Explaining White House Chief of Staff Departures

As Pete Rouse assumed the mantle of the "2nd most powerful person in the US" I grew interested in the institutional factors that predict White House Chief of Staff turnover and tenure.  I compiled a list of the chiefs of staff, along with a number of macro level factors that are likely to bear on the presidential office (more on this below). Before going further, please note two important caveats:

1)      On a theoretical level, this is an incredibly murky area.  Like many other things relating to the executive office, the arrangement of the staff is largely dependent on who sits in the Oval Office.  Eisenhower had a hierarchical organization styled after the military; Ford and Carter were incredibly decentralized (Carter didn't even appoint a chief of staff until 1979); while Reagan and George W. Bush adopted a "spokes-in-a-wheel" arrangement with a small group of trusted staffers rather than any one individual.  The flexibility of the organizational structure is also reflected in the various expectations for a chief of staff: Andrew Card’s resources were taxed less when he was able to share the burden Karl Rove.  As a result, it's not likely that all chiefs of staff depart under the same circumstances.

2)      From an empirical standpoint, the level of detail in my data collection only scratches the surface for what's necessary in a ‘real’ scientific analysis.  Just consider this the preliminary analysis you run on your data to make sure that there's something interesting going on before really investing time in a project.

The chief of staff position is a rather volatile one: the modal length of tenure is only 2 years, but there’s still quite a bit of variation.

Time in Position Stay On Left Total
1 Year 23 2 25
2 Years 13 10 23
3 Years 8 5 13
4 Years 5 3 8
5 Years 3 2 5
6 Years 0 3 3
Total 52 25 77

So, what explains the factors surrounding chiefs of staff departures?  This analysis hinges exclusively on predicting likelihood of departure as a function of macro level institutional variables, listed below, along with their expected sign:

1) Presidential Approval Rating (-). If the president is unpopular, it could be due to an inefficiency in the way the staff is arranged (Carter appointed a chief of staff only after it became exceedingly obvious that he wasn’t capable of doing the job alone).  Even if the staff arrangement isn’t the problem, a good way to signal a departure from the old way of doing things is to reorganize one’s staff.  Similarly, if the president is enjoying high approval ratings he is unlikely to make many organizational changes voluntarily.

2) Legislative Success Score (-).  Again, if the president is successful in dealing with Congress, the likelihood of reorganizing one’s staff should be lower than if the president were struggling in getting his agenda passed on the Hill.

3) Presidential Scandal (+).   Duh.


4)Tenure (-).  The longer the chief of staff is around, the more likely he is to retire.

5) Size of the Executive Office (as measured by annual budget for Executive Office in billions of dollars) – entered as a control variable.

Likelihood of Chief of Staff Departure
Variable Coef Std. Error P>|Z|
Approval Rating -0.073 0.03 0.01
Scandal? -0.207 1.01 0.84
Cong. Success Rate -0.013 0.04 0.77
EO Budget -0.001 0.00 0.30
1st Year -2.665 1.33 0.05
2nd Year -0.207 1.15 0.86
3rd Year -0.266 1.22 0.83
4th Year+ -0.095 1.31 0.94
_cons 4.839 2.77 0.08
n(obs) = 74
LRchi2(8) = 19.47

The (unsurprising) result? The approval rating of the President matters.  A lot.  The greater the public's support for the President, the lower the likelihood of resignation.  The only other factor that falls out is if the chief of staff is in his first year: only 2 chiefs of staff left office within a calendar year of their appointment.  Other than that, these macro levels don’t wind up having much influence on chief of staff departure.  Nonetheless, to better illustrate the importance of presidential approval rating in predicting likelihood of chief of staff departure (and to show off a relatively pretty picture), take a look at the predicted probability of resignation chart, below.

You see that across all levels of presidential approval, the likelihood of a first year chief of staff departure is quite low, but all other years show quite a bit of sensitivity to levels of presidential approval.  When holding all other variables constant, the likelihood of departure at year two or anytime after year four is nearly the same.   Nonetheless, the likelihood of resignation, even in one’s third year or fourth at the post is low under a popular president.

The major exclusion in the model here is the cause of departure and perhaps what the outgoing chief of staff went on to do after.  The circumstances around Emmanuel’s departure differ when compared to H.R. Haldeman or Sherman Adams, who both left amid scandals.