The Remarkable Stabilization of American Politics

That may seem a strange title in the middle of the most topsy-turvy and ground-breaking presidential primary in, well, all of American history. However, I ask you to bear with me. The other day, I was curious about the relationship between the national House vote and the partisan division of the House. I headed over to Wikipedia, where someone has compiled the national popular vote for each election going back to 1942. It turns out that the partisan division of the House tracks pretty closely with the national popular vote. Typically, an increase in the Democratic share of the vote will produce and increase in the Democratic share of the House. Equally, sizeable shifts in the partisan division of the House rarely occur without sizeable shifts in the popular vote. Only twice, in 1942 and 1996, has the party that won the popular vote lost the majority in the House and in each case, that error was corrected two years later. Overall, that's good news for our democracy, and bad news for the thesis that says that gerrymandering has produced unresponsive elections.

Most of the variance between the popular vote and the House composition can be explained by the majority-compounding power of districted elections. 53% of the vote is often transformed into more than 53% of the seats in any districted election. For example, take a state that voted 53% Democratic in a given election. If that state is divided into 10 perfectly representative districts, then each district will vote 53% Democratic and the delegation will be 100% Democratic. Clearly, districts are not uniformly drawn so as to be perfectly representative of the state's partisan divide. There are natural concentrations of voters of one or the other party and districts are drawn in a haphazard amalgam of disparate methods. Still, the general idea still holds true. That's why Massachusetts has gone six straight elections without electing a single Republican congressman, for example.

It also turns out that a district system compounds larger majorities at greater rates than it compounds smaller majorities. That's why our 52% in 2006 translated into 53.6% of the House's seat while both the 1974 and 1976 elections produced 2/3 majorities off of 57.1% and 55.5% popular majorities. It's hard to determine the "natural" rate of compounding (because of the problems outlined above), so we don't know if the few points difference between 2006 and 1974-76 is entirely responsible for the disparate results. It's also possible that periods in which one party wins large majorities, especially in years ending in "0" they are likely to dominate state legislatures and therefore to use the redistricting process to gerrymander themselves a larger majority. There's a little evidence for that in the data, but its hardly conclusive and at times contradictory. Without another convincing narrative, we are left with the theory that districts naturally compound majorities at increasing rates.

So far, all I've shown is that the partisan division of the House is strongly tied to the popular vote in the preceding election. We know that there are structural distortions built in to the districting system (population concentrations, gerrymandering, entrenched incumbents in safe districts of the opposite party, etc.), so the popular vote and the seat shares will not track together perfectly. So, why the title about stabilizing American politics?

Well, one of my secondary motivations for looking into past popular vote totals was to see if there were any periods in which the compounding factor was significantly greater of less than in other periods. Specifically, I was looking to see if Democrats' majorities were especially compounded in the 1960s and 1970s. I was looking for the effects of gerrymandering, which, as I said above, I did not find. I was looking, however, because the partisan divisions of the House have fluctuated with less frequency and power since the 1980s than they did before and more and better gerrymandering is often used as an explanation. As we've seen, gerrymandering is not to blame.

So, why did House elections become less dynamic after the 1980? It's actually pretty simple: voters choices became less dynamic. It's not a sexy answer, but it turns out that voters pretty much decide House elections. It's not just House elections that have become less dynamic, either. If you were to chart presidential approval ratings (and I did) you would see that those ratings have become less susceptible to fluctuation, as well. Before Reagan, it was typical for a president to experience multiple spikes and falls in their ratings. Starting with Reagan, however, we've seen three presidents, Reagan himself, Clinton, and Bush 43, whose ratings followed a long smooth curve for all or nearly all of their presidencies. Across presidencies, the variations from a hypothetical median line have become far less drastic since 1980. Even presidential election results seen less fluctuation. Take in two halves the 16 presidential elections in the period covered by Wikipedia's House data, and that change is clear. The first 8 elections saw Democrats win 53.4%, 49.6%, 44.3%, 42.0%, 49.7%, 61.1%, 42.7%, and 37.5%, for a range of 23.6%. The other 8 elections show Democratic results of 50.1%, 41.0%, 40.6%, 45.6%, 43.0%, 49.2%, 48.4%, and 48.3%, for a much smaller range of 9.1%.

It turns out that we're not only in a period of close partisan division, unlike any since the Gilder Age, but our politics are also remarkable stable. Assuming this president finishes his term, we'll have gone 28 years with only 4 presidents. That's something that, sans FDR, hasn't happened since the day of Andrew Jackson. Even in that early period, as well as in the 1930s and 1940s, the House was politically volatile. The 1800-1824 period was the most stable of the early years, but wasn't close to the stability of 1982-2006.

So, why such unprecedented stability?