The 2022 midterm elections are coming up (remember to vote on November 8^{th}). With every election, there’s always a lot of talk about how important it is to vote (*true*), and that if the other side wins it means the end of the world (*potentially less true*). There are always important issues at play, and the results of this election will have far reaching consequences. And people naturally have questions about what all of this means. I’m going to chicken out and sidestep most of those questions. Instead, we are going to focus on what the midterm elections mean for the markets and your investment portfolio.

In the past, we’ve done some pretty deep dives about which party is better for the stock market, what you should do with your portfolio during an election, and what the markets have looked like based on the various permutations of who controls which branch of government. But I want to focus in on what happens with midterms.

Let’s actually start with what we found with the Presidential elections. In short, the average return of the market was slightly lower in Presidential election years than during non-Presidential election years, but not by much. There’s some interesting stuff going on with what those returns look like based on whether the Presidency switched parties or not, but what about the midterms?

**Do the Midterm Elections Impact the Markets?**

Historically, we have not paid as much attention to midterm elections as we have Presidential elections. But they are still big events. They still – for a very good reason – dominate news coverage. And as such, it makes sense that they would have an impact on what happens in the market. So, do they?

Well, maybe.

With that exhilarating lead in, let’s get even more crazy and take a look at some data. For this analysis, we’ll be using the S&P 500 Index as a proxy for the US market, and really focusing on two big data points: the average return and standard deviation. The average return will help us understand what return we would expect, and the standard deviation will help us understand how much, well, deviation, there is in those returns.

Number of Years | Average Annual Return | Annual Standard Deviation | |

No Elections | 48 years | 14.57% | 20.32% |

Election Year | 48 years | 10.09% | 18.44% |

Presidential Election Year | 24 years | 11.57% | 16.32% |

Midterm Election Year | 24 Years | 8.60% | 20.23% |

*For illustration purposes only. Returns are the S&P 500 Index. Data from 1926 – 2021. No Elections means any year where there was not a Presidential or Midterm election in the US. Election Year means any year where there was either a Presidential or Midterm election in the US. Indices are not available for direct investment. Past performance is no guarantee of future performance.*

Looking at this, the first thing that sticks out is that both the average annual returns and standard deviation are lower in election years than during non-election years. And the effects look pretty sizable.

Taking the returns question first (since that’s what everyone is going to focus on), we would say that there is a premium of almost 4.5% per year for years where there are no elections. That’s a big deal. We can decompose it a little bit and point out that the real driver is the midterm elections. On average, the returns of the S&P 500 during non-election years beat the returns during years with a midterm election by very nearly 6%. That’s absolutely massive.

But there are two things to think about here:

*Is this actually real? Or is it just a trick of the data?**Assuming it is real, what would you do about it?*

Let’s take a look at these questions in reverse order. Let’s suppose that we know for a fact that the returns of the market are going to be lower on average in years with a midterm election (which we’ll deal with in a minute), what would you do differently? What changes would you make to your portfolio? Based on this data, the average return was still 8.6%. Not as high as it is in other years, but it’s still significantly positive. There might be some people on the margins who would reduce their equity exposure because the stocks weren’t going to beat bonds (on average) by as much, but what are your other options? Would you go to cash because the expected returns aren’t as high as in other years? What would improve your portfolio based on this information? These are the returns on offer (at least if we assume that this relationship is real). We don’t want to make changes to our portfolios just to make changes to our portfolios.

Now, the big question – is it real? Based on this information, should we conclude that the stock market does worse in election years?

**No.**

Or at least not with any real degree of certainty.

When we’re working with data about the financial markets (or pretty much any data for that matter), we always want to keep in mind how noisy everything is. We want to ask what the chance is that the average returns during non-election years were higher simply by chance.

And I’m asking this question because it’s pretty likely that it really is just noise in the data. When you run the numbers, there’s a little bit better than a one in four chance that the difference between the returns of election and non-election years was just there by happenstance. And even if we break it down, there’s a 51% chance that the “true” average return of the markets during Presidential election years, and a 25% chance that the “true” average return of the markets during midterm election years, were the same as during non-election years. And unless someone can come up with a *really* good story for why we should expect the S&P 500 to have lower returns during election years, I wouldn’t read too much into this.

**But what about the difference in the standard deviation?**

Let’s start with a simple question. Does it make sense that the market would be less volatile in election years than in non-election years? Would you predict that future election years would continue to have lower volatility?

That seems pretty unlikely. The financial markets move based on new information. And there is a ton of information spewing out during an election campaign – some of it may even be meaningful. I would suspect that, in the future, we would see more volatility during election years, especially Presidential election years, not less.

**So, what might be going on here?**

Well, my first instinct is always to look at the number of observations that we’re working with. The smaller the number of observations, the easier it is to get weird results. In this case, we’re looking at 96 total years of data – 48 non-election years, 24 Presidential election years, and 24 midterm election years.

The standard deviation in midterm election years is basically the same as during non-election years, but Presidential election years (which I would have expected to be the most volatile) have a really low standard deviation. 24 observations are not a tiny sample, but it’s not a huge one either.

Most likely, this can be chalked up to randomness being tricky. Weird things happen – and when we run the numbers, there’s a better than 10% chance that the standard deviation of the returns in Presidential election years was just lower than the standard deviation of the returns in non-election years by chance. I think I’m willing to take those odds.

**Does This Mean Anything?**

It’s easy to look at numbers and think that they are definitive – this number is bigger so it’s better (or worse). But it doesn’t always work that way. Statistics, how we use numbers to understand the world, usually only describes what happened. It’s a lot harder to tease out what it means – if anything.

Statistics can show how likely something is to have happened by chance. Typically, we want to see less than a 5% chance that a relationship is just a fluke of the data before we admit that there’s something there – before we say that the relationship is “statistically significant.” But think about the flip side. If there’s a 95% chance that something is *not* random, that means that there is a 5% chance that it *is* random. We’re willing to accept a one in twenty chance of accepting something as true that might just be random. If we run enough tests on a data set – and if we’re talking about the financial markets, we run *a lot* of tests – we’re going to accept a whole bunch of fake relationships.

And for a lot of those relationships, you could probably find a way to tell a pretty good story. Especially if they conform to what you already believe about the world. In our day-to-day life, we don’t generally think too critically about news stories that line up with what we already believe. But we look for every possible way to poke holes in stories that challenge our beliefs. We are predisposed to find the problems with the arguments that we don’t like (or anything that supports those arguments), but we let bad arguments that we agree with sail on by. We are all incredibly good at motivated reasoning.

So, when we look at a relationship, especially one that’s “statistically significant,” we need to be asking not only how likely something is to be there by chance, but also *why* it would be there. We need to step back and ask what a result tells us about the world, and if that actually makes sense.

This is something that we talk about a lot with regard to our investments. The financial markets are really good at impersonating a random number generator – which lends itself to coming up with crazy stories that you can “back up” with data. For instance, lately I’ve been getting a bunch of emails about a fund that is trying to capture the “night effect.” If you massage it enough, you can get the data to say all sorts of weird things.

Generally, when we are looking at a new investment relationship, we use a five-part test to evaluate whether a relationship is actually there, and that it’s useful. And the first part of that test is thinking about *why* that relationship should be in the data. Your analysis doesn’t stop when you run the numbers. That’s where it starts.

**What Do You Do With This?**

Other than developing a healthy skepticism of statistics and being wary of conclusions drawn from small numbers of observations, not all that much. From a strategic, allocation level perspective, your investment portfolio should be designed around investing for the long term. Well, this is part of the long term. We’ve got federal elections every other year, and you probably have local elections the other years as well. These are not exactly unexpected events.

As we’ve discussed many times before, one of the keys to a good investing experience is staying disciplined. It’s much better to stick with a mediocre investment strategy than it is to bounce in and out of the perfect strategy.

You need to be in your seat to get the returns you deserve. And elections don’t change that.