Democracy is our secular religion. Virtually every government on the planet at least pays lip service to the idea; even autocrats feel the need to maintain a veneer of democracy through rigged elections. But for many (in the Western world, at least) 2016 was a year that shook their faith. While the US in particular has significant problems with the design of its political institutions that are more to blame in this instance, a broader point remains: it’s time we began reevaluating democracy.
Democracy for Realists, which came out last year, is a work by political scientists Christopher Achen and Larry Bartels seeking to do just this. Unlike some other recent works, summarised here, these two are not necessarily out to reject democracy. But they do seek to correct the narrative of how the democratic process operates. The failure lies not in democracy itself, but in the fact that we have designed political institutions with a flawed understanding of how democracy actually works.
The basic structure of the book is a systematic dismantling of most understandings of democracy. Having torn democracy down, they conclude by building a ‘realist theory of democracy’. Exciting stuff! I’m still reading the book, and so will be using this space to work through their argument.
To begin: what exactly is it about democracy that we like so much? Many of the characteristics that people like democracy for really have their origin in liberalism: equal rights and rule of law, for example. There are many instances of illiberal democracies, where the majority tramps on the rights of minorities, and a country like Singapore may be considered a liberal autocracy. It’s not inherent that the two go together, and indeed there is significant tension between liberalism and democracy, as Dani Rodrik and Sharun Mukand argue.
Setting that side then, what is unique to democracy that we find so appealing? A common answer might be that in a democracy ‘the people decide’, and this is the origin of the ‘folk theory of democracy’ and where Achen and Bartels’ argument begins.
Under the folk theory, democracy is a way of aggregating preferences from the populace. If we have five people deciding on whether to get vanilla or chocolate ice cream, putting the matter to a vote will reveal the group’s overall preference. Because (alas) the world is not so simple, and for all of human history a system of direct democacy was wholly impractical, in practice democracies have used a system of elected representatives to convert the will of the people into action. Under the folk theory of democracy, representatives are understood to be, well, representatives: they are to express the views of their constituencies, regardless of their own feelings. They are a vessel. (This contrasts with the trustee model of representation, which we will come to in a later post).
Let’s say next year we’re passing a resolution on the official ice cream flavor of our nation. In the interim there is an election for our congress. Under the folk theory of democracy, people will vote for the party that represents their preferred flavor. If more people like chocolate best, they will vote for the party that is pro-chocolate, and we will end up with chocolate as our official ice cream flavor.
Obviously this is a simple example, but the intuition can and has been extended and formalized in the ‘spatial model’ of voting. The spatial model comes from economics, and is similar to models of competition in industrial organization. The canonical example is to imagine a stretch of beach. Beachgoers are randomly distributed across the beach. Now add two identical ice cream vendors to the mix. Where will they set up to capture the biggest market share of hungry customers? The answer is both will set up right in the very middle, giving each one half of the beach. (This assumes a uniform distribution; if the beach goers were distributed normally, then the vendors would set up at the location of the median beach goer.)
Likewise, we can imagine ideology to be a single continuous variable, like the narrow strip of beach, and political parties to be the vendors. To capture the biggest market share, both parties will set up at the location where the average ‘distance’ to any given voter is minimized: at the median. If the vendors can carry only one flavor, both will choose chocolate, because the median voter is pro-chocolate.
This model explained nicely the non-ideological parties of the postwar United States that dominated until the 80s and 90s: both were trying to maximize their market share, and so set up camp in the middle of political space. But there has been a well documented trend towards political polarization that challenges this model.
The challenges to the spatial model are both theoretical and empirical. Theory presents two major hurdles: first, the assumption of unidimensionality, and second, the challenges of Arrow’s theorem.
The idea of ideology as a single variable, with a left and a right, is of course common and part of our standard political lexicon. But for a long time there have been advocates of adding additional axes to political ideology, as can been seen in the many online political ideology quizzes. And when additional dimensions are added, the functioning of the spatial theory collapses: there is no longer any unique solution to the median voter question.
Furthermore, the work of Ken Arrow demonstrated that in any voting system with three or more choices there are no unique answers to the question of what the majority ‘wants’, because the design of the voting system can itself affect the outcome. Plurality voting can give a different answer from runoff voting: which is the true ‘will of the people?’
Let’s go back to the ice cream example. If we add a third flavor, suddenly things get more complicated. Let’s say 40% vote for chocolate, 30% for vanilla, and 30% for strawberry. In a plurality system chocolate would still win, despite the fact that 60% of people voted against it. Ranked or runoff voting systems try to deal with this by having voters make second or third choices, and transferring votes until one candidate receives a majority, but these systems can also lead to strange outcomes.
Now, even if theoretically we can dream up edge cases where voting is inconclusive, maybe in the real world democracy does well enough. Alas, no, as the empirical evidence shows that most of the time, voters don’t have particularly strong ideology. Strong partisanship, yes, but actual views on policy (when they exist) seem to have little correlation with voting patterns. As Achen and Bartels put it, ‘the folk theory is of little use in understanding actual democratic politics’ (p. 36), and I can think of no better proof than the 2016 election.
In sum, the idea behind the folk theory of democracy, that the public has coherent and knowable policy preferences and makes electoral decisions based on those preferences, is not tenable in the face of major theoretical and empirical evidence to the contrary.
Next time, we’ll deal with the argument that, by cutting politicians out of the process, a more direct democracy would lead to better outcomes, and then move on to the retrospective theory of political accountability.