The retirement of Justice Anthony Kennedy and nomination of conservative Brett Kavanaugh to replace him is poised to give Republicans a more reliable majority on the Supreme Court. But a 5-4 majority is a narrow one. A Democratic victory in the 2020 presidential election coupled with the death of a single Republican appointed justice would produce a reversal in power on the Court. The current state begs the question: precisely how likely are Democrats to regain the majority?
Despite the complexities of the legal and political processes that produce Supreme Court candidates, this question is surprisingly approachable statistically. The central variables involved are not from the domain of legal theory or constitutional interpretation, but rather from areas familiar to a statistician: life expectancy, behavioral preferences, and election projection. By consulting actuarial tables, the record on judicial retirements, and the history of presidential election results, I have endeavored to construct a quantitative understanding of the likelihood of changes of control on the Supreme Court.
How Does the Balance of Power on the Supreme Court Shift?
So how does the Supreme Court actually change composition? Because Justices have life terms, the only way that a seat becomes vacant is if there is a retirement or, more morbidly, a death. Looking at recent history, we see that justices strongly prefer to retire when the president shares their ideological perspective. Nine justices appointed since 1965 have retired, and of those only 1, Justice Thurgood Marshall, retired under an ideologically unfavorable administration. This inclination to retire under an administration likely to appoint a successor with similar views means that retirement is unlikely to change the Court’s ideological composition. Rather, retirement will likely perpetuate the current balance of power.
The most likely way for a seat to flip from one party to the other is a death. If a Justice from the opposite party of the president dies, the president can appoint a successor from his or her own party. Hence the Court balance of power shifts by one seat in the direction of the “in-party”, the party that controls the White House.
Thus we are left with a morbid strategic struggle for control over the Supreme Court. Each side awaits deaths from the other side, which hopefully occur while their own party is in the White House. Meanwhile, retirements from friendly Justices allow a party to replenish their ranks with younger members, who are less likely to cede a seat to the opposition in the near term.
To calculate the probability of the Supreme Court having a certain composition over time, I run 10,000 simulations of the next 100 years, modeling different possible iterations of the Court. Each iteration involves a simulation of the relevant events of those years: presidential election results, deaths, and retirements. The likelihood each event occurs is estimated by a means I will describe shortly. In this way, we can analyze how events might compound and combine to produce different possible Courts over time. I describe the simulation procedure in detail here, along with my assumptions and justifications.
1. Deaths and Appointments
I begin in the year 2018 with the current Supreme Court Justices, including the yet-to-be confirmed Judge Kavanaugh who I assume is also on the Court.
At the beginning of each year, I increase each Justice’s age by 1. Then I simulate whether or not a Justice dies over the course of the next year using a mortality table provided by the U.S. Social Security Administration. The probability that they pass away is solely based upon their age and sex. You can view the table here.
If a Justice does pass away, I replace them with a new Justice of the same partisan alignment as the current president. For example if a death occurs in 2019, the new Justice will be Republican aligned. Throughout, I assume that an incoming Justice has the same party as the current president, regardless of the composition of the Senate at the time.
The new Justice is randomly assigned an age and gender, according to the historical data. The age of an incoming Justice appointed after 1965 was roughly normally distribution with mean 54.1 years and standard deviation 5.3 years, so I use this distribution for age. The Court currently has 3 female members, so I make the general assumption that there is a 1/3 probability that the incoming Justice is female, and 2/3 probability of being male.
In a non-presidential year, I simply simulate deaths. At the end of the year, I count up the number of Democratic and Republican justices and record these numbers. I refer to the Court’s balance of power in a given year as the number of justices on each side at the end of that year. The only complication is that I also assume a hard-cap on a Justice’s willingness to remain on the court at age 90. Therefore, if the new year brings any judge to age 90, I simulate their retirement and a new nomination, irrespective of the party of the president.
2. Presidential Elections and Retirements
Presidential election years are more interesting. In these years I simulate two other events, in addition to deaths: (1) the result of the presidential election and (2) whether or not Justices of the same party as the incoming president decide to retire.
To determine the likelihood that either party will win the presidency, I looked back at all American presidential elections since Truman’s victory in 1948. If we consider whether or not there was an incumbent president running for re-election, and which party was the in-party (the party previously in power), we see a few clear patterns. The electorate has tended to re-elect incumbents, except when the incumbent is representing a third or greater consecutive term for the in-party. In years when there is no incumbent running, the electorate has favored the candidate of the out-party (the party not currently in the White House).
The table below demonstrates these trends:
With these figures in mind, I use the following probabilities to determine the party of the incoming president in an election year.
- Incumbents have a 2/3 chance of winning re-election. There are two exceptions: If the incumbent shares the same party as the previous president, they are given a 1/3 chance at re-election. Secondly, in the 2020 election, I give President Trump (or more generally the Republican Party) a 50% chance at winning re-election due to betting markets projecting the race at approximately those odds. (As of 8/5/18, predictit.org has Republican odds at about 46%.)
- If there is no incumbent, I give the out-party, the party not previously in the White House, a 2/3 chance to win.
Why do I use 2/3 for the probabilities rather than the actual historical frequencies? Well, we are considering a small sample of presidential elections in “modern” times so I also factor in my prior beliefs. In the absence of a larger sample of presidential elections, I shift the probabilities more towards a state of uncertainty (i.e. closer to 1/2).
The important point is that there have been two tendencies: the electorate has usually re-elected the incumbent but also has favored the out-party after 2 or 3 consecutive terms of a single party control.
After determining the party of the president, I determine which Justices of the same party decide to retire. For those Justices who share a party with the president, I assign them a probability of retirement determined by a logistic regression modeling the probability of retirement by age. I note that I am estimating the probability that a Justice will retire during the president’s full four year term, not just during the current year. This model was built using data from Supreme Court justices appointed since 1965. I make the added assumption that Justices 90 or over, from either the president’s party or the out-party, always retire.
In the event that a Justice does retire, a new Justice is generated in the same manner as in the event of a death. Of course, the incoming Justice shares a party with the president.
Also, I should note that I am making the assumption that a Justice will not shift allegiance to side with opposite party of the president who appoints them. This has not been a completely unbroken rule in modern times, as two Justices who were appointed by Republicans came to be reliably vote with the liberals (Souter and Stevens). While there is always the possibility that a Justice in the future will follow this same path, I believe this probability is small due to the increased pressure on the president to nominate a Justice who will remain voting with the party’s bloc.
Each iteration continues year-after-year over the full 100 year period, though I focus my analysis on the results over the next 30 years. This is because these are most relevant years to contemporary readers, and, of course, because I have the highest confidence in near term projections.
In summary, during non-presidential election years, only deaths and retirements at
90 are modeled. In a presidential election year, the party of the president, subsequent retirements from age or inclination, and deaths are all modeled. Each of my 10,000 iterations runs for 100 years, and I record balance of power at the conclusion of each year.
Results and Analysis
The results are perhaps depressing for liberals (or exciting for conservatives), but this depends on one’s prior expectations. I find it most informative to examine the balance of power on the Court in 4 year intervals, matching the rate of presidential elections:
The Court will almost certainly be majority Republican before the 2020 election. The only question is whether the conservative majority will be 5-4, as it is now, or be more lopsided. Due to the advanced ages of Justices Ginsburg (85) and Breyer (79), my model gives the Republicans a not insignificant 33.8% probability of holding 6 or more seats on the Court before the 2020 election. Considering the uphill battle Democrats already
face in shifting the new 5-4 balance, this projection should concern Democrats aspiring for a quick reversal.
In their most optimistic scenarios for a quick turnaround, Democrats can hope for a 2020 Democratic presidential victory and the death of at least one Republican Justice over the period from 2021 to 2024. Unfortunately for liberals, the most senior Republican Justices are relatively young. The two oldest Republican aligned members, Justices Thomas and Alito, are 70 and 68 years old, respectively. My model estimates the probability of Democrats having a majority on the Supreme Court in 2024 at 12.1%. If a Democrat does in fact win the White House in 2020 (I assign this event 50% probability), the odds of a liberal court in 2024 approximately double to 24.3%.
Of course, if Republicans instead maintain control of the White House in 2020 then they will have even more opportunities to press their advantage on the Court. In a “Republican victory in 2020” scenario, Republicans have 6 or more seats 75.8% of the time by 2024. Even in the base scenario (assuming Republicans have a 50% probability of winning the White House in 2020), conservatives have approximately a coin flip (49.6%) chance of having 6 or more seats in 2024. A 6-3 or more lopsided Court is a real possibility.
The graph below demonstrates the importance of the 2020 presidential election to both parties.
Returning to our base model, where each party has a 50% chance of victory in 2020, the chances of a Democrat controlled Supreme Court in 2028 and 2032 are 23.3% and 28.7%, respectively. Thirty years from now in 2048, Democrats have a respectable 40.2% chance of having a majority.
Finally, apart from the grander contest over control of the Court, I find it interesting to quantify how vulnerable each individual seat is. Below I display the probability that each Justice will lose their seat to a new member from the opposite party, over the next 30 years. Naturally, the older justices are at a greater risk of passing away under an opposing president and ceding their seat.
The Supreme Court balance of power is amenable to a simulation based modeling approach. Under this approach, we can see that Republicans have a stronger near term edge than at first might be supposed from their narrow 5-4 majority. They have an opportunity to capitalize on their White House control to potentially obtain a sixth seat, a possibility that is made more likely by the advanced ages of Justices Ginsburg and Breyer (85 and 79 respectively). Democratic hopes at a near term Court reversal rely heavily on the 2020 election, which at this point looks like a toss-up. A lot is on the line, and the relative strengths of both sides will be much clearer in January of 2021 when the next president is sworn in.