’Connell 2009). Presidential appointment choices shape both stability and change in executive governance. The duration of executive appointee service has broad consequences for public agencies. In 2005, Hurricane Katrina made land fall in the Gulf Coast region of the United States, causing vast destruction to both life and property. What followed was one of the largest failures in government response in U.S. history. At the time of the disaster, it is estimated that 15 to 20 percent of the positions in the Federal Emergency Management Agency (FEMA) were vacant (Bertelli and Lewis 2013). Without stable leadership in the upper-level positions, an agency is poorly equipped to handle the routine policy and management tasks required of all federal agencies. The 2006 bipartisan congressional report, Failure of Initiative, noted that the confusion within FEMA, and among its leadership, was exacerbated by such problems relating to “inexperienced” political appointees (U.S. House of Representatives 2006: 298). Although the failures associated with an inadequate emergency response was numerous, clearly much of FEMA’s failures can be traced to inexperienced leadership resulting from high levels of personnel turnover.

Considerable variation occurs in the tenure distribution of U.S. federal executive leadership appointees analyzed in this study. For instance, the Department of Education has a median leadership appointee tenure of 718 days, with Assistant Secretary Albert Bowker [Carter appointee] serving a low of 206 days, while Secretary Richard Riley [Clinton appointee] served a high of 2,920 days. Conversely, the Office of the U.S. Trade Representative has a median tenure of 1,352 days, with Rob Portman (George W. Bush appointee, U.S. Trade Representative) holding the position for 403 days and U.S. Trade Representative Mickey Kantor (Clinton appointee, U.S. Trade Representative) 2,627 days. As it pertains to individual appointees, Leslie Kramerich (Clinton Appointee, Assistant Secretary of Labor) served for 20 days, while Ernest Ambler (Carter appointee, the Director of the National Institute of Standards and Technology, Commerce Department) held the position for 4,074 days.

The few systematic empirical studies analyzing U.S. federal executive appointee turnover offer invaluable insight into how political and agency conditions shape appointee tenure service (Dull, et al. 2012; Wood and Marchbanks 2008). Yet, the loyalty of presidential appointees is treated as unobservable to presidents when evaluating appointee choices, and hence do not consider its consequences for tenure longevity. For instance, Wood and Marchbanks (2008: 380) note that “The problem is that the president can never know the appointee’s true values and preferences. Therefore, the president’s choice requires an element of trust on the appointee’s reason for serving. Under such conditions, errors in staffing may occur due to asymmetric information. Moreover, some appointees may engage in deception and misrepresent themselves. Thus, there will always be a proportion of presidential appointees who fail to effectively represent the president and/or exit the administration early. However, due diligence in initial selection decisions can reduce the severity of the adverse selection problem.” Yet, recent research shows that both the type and caliber of political appointees is explicitly evaluated by presidents (Krause and O’Connell 2016, 2019; Ouyang, Haglund, and Waterman 2017; Waterman and Ouyang 2020).

The present study focuses squarely on appointed administrative leaders that provide both critical direction to federal agencies, and hence serve as the fulcrum between presidents and the bureaucracy largely responsible for U.S. executive branch policymaking (e.g., Heclo 1977; Seidman and Gilmour 1986). Prior studies of appointee turnover mostly comprise of positions with extremely limited policy influence, such as appointed officials in non-policymaking positions within federal agencies (e.g., Chief Financial Officer or General Counsel), or those whose position does not rise to a level of policy direction and leadership necessitating Senate confirmation. Next, an incentive compatibility logic is proposed to explain turnover among executive leadership appointees in U.S. federal agencies.

Understanding How U.S. Presidential Appointments Shape Both Continuity and Change in Executive Administration

American presidents’ employ appointment powers as an instrument to facilitate their policy objectives (e.g., Moe 1985). Presidential appointment choices therefore have tangible consequences for executive administration (e.g., Aberbach and Rockman 2000; Lewis 2008; Ouyang, Haglund, and Waterman 2017). Appointee characteristics are shown to have direct implications for administrative performance. During the Obama administration, non-policy-oriented, patronage-type appointees were placed in ideologically aligned agencies to limit their adverse impact on administrative performance (Hollibaugh, Horton, and Lewis 2014). Because presidential loyalist appointees earn lower performance ratings than less loyal counterparts (Gallo and Lewis 2012), presidents often make tradeoffs between administrative responsiveness and effective performance when choosing PAS executive nominees (e.g., Hollibaugh 2015; Krause and O’Connell 2016, 2019).

Because these appointee characteristics are known at the time of nomination decisions, presidents can shape both continuity and change in U.S. federal agencies through appointment choices. This is an important element of a president’s executive branch governance strategy that has yet been fully understood or appreciated. In a sample of 860 administrative leadership positions (i.e., upper-echelon political executives) from the 41 major U.S. federal executive agencies from the Carter through G.W. Bush presidencies analyzed in this study, roughly 38% of these leadership appointees serve 2 years or less, while nearly 62% serve three years or less, and 84% serve 4 years or less. ¹ Yet, appointee tenure is not solely determined by presidents. The length of an administrative leadership appointee’s service to the president will also depend upon both incentives and opportunities for remaining in office. The confluence between the type of appointees’ chosen by presidents and the context in which these individuals serve an administration is critical for determining leadership stability within U.S. federal executive agencies.

A president’s ability to achieve desired policy goals crucially depends upon coherent executive administration, but also the requisite stability in terms of both direction and effort. Presidents must partner with political executives such that both actors are in a mutually beneficial organizational arrangement for purposes of generating stable agency leadership. Otherwise, each actor has an incentive to “defect,” and hence accelerate the turnover of administrative leaders. The logic sketched out below outlines how the incentives for both the president and loyalist executive appointees are critical for predicting administrative turnover within U.S. federal executive agencies. Alternative theoretical mechanisms are explicitly considered that may account for administrative turnover that is distinct from the incentive compatibility logic proposed in this study (see Online Appendix [see Appendices B, C, and D]).

Proponent versus Caretaker Loyalists: On Leadership Stability in Public Agencies

Appointee loyalty to the president is premised on both the ideological consistency and non-policy fealty exhibited by the appointed leader to the president (Krause and O’Connell 2016, 2019). That is, appointee loyalty to the president reflects the extent that they are congruent on policy with their appointing president, as well as exhibit an organizational loyalty to the administration by their willingness to serve as a “team player.” With respect to appointee loyalty, presidents prefer those whom they believe will better reflect responsive competence in executive administration, ceteris paribus. Responsive competence refers to the necessity of modern presidents’, utilizing senior administrative appointments, to consolidate authority and coordinate executive action consistent with presidential policy objectives (Moe 1985; Nathan 1983). This is an especially critical issue for obtaining coherent executive administration. Less loyal administrative leaders are more prone to side with careerist agency personnel on administrative policymaking matters than presidents (Heclo 1977; Seidman and Gilmour1986), thus undermining the latter’s efforts at attaining their policy objectives. Administrative leaders can engage in bureaucratic delay, and thus undermine the long-run effects of these policies from the president’s vantage point. For example, EPA head Christine Todd Whitman’s effectiveness at delaying presidential-advocated reforms of the New Source Review (NSR) process used for evaluating new environmental rules undermined the George W. Bush administration’s policy efforts at relaxing air pollution emission standards (Krause and Dupay 2009).

Yet, appointing more loyal administrative leaders is not a guarantor of greater leadership stability in executive agencies. This is because an appointee’s tenure is not only determined by the president, but also of their own accord. Given president’s desire for sufficient coherence and stability required to enact their programmatic goals, the constant churn among U.S. federal executive agency leadership positions is not desirable from the president’s perspective (e.g., Aberbach and Rockman 1988, 2000; Bertelli and Lewis 2013; Heclo 1977; Resh 2015; Seidman and Gilmour 1986). This study maintains that the willingness of executive appointed leaders to remain in office is mutual decision by both these individuals and the president. These individuals have established career backgrounds that that afford them with a nontrivial degree of control over how long they are willing to serve a president for a given administrative position. This is because they are typically in either a financial or career position where their credentials ensure that they have attractive employment opportunities external to the U.S. federal executive branch. ²

Incentive compatibility in this context refers to presidents wanting loyalist executive appointees to facilitate their policy objectives through administration, while loyalist executive appointees wish to assist the president in carrying out their policy goals. Whether loyalist executive appointees represent a proponent or caretaker crucially depends upon whether the agency that the president assigns them to a leadership position at the time of nomination is either “on” (i.e., policy priority) or “off” (i.e., non-policy priority) the president’s policy agenda. Individuals serving in presidential policy priority agencies represent proponents who play a vital role in advancing the administration’s policy goals. For instance, Secretary of Education Richard W. Riley, served 2,920 days as a “proponent” loyalist executive appointee (74^th percentile of Presidential Loyalty among this sample of executive appointees) with respect to educational reforms and initiatives that he spearheaded on behalf of the Clinton administration as a stated policy priority at time of his nomination. ³ Conversely, an appointee exhibiting robust loyalty to the president, albeit serving in an agency that is a non-policy priority agency, serves as a caretaker insofar that they are of marginal value in achieving administration policy goals in their appointed position. One example of a caretaker loyalist is U.S. Trade Representative Robert J. Portman (98^th percentile of Presidential Loyalty among this sample of executive appointees) who served only 403 days before being appointed to a relatively more prestigious position as Director of the Office of Management and Budget (OMB) within the George W. Bush administration.

Presidents, however, are constrained in their efforts to maximize loyalty from appointed administrative leaders. Because presidents do not enjoy unilateral control over the executive appointment process, they might seek to offer concessions on administrator loyalty in exchange for the Senate granting greater policy or budgetary authority (Bertelli and Grose 2011; McCarty 2004). Further, presidents consider a range of agent-related attributes beyond loyalty when making appointment choices, and hence might discount loyalty in favor of other beneficial attributes such as managerial or policy-related experience (Hollibaugh, Horton, and Lewis 2014; Krause and O’Connell 2016, 2019; Waterman and Ouyang 2020). Outside of the president’s own discretionary calculus, presidents confront limited pools of potential loyalists to draw upon when selecting appointees (Dewan and Myatt 2010), especially for upper-echelon political executives.

Both the proponent and caretaker designations are formed by mutual commitment, anchored by the president’s policy priorities known at the time of nomination that represent the terms by which the appointee has agreed to serve. ⁴ In turn, the relatively short tenures of these appointed executive agency leaders suggests that presidential policy commitments tend to be sticky during these respective service periods. Mutual commitment is solidified through both annual and multi-year presidential budgetary commitments, which provide appointees with acute insights into policy and agency priorities within the administration (Ali 2020; May, Workman, and Jones 2008; Pasachoff 2016). Further, agency executives are more apt to remain within agencies in which they are actively a part of the policy generating and implementation process, while displaying a stronger proclivity to depart from their positions when presidential policy and agency priorities shift (Bertelli and Lewis 2013; Bolton, de Figueiredo, and Lewis 2021; Wood and Marchbanks 2008).

Incentive compatibility between presidents and unelected executive branch officials confer benefits to each actor (e.g., Gailmard and Patty 2013, see Dixit 2002 for a more general summary discussion as it relates to public sector/governmental applications). Presidents strategically desire greater administrative leadership stability for policy priority agencies that require a proponent whom they depend upon to facilitate their policy priorities compared to non-policy priority agencies served by caretakers. Similarly, proponent appointees enjoy greater benefits accrued from service in the form of advancing a president’s policy goals compared to caretaker appointees whose agency does not constitute a presidential policy priority. An incentive compatible arrangement that mutually benefits both the president and appointed leader is necessary to ensure greater continuity that is often required to ensure that the presidents’ policy goals can be faithfully executed.

The importance of incentive compatibility between presidents and their appointed agency leaders are reflected in their respective interests and talents. The Presidential Loyalty latent measure of Krause and O’Connell (2016, 2019: 532–533) consists of two indicators that reflects a shared political orientation between the president and the appointee (i.e., nominee is a co-partisan of the president and made a campaign donation prior to their nomination). This latent trait measure also employs a set of binary indicators to account for fealty, or non-policy responsiveness, that appointees may have towards the president (i.e., whether the nominee previously served in an appointed position by an administration of the same party, the nominee had prominent experience with national or state party organizations of the president’s party, previous electoral experience, and whether the nominee served in the executive branch of a state during the time when the president previously served as governor). These appointee loyalty measures are employed to analyze the mean differences in appointee characteristics known at time of nomination between “higher” loyalist appointees assigned to a position within an agency that is an administration policy priority at time of nomination [n = 223 leadership appointees] compared to those whose nomination does not represent such a priority at time of nomination [n = 113 leadership appointees]. ⁵ Strong loyalists in non-priority agencies are not noticeably more loyal than those placed in presidential priority agencies (mean difference: −0.041 standardized units, t = −0.49, p = 0.623). Strong loyalist executives appointed to presidential priority agencies do not exhibit noticeably stronger management skills reflected by qualifications (Managerial Competence) than counterparts placed in non-policy priority agencies (mean difference: 0.136, t = 1.136, p = 0.257). This latent measure captures the relevant public sector managerial skills of the appointee constructed from information regarding whether the appointee recently lived in the DC area (i.e., “social embeddedness” with those both living and working in the U.S. federal bureaucracy), had substantial federal government management related professional experience (defined as being at the assistant secretary or equivalent level outside the federal government), possessed any prior federal management experience, or whether the nominee had prior agency-specific managerial experience via an appointed PAS or Non-PAS position (Krause and O’Connell 2016, 2019: 533–534). Yet, such a policy expertise differential derived from qualifications (Policy Competence) is observed favoring proponents vis-à-vis caretakers (mean difference: 0.26, t = 2.61, p = 0.009). Policy expertise is measured in terms of skills and qualifications as a latent measure (Policy Competence) that gauges the nominee’s policy-relevant specific expertise for the agency. This expertise is represented by indicators consisting of the nominee’s educational background, whether the appointee had prior professional experience in the policy issue area, whether the appointee had prior policy-relevant employment in the federal government, had any prior civil service experience, or whether the appointee had any prior civil service position within the agency that they were appointed (Krause and O’Connell 2016, 2019: 534–535). Proponents exhibit stronger policy-specific human capital attributes relative to caretakers.

Although no systematic presidential loyalty differences favoring executive appointees placed in priority agencies over non-priority agencies are observed from these data, clearly “proponent” type loyalists enjoy both a greater capacity and willingness to facilitate the president’s policy objectives compared to caretaker loyalist counterparts. Analogous to Gailmard and Patty’s (2013) endogenous theory of executive policy investment, higher skilled loyalists will invest greater service time in their appointed leadership positions when facilitating the president’s policy priorities relative to less skilled loyalists who are not afforded the same opportunity to serve the president’s policy interests. Therefore, stronger loyalist appointees care more about their agencies being valued in policy terms from the president’s perspective than compared to weaker loyalist counterparts. In short, presidents and administrative leaders both desire continuity of the latter’s service—but only when it serves each parties’ mutual interests. Otherwise, the president and administrative leader each have an incentive to “defect.” Hence, the connection between appointee loyalty and tenure in office will enhance leadership stability in the presence of incentive compatibility between a president and loyalist executive appointees. This incentive compatibility logic yields the following testable prediction for explaining administrative leader turnover:

Incentive Compatibility Hypothesis: Appointed leaders’ loyalty to the president is associated with relatively longer tenure when their service is predicated on administering a presidential policy priority compared to ‘off-agenda’ items.

That is, agency leadership continuity is conditioned by the presidents’ premium for responsive competence for those agencies central to their policy agenda, as well as loyalist appointee’s desire for facilitating such policy investments on behalf of presidents.

Data and Empirical Strategy

The empirical analysis focuses on 860 U.S. federal executive branch agency leadership appointees covering a sample of 41 policy-oriented organizational units comprised of presidential units, executive departments, executive bureaus and offices, and independent executive agencies that cover a 32-year period (1977–2008) across five presidential administrations (Jimmy Carter through George W. Bush). These data cover the sample of U.S. federal agencies and appointed leadership positions that contain latent trait measures of presidential loyalty, managerial skills, and policy expertise (see Krause and O’Connell 2016, 2019). The focus on leadership PAS positions is simply due to these individuals being largely responsible for both setting and directing policies administered by U.S. federal agencies. Turnover among both mid and lower level appointees is less consequential for the broader operations of federal agencies. Moreover, “outside” career options leadership political executives that comprise our sample are not only more abundant and lucrative, but also less variable, compared to career and lower level political executives. These data are also restricted to executive agencies since they serve “at-will” of the appointing president, and hence, account for challenges attributable to confounding from alternative appointment constraints (e.g., partisan balancing requirements), terms of appointment (e.g., staggered terms), and other unobserved characteristics relating to tenure (e.g., term renewals) that undermine comparisons with non-executive agencies.

The dependent variable is Appointee Leader Tenure Duration and has a median/mean of 907/988 days (2.485/2.707 years), a standard deviation of 588 days (1.611 years), with a minimum of 20 days and a maximum of 4,074 days (11.162 years). ⁶ Figures 1A and 1B display the histogram plot (with kernel density overlay) and univariate Kaplan-Meier nonparametric survival function. Figure 1A reveals some positive skewness in the distribution of administrative leadership tenure (skewness = 1.20), with service exceeding 1400 days (3.836 years) becoming less common compared to those individuals serving a shorter duration. Similarly, Figure 1B shows a corresponding sharp tapering off of the survival estimates among this sample of U.S. federal executive agency leadership appointees beyond the 1,460 days (4 year) service mark (with the vertical red line representing the unconditional median days of service [907 days]). Much of this distinction occurs for those individuals who serve across terms of a given presidential administration. ⁷ OPEN IN VIEWER

Empirical Findings

Table 1 presents the main regression estimates from the Cox semiparametric and Weibull parametric models, respectively—without (Models 1 & 3) and with (Models 2 & 4) inclusion of both agency and presidency unit effects. The findings relating to the control covariates offer additional insights for explaining variation in administrative leaders’ duration of appointed service for a given position. Neither the policy nor managerial qualifications of these individuals systematically affect their tenure of service based on the null findings for the Policy Competence and Managerial Competence covariates. Clearly, individuals serving in top administrative leadership positions (Agency Position Type) are anywhere from 40.2% [Model 3] to 49.0% [Model 2] less likely to depart on a given day compared to those in subordinate level administrative leadership positions. ¹¹ Ideological policy conflict between the president and Congress at the onset of service (President−Senate Filibuster Pivot Distance at Start) consistently increases the likelihood that administrative leaders depart their positions due to inter branch tensions. Internal party polarization within the Senate at the onset of service (Senate Party Polarization at Start) reduces the likelihood of executive appointee departure. This latter finding suggests that political fragmentation within the Senate weakens this institution’s capacity to pressure administrative leaders and their respective agencies compared to a more unified Senate across party lines. Individuals serving across presidential terms (Crossover Presidential Terms) experience a lower hazard ratio than counterparts serving exclusively within the presidential term when they were appointed. The risk of departure for appointed executive leaders is most pronounced for those individuals that begin service during the final 2 years of each presidential term (Administration: Years Three, Four, Seven, and Eight).

OPEN IN VIEWER

Table 1.

Cox and Weibull Regression Model Hazard Ratio Estimates of Administrative Leaders’ Tenure in U.S. Federal Executive Agencies.

Covariates	Model 1 [Cox]	Model 2 [Cox]	Model 3 [Weibull]	Model 4 [Weibull]
Presidential loyalty	1.394*** (0.138)	1.386** (0.161)	1.417*** (0.146)	1.375** (0.159)
Policy priority agency	1.096 (0.114)	1.152 (0.199)	1.125 (0.126)	1.154 (0.206)
Presidential loyalty × policy priority agency	0.651*** (0.072)	0.621*** (0.087)	0.641*** (0.073)	0.631*** (0.087)
Policy competence	0.983 (0.070)	1.034 (0.082)	0.989 (0.072)	1.042 (0.082)
Managerial competence	1.006 (0.060)	0.979 (0.067)	1.008 (0.059)	0.985 (0.066)
Agency position type	0.582*** (0.046)	0.510*** (0.055)	0.598*** (0.049)	0.539*** (0.057)
President−Agency ideological alignment	1.450*** (0.134)	0.651 (0.173)	1.470*** (0.138)	0.737 (0.185)
President−Agency ideological opposed	1.360** (0.136)	0.621 (0.167)	1.364** (0.136)	0.694 (0.176)
Executive subagency	1.103 (0.165)	1.768** (0.320)	1.102 (0.168)	1.685** (0.297)
Independent executive agency	0.839 (0.079)	2.061* (0.600)	0.828* (0.076)	1.743* (0.480)
Senate party polarization at start	0.0004** (0.001)	0.166E-10* (0.176 × 10−9)	0.0004** (0.001)	0.777E-10* (0.813 × 10−9)
President − Senate filibuster distance at start	1.771* (0.401)	1155.527** (2625.988)	1.950** (0.435)	893.796** (2000.161)
Presidential job approval at start	0.994 (0.003)	0.990* (0.005)	0.995 (0.003)	0.990* (0.005)
Crossover presidential terms	0.194*** (0.034)	0.165*** (0.036)	0.200*** (0.036)	0.175*** (0.037)
Unemployment rate at start	0.934 (0.044)	1.139 (0.099)	0.933 (0.043)	1.130 (0.098)
Administration: Year two	1.886** (0.377)	1.620* (0.368)	1.843** (0.377)	1.653* (0.368)
Administration: Year three	5.349*** (0.729)	3.954*** (0.877)	5.843*** (0.791)	4.384*** (0.924)
Administration: Year four	4.276*** (1.397)	3.535*** (1.196)	4.773*** (1.450)	3.943*** (1.239)
Administration: Year five	1.357* (0.169)	1.638* (0.403)	1.338* (0.176)	1.528 (0.379)
Administration: Year six	2.794*** (0.368)	3.699*** (0.910)	2.727*** (0.327)	3.451*** (0.852)
Administration: Year seven	5.076*** (1.287)	5.651*** (1.734)	5.651*** (1.338)	6.296*** (1.866)
Administration: Year eight	7.436*** (2.284)	9.027*** (3.468)	8.639*** (2.551)	10.022*** (3.815)
Total appointee loyalty effect: Policy priority agency	0.907 (0.068)	0.861+ (0.074)	0.908 (0.070)	0.867+ (0.075)
Total Within−Policy priority appointee loyalty effect	0.713** (0.095)	0.716+ (0.134)	0.721** (0.094)	0.727+ (0.135)
Agency and presidential administration fixed effects	No	Yes	No	Yes
ln (p)	_	_	2.499*** (0.080)	2.685*** (0.081)
AIC	9072.263	9021.488	1128.849	1043.701
Observations	860	860	860	860

The relationship between appointee loyalty and agency leader tenure serving in a non-priority agency (Presidential Loyalty) uncovers between a 37.5% [Model 4] and 41.7% [Model 3] higher average hazard ratio across model specifications. This statistical pattern suggests that low loyalist administrative leaders whose agencies are off the president’s institutional agenda are at lower risk for departure compared to high loyalist counterparts similarly serving in non-priority agencies. The incentive compatibility hypothesis postulates a hazard ratio <1.0 corresponding to the Presidential Loyalty × Policy Priority Agency covariate. That is, a rise in appointee loyalty is hypothesized to be associated with a lower average level of departure for administrative leaders whose agencies are on the president’s policy agenda compared to those who serve in non-priority executive agencies. Further, the total effect of appointee loyalty on their tenure hazard ratio appears at the bottom of Table 1 as a linear combination of these respective hazard ratios. These results indicate that the total risk of departure range for the Total Appointee Loyalty Effect: Policy Priority Agency linear combination ranges between −9.2% and −13.9% that attains marginal significance at the p < 0.10 level for the models containing both agency and presidential administration fixed effects. These linear combination effects are, on average, about −28% lower for explaining appointee tenure hazard risks that is decreasing in appointee loyalty within the subset of policy priority agencies (Total Policy Priority Agency Effect: Appointee Loyalty).

Figure 2 displays the substantive effects of these estimates by analyzing an interquartile unit marginal increase in appointee loyalty [1.3653] multiplied by the hazard ratio associated with the Presidential Loyalty × Policy Priority Agency interaction covariate. The estimates offer clear support for the incentive computability hypothesis since the estimated hazard ratios fall well below the 1.0 threshold associated with a null finding. Agency leaders serving in policy priority agencies experience a lower likelihood of departure than those serving in non-policy priority agencies, ranging anywhere from a low of a 44.3% ([0.557–1] *100) decline in the hazard ratio of appointee departure (Model 1) to a high of a 47.8% ([0.522–1] *100) hazard ratio decline in administrative leader departure (Model 2).OPEN IN VIEWER

Figure 3 analyzes these core estimates of interest converted to predicted median Weibull survival times with corresponding 95% confidence intervals (Models 3 & 4). These substantive effect estimates focus on the “extra days” of service differential that can be expected from administrative leaders serving in presidential policy priority agencies versus those serving in non-priority agencies based on respective interquartile (q = 25 → q = 75) and interdecile (q = 10 → q = 90) changes in the Presidential Loyalty covariate. Because predicted median survival times account for both (point) estimate uncertainty and overall model (prediction error) uncertainty, the estimates appearing in Figure 3 are naturally more imprecise than the estimated hazard ratio effects reported in Figure 2. The estimated interquartile increase in appointee loyalty on administrative leader length of service between presidential policy priority agencies versus non-priority agencies is 249 days (0.68 years) and 244 days (0.67 years) additional days of appointed service in a given position that respectively omit and include both agency and presidential administration unit effects. These effects constitute a 27.45% and 26.90% differential increase in predicted median tenure between those administrative leaders serving in presidential policy priority agencies at the time of nomination versus counterparts serving in non-priority agencies. ¹² When analyzing starker interdecile changes in appointee loyalty, this differential commitment effect is associated with a predicted median tenure increase for administrative leaders of 451 days (1.24 years) and 442 days (1.21 years) in these respective models. This interdecile unit change yields a 49.72% and 48.73% differential increase in predicted median tenure relative to the baseline of unconditional median tenure.OPEN IN VIEWER

Figure 4 displays the predicted differential survival rates relating to low (q = 10), moderate (q = 50), and high (q = 90) levels of appointee loyalty between serving in a presidential policy priority agency versus a non-priority agency. Analysis of the Model 4 estimates displays the dynamic differential in survival between proponent versus caretaker appointee types that cannot be gleaned from the preceding analysis of differential hazard ratio and predicted median survival estimates. Each of the vertical red lines represent predicted years of service time elapsed. Specifically, this predicted survival rate differential between Low (q = 10) and High (q = 90) appointee loyalty is a modest 3% [0.96–0.93] when evaluated at the end of the first year of service (T = 365 days). This differential rises to 16% [0.81−0.65] and 27% [0.55−0.28] respectively in the second (T = 730 days) and third (T = 1,095 days) years of service, before declining to 18% [0.26−0.08] at the end of the fourth year of service (T = 1,460 days). This 3 year range between two and 4 years of appointed service is substantively meaningful since it yields the greatest risk for appointee turnover that is germane for nearly half of the administrative leaders in this sample [83.84%–38.26% = 45.58%]. For longer tenured appointees, these differentials attenuate with the passage of survival time (i.e., appointee tenure), with estimated 7% [0.10−0.03] and 1% [0.02−0.01] differentials in the fifth and sixth years of service (T = 1,895 days, T = 2,190 days), respectively. Clearly, incentive compatibility between presidents and loyalist political executives’ shared policy interests yields the best prospects for continuity of administrative leadership within U.S. federal executive agencies.OPEN IN VIEWER

Taken together, these findings provide quantitative leverage demonstrating the conditions whereby presidents can feasibly hope to obtain from enhancing greater stability among their appointed administrative leaders based on what is known at the time of their nomination choice. Put simply, these findings illuminate how presidential appointment choices affect the extent that they instill stability in executive administration. These findings indicate that presidents obtain a conditional tenure premium that presidents obtain from loyalist executive appointees—one that can only be realized when presidents’ policy priorities are linked to an appointee’s agency that is known at the time of formal nomination. That is, incentive compatibility between presidents and their administrative leaders facilitates the president’s goals of responsive competence insofar that they can attain greater stability. When such conditions are lacking, presidents place less emphasis on responsive competence since either an agency is either not part of the president’s policy agenda or a strong loyalist is not appointed. For non-priority agencies, presidents appear to be rationally less averse to administrative leadership turnover since this subset of executive appointees are less critical to advancing the president’s policy agenda.

Discussion

A critical, yet underappreciated aspect of executive power pertains to the duration of service of appointed administrative officials. “Duration,” as defined by Alexander Hamilton, is synonymous with governance stability that is vital for ensuring that policies are implemented to achieved intended outcomes. The concept of “Duration” is central to the proper functioning of subordinate executive offices charged with responsibility for policy administration (Caldwell 1964: 90; see also Green 1990: 514), and also a means of avoiding both chaos and disruption in executive administration as duly noted in Federalist 72 (Rossiter 1961: 436; see also, Green 1990: 514). This issue is a critical one to effective governance in the modern U.S. administrative state given the tension between highly transient executive appointees and relatively permanent career officials (e.g., Aberbach and Rockman 2000; Heclo 1977; Lewis 2008; Resh 2015), exacerbated with the emergence of “acting” executive leadership officials (e.g., Kinane 2021; O’Connell 2020).

Although past studies bring to bear important empirical insights on longevity of service among U.S. federal appointees (Dull, et al. 2012; Wood and Marchbanks 2008), they do not consider the policy-based incentives appointees confront. Presidents’ appointment choices have direct consequences for ensuring continuity among politically responsive agency leaders in those administrative settings most vital to their policy agenda. Because executive-level political appointees typically serve rather short tenures that only represent a small fraction of a president’s time in office, coherent executive branch governance requires stable administrative leadership if efforts at reconciling the fundamental tension between political leadership (“mandates”) and bureaucratic professionalism (“mandarins”) are to be effective (Aberbach and Rockman 1988). Although presidents do seek to use administrative means to achieve policy goals (e.g., Nathan 1983; Moe 1985), presidents are additionally constrained beyond canonical control-expertise tradeoffs inherent to principal-agent relations. Specifically, presidents cannot attain Alexander’s Hamilton’s aim for adequate “duration” in executive administration from merely appointing loyalists, unless it is coupled with a firm mutually beneficial commitment coinciding with an executive appointment. Hence, presidents must prioritize their efforts for effective politicization of executive agencies by locating loyalist appointees in policy priority agencies to ensure administrative leadership stability.

Supplemental Material