Dave Wascha writes,

I’m auditing a grad course in System Dynamics at the University of Washington. For our final project we’ve chosen to model the structure of the US Constitution and its impact on voting behavior over time. I’ve graphed the margin of victory for presidential elections from 1856 to present and the result is an oscillating pattern that is a common artifact of System Dynamics. It implies that there is too long of a delay in the feedback loop (elections). We see this across many domains like boom and bust economic cycles, traffic patterns, even the popularity of restaurants. I’m sure you’re familiar with it.

For our project I’ve stipulated (not based on much) that I believe it to be the case that voters cyclically vote the majority party out of the executive branch and/or the legislative branch after some period of time where discontent builds. There are potentially three interesting oscillating patterns at play with one another. The executive branch, the legislative branch and a long arc feedback loop in the judicial branch. These are naturally designed as a system of checks and balances which is why I think it is such a great use of System Dynamics to take a look at.

Both parts of the legislative branch have different “policies” in System Dynamics parlance, around term lengths and limits and of course the executive branch is limited to two terms of 4 years each. Every voting opportunity represents a cycle of feedback and as I mentioned before since there is an oscillating pattern it means that there is a delay in the system. I would like to explore whether this delay is a function of term lengths and limits and if so what the potential implications are of changing the policy around term lengths and limits. The goal would be to come up with a new policy that minimizes the amplitude of the oscillation. In this case the new policy could be a change in term lengths or limits or both.

My only comment is that the recent close elections can be attributed to increased political polarization: as William Leblanc explained to us in our seminar last week, with more voters having strong feelings about the two parties and fewer voters in the middle, you’ll get smaller swings.

"For our project I’ve stipulated (not based on much) that I believe it to be the case that voters cyclically vote the majority party out of the executive branch and/or the legislative branch after some period of time where discontent builds."

I don't know much about system dynamics beyond basic fourier stuff and differential equations, but that doesn't sound like a refutable hypothesis with election data. It could always be that a change in government was caused by a wave of a frequency such that it only synchronized with that election in your data.

There is an article on cycles in American electoral politics in the next APSR (Feb 2008).