Tuesday, February 25, 2014

Where Do Parties Come From?

The following is a guest post from Darren Schreiber, a political scientist at the University of Exeter. He tweets at @polneuro.

On a blog about parties, it is worthwhile to consider where they come from.  A line of work going back to the 1970s has investigated the minimal conditions that will induce strangers to form groups and engage in discrimination against others.  It turns out that trivial signals like colored t-shirts or small stickers will lead people to form alliances against those with a different t-shirt or sticker.  In a paper that has just been published in Political Research Quarterly, I begin with the idea of individuals creating such small alliances and use it to explore how political parties form and compete.

I show that many features of party systems emerge naturally with just five basic rules to guide behavior and some trivial assumptions about the set up.  The strange thing about the paper isn’t the assumptions or the features that emerge; both are really conventional from the political science standpoint.  The surprising thing is that I can start with such simple, even stupid processes guiding political choices, and end up with a model that does many of the very sophisticated things we observe in political parties.

I begin with a set of typical political science assumptions about how voters might make political decisions.  The consequences of these assumptions are then explored using a computer simulation.  In this thought experiment, I imagine voters arrayed in a space based on their preferences about particular political issues.

So we could think of a group of voters standing in a line representing how big of a budget the military ought to have.  A voter who wants to spend a billion dollars on the military stands to the right of a voter who wants to spend nothing on the military and to the left of someone who wants to spend two billion dollars.  Voters look to see who is closest to them on this line and find out who is closest to their own preferences on the military budget.  If two voters both agree that they are closest to each other, then they form a tiny coalition of two.  We will imagine that when a voter who wants a billion dollar military budget forms a coalition with a voter that wants a two billion dollar budget, then the coalition they form will split the difference and seek a budget of $1.5 billion.

In the next step of the model all those coalitions look around and see which other coalition or lone voter is closest to them on the military budget line.  With agreement, they then can form a coalition of coalitions.  This process repeats until a coalition has formed that is large enough to command a majority.  If we label the largest coalitions of coalitions a party, then this simple process is enough to generate a two party system most of the time (starting with different random mixes of voters arrayed on the line).  If the party only requires a plurality to control the policy then we still get a two-party system 44% of the time.
Coalitions forming in two dimensions.
Vs are voters and Cs are coalitions.
We could label these simple coalitions “policy-seeking” since they are only pursuing their own policy agenda and are not open to making a strategic compromise beyond seeking the average budget preference of their membership.  But, it is simple enough to imagine an “office-seeking” coalition that is willing to move its budget policy just so that it can gain enough support to hold office and get to implement that policy.  If all the coalitions are office-seeking, then we get a two party system 94% of the time and another funny thing happens, the parties nearly always converge with their budget proposals at the median voter (the voter who is in the middle of the distribution if we count in from the right or the left side of that line).  In the paper, I go on to explore phenomena like party realignment and the tendency of winning coalitions to have the bare minimum number of votes needed to control the system rather than having super-majorities.

My main research for the past decade has been using functional brain imaging to study how people think about politics.  I started this project with the idea that I would need to program some kind of sophisticated artificial intelligence type algorithm in order to create simulated voters that would do the complex things we observe in political parties.  It turned out, however, that voters with really limited insight and trivial decision-making processes were sufficient to create a party system that had many of the characteristics that we observe in real politics.

No comments:

Post a Comment