Presidentialization is still a thing

With the second impeachment of Donald Trump, we can say that one piece of good news is that Samuels and Shugart (2010) are still right. In our book, Presidents, Parties, and Prime Ministers, one of our claims is that parties in presidential systems face a severe dilemma: On the one hand, they need leaders who can win a separate popular election. On the other hand, the leaders selected for that purpose may not always share the goals of the party, but the party is basically stuck with the president, given the fixed term. While impeachment and removal are usually available under constitutional provisions, it is almost an iron law that parties do not vote to impeach their own president.

On 13 January, and in the wake of the insurrection of 6 Jan., this theory was put to a severe stress test. In fact, the day before the impeachment vote, it looked like the dam had broken and there would be many defectors from the Republican Party, who would join with Democrats and vote to impeach. The biggest blow was Liz Cheney, with the no. 3 position in the GOP House leadership, announcing she would vote to impeach. That seemed like it could give cover to others who wanted to break with the president after his reprehensible actions the week before. The New York Times reported that Kevin McCarthy, the minority leader, “and other party leaders have decided not to formally lobby Republicans to vote “no”.” Moreover, according to the same report, the Republican Senate leader, Mitch McConnell believed Trump had committed impeachable offenses.

Yet, in the end, there were only ten defectors. While this is the highest number of Representatives from a president’s own party to have joined an impeachment vote in US history (all four such votes), it is only about 5% of the total number of party members in the House. Normally, we would think of 95% unity as pretty high, and thus the case of Trump’s impeachment conforms, so far, to the theory: the president’s party does not vote in favor of a process that could lead to removal of its own leader, the president.

By contrast, the book shows that for about a third of prime ministers in parliamentary systems the manner in which they leave office is due to their own party replacing them between elections. Fundamentally, prime ministers do not have fixed terms and are agents of their own parties. Presidents, on the other hand, typically cease to be agents of their parties upon being nominated and especially upon winning the presidential election. This is the key argument of the book: “Presidentialization” effectively reverses the principal–agent relationship, as party members have strong electoral and other incentives to follow the lead of the president whose term does not depend on their ongoing support.

Presidents’ parties may not always support the president’s legislative initiatives (although in most cases, they follow the big ones, even when such initiatives deviate from normal party priorities–see Chapter 8 of the book), but they do usually hold the ranks together when it comes to a co-partisan president’s continued tenure in office. Apparently, even after incitement to insurrection over refusal to accept a lost reelection bid, and even with only a week to go in the term.

In connection with the above argument, some have asked what about Richard Nixon? Had he not resigned, it would have been a bipartisan impeachment and removal. This is probably correct. We also have other cases in our dataset of presidents who resigned for one reason or another. Obviously, in these cases, we are unable to observe an impeachment vote, so they are outside our theory. We can thank Nixon and others for sparing their parties the need to violate an iron law!

More seriously, there is probably, theoretically, some floor of presidential approval below which the dynamic changes. I do not claim to know where that floor is, but Nixon probably breached it when his approval hovered near 20% at the end. Given the small N problem, this remains entirely speculative. The logic might be something about tipping points of support in the party member’s own constituencies, as opposed to a parliamentary party, which typically has a more collective leadership that looks out for swing voters who determine its ability to retain executive control in future elections. And in multiparty systems, this modelling would get even more complex. Lots of PMs lose office due to coalition collapse. Presidents rarely go out that way. There is the case of Dilma Rousseff in Brazil (2016), but it conforms to the theory: her party voted 0-10 in the Senate against conviction. Ultimately, her problem was that her party had only 10 of the 81 seats! (They had also voted 0-60 against impeachment in the Chamber of 513 total members.) There was also the case of Park Geun-hye in South Korea in 2016, where some unknown number of members of her party may have voted to impeach. The reason it is unknown is the vote is secret. If the logic of members not dumping a co-partisan president is tied to electoral incentives (fates of legislators tied to that of the president), then a secret vote would break that. In the book we also mention the case of Raúl Cubas Grau in Paraguay (1999), forced out during an impeachment vote by his own Colorado Party. In this case, the party held a super-majority, and could do it alone without fear of electoral blowback. We discuss some other cases with splits in a party. The bottom line is that there is nothing routine about impeachment, and the calculation of president’s co-partisans is usually that it is unwise to break with the leader who won your own voters’ support in the most recent election. Trump’s case would be the only one I am aware of in which the most recent presidential election was one he had lost, but we still saw a very high degree of overlap between vote for House GOP winners and votes for the president, meaning that a break is essentially saying to voters, sorry, you voted for a crook, so let us set things straight for you.

So 13 January may not have been a good day for American democracy, but it was a good day for comparative institutional political science.

7 thoughts on “Presidentialization is still a thing

  1. Impeachment/conviction is one instance for which a secret ballot might be more appropriate. The reports that many GOP congressmen voted against impeachment because they literally feared for their lives and those of their families are terrifying. Maintaining party discipline by threatening to “primary” someone is one thing – doing so by rousing the mob to (credibly) threaten physical retribution is quite another. Given that a sitting president heads the coercive power of the state (even if s/he not a populist with a private army), and might also control the distribution of various carrots to accompany those sticks, I wonder whether allowing members to vote their conscience in a secret ballot might be a useful exception to the norm of transparency.


    • Yes, I agree. This is one case where I’d favor a secret ballot for exactly those reasons.

      And, as I was drafting this, I was wondering if this case of minimal partisan defection should not “count” due to reports of members feeling threatened in more than their future electoral careers.


  2. Parliamentarianism is a thing as well.

    What about Motion of no confidence in a parliamentary democracy? I imagine that it works differently if the parliament is elected by PR vs a Plurality Majoritarian system. Are intra party coups more common with Plurality Majoritarian systems vs Multi party majority coalition/minority coalition governments where governments are more prone to break down due to minor coalition partners leaving government thus exposing the government to possible motions of no confidence that lead to a new government being formed or snap elections?

    Perhaps having secret ballots of votes of no confidence helps or hinders matters?


    • Rob, I am not sure what that first line of your comment is intended to convey.

      “Presidentialization” is the reversal of the principal-agent relationship whereby legislators lose the incentive to challenge the party leader’s position. Its logical opposite would be “parliamentarization” but I doubt it is even possible absent an actual parliamentary system, or at least a mechanism for no-confidence votes (as in hybrid systems).

      Now, to other points you raise… a party replacing its leader is NOT a coup! The point of all this is that parties choose leaders to be their agents and executive candidates. Under the normal working of a principal-agent relationship, the party-principal can replace the leader-agent. A process of a principal replacing the agent that is accountable to it is very much the antithesis of what is meant by the term, coup. Some journalists in the UK and elsewhere misuse this term, but we should not follow their lead.

      On the other hand, the separation of origin and survival inherent to presidentialism can (effectively) reverse this normal functioning of an organizational principal-agent relationship, both because leadership selection has to take account the potential leader’s ability to win a separate election, and because of the fixed term once their chosen leader has won the election. So “parliamentarization” would simply be the regular functioning of the intra-party accountability relationship. Unless a party somehow finds a way to work around the separate survival problem and hold its leader-president accountable, I do not believe “parliamentarization” can be achieved under presidentialism. Under parliamentarism, on the other hand, it is just normal intra-party politics. (Not to say there is not variation in its degree across countries, parties, and prime ministers. As to your question on majoritarian or PR systems, see “Do parties dump their own PM less often in Westminster systems?“)


      • Also, conceptually, we need to distinguish a vote of no confidence (voted by the full parliament) from an intra-party replacement of the PM. They have the same outcome, but the former are usually initiated by the opposition and/or coalition partners, whereas the latter are purely internal party affairs. A few of the former begin as the latter, but most of the time, these can be distinguished in reality as well as conceptually.


  3. “There is the case of Dilma Rousseff in Brazil (2016), but it conforms to the theory: her party voted 0-10 in the Senate against conviction. Ultimately, her problem was that her party had only 10 of the 81 seats!” so perhaps extreme party fragmentation is actually a good thing for presidentialism?

    “There was also the case of Park Geun-hye in South Korea in 2016, where some unknown number of members of her party may have voted to impeach. The reason it is unknown is the vote is secret.” I think we can be certain some of her own party’s MPs voted against her – her party had won 122/300 seats in the election that took place earlier that year, and from what I’ve been able to find, they still held those seats, apparently with a few independents having joined the party since the election and before Park was impeached. 234/300 MPs voted to impeach (the subsequent trial was in the Constitutional Court). Soon after the impeachment (and even before the final verdict), about 20 MPs left the party, and a handful more would leave later.

    Given impeachment does cause a lot of damage to a party’s reputation, that probably still hangs over an individual legislator’s vote, even when their individual vote is secret.


  4. Pingback: Our Political Reality

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.