How Do Presidential Elections Affect The Markets?

Elections cause a lot of noise. That’s pretty much true no matter how you look at it. Just like presidential candidates, the markets are always looking to the future.

Markets move on expectations and new information. In other words, they are constantly reacting to what is happening now and how the market thinks current events will affect future events.

Political campaigns are full of new information—some scandal or triumph occurs, or a candidate gives a particularly stirring speech or changes a stance. As you can imagine, all this new information causes the markets to jump around more than usual.

And it doesn’t even particularly matter if what the candidates are saying would result in good policy or not. It’s new information. How the market reacts to this new information will vary, but the market will move simply because new information crops up. However, it’s important to remember that the defining feature of a campaign is uncertainty. Except in rare circumstances, no one (including the market) knows who will win or which set of policies the new president will (try to) implement.

So it stands to reason that investing during an election would be particularly fraught (not that investing during “normal” times is smooth sailing).

But let’s actually test this.

Are there meaningful differences in stock returns during political campaigns? Are the markets actually more volatile?

Logically, we would expect that election campaigns would lead to higher volatility simply because a lot of stuff is happening. Though the direction of influence on stock returns isn’t quite as (apparently) obvious. But let’s start picking at the data.

Do Elections Hurt Stock Returns

Let’s start by looking at stock returns. You can tell all sorts of stories about why campaigns would lead to either better or worse returns, but the nice thing about all of this is that we have about a hundred years of data to look back at.

For this analysis (and the rest of the way through) we’ll be looking at the S&P 500 Index, which we have data for all the way back through January 1926. To set a baseline, the average annual return of the S&P 500 Index from 1926 through 2023 was 11.55% per year, with a standard deviation of 18.63%.

So what happens when we explicitly compare against elections?

Let’s start basic by just comparing election years (both Presidential and midterm election years) versus non election years.

Looking at this first cut, it certainly looks as if election years are not a great time to be investing. And we have a pretty good amount of data for both election and non-election years – we have 49 years of data for both groups. But it’s important to remember just how volatile stocks are (which we’ll get to in a bit). Even with this 4.26% annual premium for non-election years, we actually can’t say with a high degree of confidence that there is a real difference. There’s a decent chance that this is just normal randomness¹.

But what if we cut the data slightly differently?

What if, instead of looking at the entirety of the year when an election happens, we just look at the campaign season? While it certainly feels like we have entered the age of the perpetual campaign (though I grew up in New Hampshire, so maybe I’m biased), it’s really the back half of campaign years when most people start paying attention. So what if we looked at just the returns from July to October² in election years?

There are a couple of stories to talk through here, but the standard deviation during election years, and especially during campaign seasons, has been higher than during non-election years. In fact, the increased standard deviation has been purely during the campaign season – to the point that the Non-Election Year and Non-Campaign seasons during election year standard deviations round to the same number³.

First off, the numbers really seem to like a difference of roughly four percentage points with regards to the election returns for some reason. But aside from weird coincidences, there’s a more important point to make. The difference between the Non-Election Year and Campaign season standard deviations is statistically significant – from a statistical perspective, we can be reasonably confident⁴ that this is not just from random chance.

There really does seem to be something to the idea that the stock market is more volatile during the height of campaigns.

But what should we do with all of this information?

So What Does All of This Mean?

Let’s start with the difference in standard deviation. During the four months we’ve categorized as campaign season, there does seem to be an increased level of volatility in the stock market. And more importantly, there’s a pretty reasonable explanation for why there’s increased volatility. But it’s important to keep some perspective in terms of what this means from a practical perspective. If we keep everything in terms of annual numbers, there isn’t a huge difference in the distribution of returns in practical terms.

As I mentioned earlier, an easy way to conceptualize standard deviations is that you can expect about two thirds (technically 68%) of observations will be within one standard deviation of the average. Well, let’s see what that means here.

As a reminder, over the period we’re looking at (1926-2023), for Non-Election Years the average annual return was 13.68% and the annualized standard deviation was 17.83%. And for the Campaign Season the average annual return was 12.37% and the annualized standard deviation was 22.22%.

Again, there’s a statistically meaningful difference here, but I’m not really sure that most people would actually notice it. Especially considering we’re only in campaign season for four months at a time.

To my mind this is a bit of the cost of simply being invested in the stock market. We know going in that short term returns are going to be noisy and just plain weird (to use some technical terminology). And this is just one more manifestation of that.

The question around the differences in historical returns is a little more nuanced though. The obvious question is whether you would want to get out of the stock market (or just reduce you exposure to stocks) during the non-campaign months of election years.

There are a couple of things to think about here.

The first is what your alternatives are. If you’re not going to invest in the stock market because the historical returns have been lower during the non-campaign season of election years, what else would you invest in. Stocks still have a positive return during this period, and they still beat bonds. From 1926-2023, Five Year US Treasury Notes have an annual average return of 5.06%. By switching to bonds you would be giving up even more return.

The other question is whether you think this pattern will persist. We started this discussion earlier, but it’s worth diving into fully, because it has some really important implications for how we look at any investment data.

In the short term the stock market (and financial markets generally) are basically random number generators. It’s really tough to tell the difference between daily stock returns and a coin flip. Which means that you can find all sorts of really great stories in the data. The problem is a lot of them aren’t real – and those stories can last for a long time. Even long enough that straightforward statistical hypothesis testing will say that it is very unlikely it’s random chance.

But it’s important to remember that the standard test of statistical significance is whether there is a 5% or less chance that whatever difference we’re looking at is actually there. Which means that there’s still a 5% chance that it isn’t there. This means that (up to) 1 in every 20 “statistically significant” findings could just be noise.

And think about how many different investment stories we are constantly looking at.

We are going to find really compelling back tested data.  We’re going to find some really interesting relationships that tie together incredibly cleanly.

And some of these can be a lot of fun – there are whole websites devoted to finding and highlighting these spurious correlations. And they’ve even started adding AI explanations – sometimes even full AI generated papers – describing why the correlations⁵ exist.

But it’s important to take that step back and ask if something makes sense. Can you tell a good story that explains why we see a difference in returns?

I think there’s a pretty good case for believing that stock market returns are a little bit more volatile during the Campaign season. There’s a lot happening then.

Bu, I’m not so sure about the returns we’re seeing around elections. Asking why the returns are the way that they are is only the first step in deciding whether a return difference will persist.

All of this is a long way of say to take these sorts of numbers with a whole truck load of salt.

Would I be surprised if we see the same pattern of returns for the upcoming election? No, not particularly. But would I expect the pattern to continue? No. I wouldn’t.

The fundamentals of long term investing for retirement still win out. Instead of making short term adjustments to try and beat (or outsmart) the markets, focus on building a portfolio that works for you and that you can hold yourself to over the long term.

This is not an easy process – and it’s not always easy to stay disciplined even after you set up your long term portfolio – but spending the time to design an investment portfolio that fits both your risk tolerance and broader retirement plan will help you move towards the retirement that you’ve always wanted.

Trying to get one over on the market won’t.

To find out more about how to build an investment portfolio that works for you, read our eBook 9 Principles of Intelligent Investors.

1. For the stats folks, the p-value on the difference is 0.13. For the non-stats folks reading the footnotes, that means that there is (basically) at 13% chance that there is no real difference in the average returns, and it is just there by chance. Typically, people use 5% as the cutoff for when something is “statistically significant” though there’s no magic to that cutoff. But 13% is pretty high to reject the possibility of chance. ↩︎
2. Even though the elections are in November, I’m ending the campaign season in October because I don’t want to get into the effects of the results of the election[link]. ↩︎
3. They are different, just very close. The standard deviation for Non-Election Years was 17.827%, and for the Non-Campaign season in election years it was 17.834%. ↩︎
4. At a better than 2.5% level of significance. ↩︎
5. It’s a rabbit hole – but a fun rabbit hole. ↩︎