People assume all sorts of wild stock market returns when they make their financial calculations. Here are some numbers that show up on web searches:

6 %

8.4 %

10 %

10.1 %

11.3 %

11.5 %

13.6 %

16 %

These are all correctly computed under their respective assumptions, but they are very misleading because whatever those assumptions were, they’re not relevant for most calculations. You should assume 4 % in your calculations. Here’s one way to arrive at that:

• The S&P 500 has historically yielded 12 % per year before inflation.
• But that’s the arithmetic mean. Returns compound geometrically, so what we actually want is the compound annual growth rate. The typical rule of thumb is to remove half the squared volatility from the arithmetic mean. Volatility is notoriously hard to estimate, but let’s put it at 15–30 %. This leaves us with a cagr of 8–11 %.
• During the same period, inflation has been 3–4 % depending on which source you trust. This brings the real (inflation-adjusted) returns down to 4–8 %.
• On top of that, we have holding costs, transaction costs (e.g. rebalancing), and taxes. These are different in different localities and vary by investment strategy, but let’s say they remove 1–3 % of return. The net cagr is, then, 2–6 %.

In the middle of all our assumptions we find 4 %. That’s a much more reasonable number.

Comments

2026-05-25: Jesse Onland

I think the thesis of this post is correct, but involves some important errors.

1. The S&P 500 is not the stock market. This is a very common error but a serious one. The S&P 500 is an index of large-cap US equities. The stocks in the index account for only about half of the global public equity market by market cap. There are reasons to think that large-cap stocks have systematically different return characteristics than stocks in general.
2. It is best not to try to account for tax and transaction costs in a headline figure like this because these differ considerably from investor to investor.

pwl Capital, a Canadian wealth management firm focused on evidence-based investing, has a nice page on estimating the equity premium.

(Jesse Onland has a few data science projects and a very impressive and intriguing link collection at their website.)