Monday, October 11, 2021

Polarization and The Median Voter Theorem: A Natural Experiment

Polarization and animosity between our two political parties has been rapidly increasing over the past two decades, both among voters and their representatives. In other words, the distribution of voter preferences over the Democrat-Republican spectrum is shifting from single peaked to double peaked.
This intense change in voter preferences provides a natural experiment to test characteristics about voter utility and behavior.

In class, we proved that, given certain assumptions, the median voter theorem holds under any distribution of voter preferences. Since the two new peaks of voter preferences are basically the same size, the median voter stays in the same place and policy should remain unchanged. In particular, we had to assume that abstention was not allowed or at least it was distributed evenly across the distribution, not concentrated in the tails by voters dissatisfied with centrist policies. If there were ideologically motivated abstensions, then a shift of voter preferences to opposite sides of the spectrum would spread the policies of each party apart as well. We can test which one of these models is more accurate by looking at the outcomes of voter turnout and government policies.

Do we see ideologically motivated abstention and diverging policy? No. In fact, it looks like the most extreme voters are also the most politically active.
Additionally, on most important metrics, the policies actually implemented by either party are still similar. Even though the rhetoric between Trump and Biden was more adversarial than ever, both have kept the border closed, both have kept high tariffs, both funded planned parenthood, neither instituted a national vaccine or mask mandate, neither raised the minimum wage, both spent trillions of dollars, both wanted to leave the middle east, and both highly regulated business.

So if the median voter theorem is still holding and policies haven’t diverged, why are the most extreme the most likely to vote and why do the parties and their voters seem to hate each other more than ever? The answer goes back to where voters get their utility. A parsimonious explanation is that voters are motivated more by tribal allegiances and an inherent utility from voting for a favorite team than by policy considerations. The most extreme voters are the most entrenched in the belief that their party is great and the other is a threat to American society, so they get the most utility from voting for their team, regardless of the policies either party passes. Parties want more of these most loyal followers so they fire up their rhetoric to energize their fans and polarize everyone to one side or the other. Hopefully this new strategy for satisfying voter preferences doesn't result in civil war!


1 comment:

Dani Szabo said...

I disagree in a few places with your viewpoint. I agree the parties have become more polarized than in the past. However, I believe we see differences in policy from administration to administration in many areas. The magnitude of the changes being relatively small (if you wanna argue that) I believe has less to do with Biden's actual stances, and more with the fact that our political system faces gridlock and Biden doesn't have a stable Senate majority (Joe Manchin being the major holdout). If you look at his infrastructure bill specifically, it would have been one of the most progressive in history, but it's un-passable because of Republican (and Joe Manchin) opposition. I think a better explanation for why the Median Voter Theorem doesn't hold in presidential elections stems from primaries. Primaries generally elect candidates who hold the median view of the party (although one might argue that even then voters with more extreme views tend to vote at higher rates in primaries over more moderate party members, skewing the candidate towards the extreme of the party). However, this median is then not the median of the general electorate, but rather the median of the party, and therefore we don’t get outcomes following the Median Voter Theorem in Presidential elections. There is a more in-depth analysis and backing for this in a working paper by Simon Anderson.