Thread by @edenhofer_jacob

1/ While many non-political-scientists look askance at political parties, the pol sci literature regards them as vital for making democracies in large societies work. Let me explain why in this 🧵.
First things first: How do we define political parties?

2/ There are many definitions, but let me highlight two particularly important ones. Downs (1957, 25), following Schumpeter, defined parties as “teams of [politicians] seeking to control the governing apparatus by gaining office in a duly constituted election.” Sartori (1976, 64)

3/ , by contrast, defines a party as a ‘political group that presents at elections, and is capable of placing through elections, candidates for public office’. Fundamentally, parties facilitate compromise between different factions who, while sharing broadly similar ...

4/ ideological commitments, differ in their specific policy views.
So, why do parties matter? See👇for my summary of the literature - I explain these four points in more detail below.

5/ 1⃣ Parties allocate the scarce resource of political power by deciding (i) activists / members can run for office and (ii) which elected candidates are appointed to which parliamentary committees, cabinet posts, or other positions. This point is emphasised by Aldrich.

6/ 2⃣ Parties allow politicians to compete more effectively for votes in (at least) two ways.
2⃣.1⃣ Parties allow candidates to coordinate their campaign efforts via the pooling of resources and by alleviating the moral hazard problem inherent in electoral ...

7/ teams (Cox et al. 2021, @JFiva + @ProfDanSmith).
Moral hazard refers to the party leadership’s (leader of the electoral team) problem of incentivising their candidates (team members) to exert the maximum possible campaign effort (supply productive
sciencedirect.com/science/articl…

8/ productive inputs), when the latter is unobservable or cannot be contracted for and only the outcome (winning or losing a seat/the election) is observable (see also Holmstrom 1982)
Why is this problematic for seat-maximizing parties?
jstor.org/stable/3003457…

9/ In some elections, notably closed-list PR ones, parties would gain higher seat shares if those with high (safe) or low (hopeless) list positions campaigned vigorously for those with marginal list positions - though they would not benefit personally. Cox et al. (2021) show

10/ theoretically and empirically that parties can solve this moral hazard problem by committing ex ante to allocating higher offices in government as a monotonically increasing function of members’ list rank.
2⃣.2⃣ Parties allow politicians to compete more effectively for

11/ votes when the policy space is multi-dimensional, as Levy (2004) argues in an elegant paper. Here is the argument.
•Without parties, candidates compete in multiple dimensions by proposing policy vectors. Most candidates face the problem that their
sciencedirect.com/science/articl…

12/ ideologically most preferred vector diverges from the vote-maximizing one (core tenet of citizen-candidate models a la Osborne and Besley).
•Candidates cannot credibly promise to implement a compromise vector that is closer to the vote-maximizing vector than their

13/ ideologically most preferred one. After all, without parties, no one can prevent them from breaking this promise once elected. Multiple dimensions create a commitment problem.
•Party-based enforcement mechanisms, such as whipping, enable individual candidates to credibly

14/ commit to a compromise vector, which is some weighted average of the ideal vectors of the different intra-party factions.
•Why does Levy’s argument fail in a one-dimensional policy space?
•Absent concerns about campaign effort, candidates would be better off running

15/ on their own with one dimension, rather than forming parties.
•Any party would consist of left- and right-wing candidates. The party’s policy platform would be some weighted average of the candidates’ ideal points. As is well-known, in stylised environments, the median

16/ position wins.
•So, candidates cannot find a compromise that would satisfy all of them relative to the median, making the party unstable.
This brings us to the third function.
3⃣ Parties help legislators solve the problem of collective policymaking in (at least) two ways.

17/ 3⃣.1⃣ There is a Coasean logic to political parties (@boixserra 2009).
•Suppose legislatures consisted of unorganized representatives. Then, for every piece of legislation, new groups of legislators would have to form to draft the ...
academic.oup.com/edited-volume/…

18/ laws and organize majorities.
This is time-consuming and would slow down legislative activity. A similar argument holds for committee work and other oversight activities.
•Parties, like firms, reduce the transaction costs of legislating and holding the executive accountable.

19/ They are vehicles for coordinating parliamentary activity.
3⃣.2⃣ Parties allow for credible vote trading (log-rolling) between parties and factions within a party. This, as Rosenbluth and Shapiro (2018) argue, can improve the coherence of policy platforms - both within ...

20/ one time (horizontal coherence) period and across time periods (vertical coherence or dynamic consistency). This is because parties internalise / take into account one policy proposal has on other domains, which individual legislators would be less likely to do.

21/ 4⃣ A fundamental problem for voters is that information acquisition is costly (see Lupia and McCubbins). Parties and party brands alleviate that problem.
If there were no parties but only individual candidates, it would take a lot of time for voters

22/ to determine what each candidate stands for on the issues that matter most to them. Party brands serve as shortcuts for voters, allowing them to infer what a candidate stands for without too much effort. These shortcuts work particularly well when party discipline is high.

23/ Let me conclude this 🧵with some recommendations for further reading. It goes without saying that the classic paper by Strom should be on the reading list of any serious political scientist.
jstor.org/stable/2111461

24/ I applied the Stromian logic in this 🧵on the CDU's incentives to join forces with the AfD.

https://twitter.com/edenhofer_jacob/status/1682696947713409024

25/ This paper by Bawn et al. is also worth a read
cambridge.org/core/journals/…

26/ I am also a big admirer of this paper by @Nolan_Mc and Schickler
annualreviews.org/doi/abs/10.114…

27/ Finally, let me draw your attention to this new
@BJPolS paper by Lindvall, Rueda, and Zhai
cambridge.org/core/journals/…