As others have said, its the volatility that kills a portfolio which relies on an average return being acheived every year. It would actually be unusual for a portfolio to produce the average return in any year. If you get below average returns in the early years the probability of failure (i.e. running out of money) goes up a lot because the reduced remaining principal has to produce ever higher returns to get things back on track.
To use an overly simplified example:
Start with $1000, and plan on an average return of 7% and a constant SWR of 4% of the original capital ($40) each year (i.e. no inflation adjustment).
If you get the expected average return of 7% each year, you will be left with $1030 at the end of year one ($1,000 +7% - $40) and $1062.10 at the end of year 2 ($1,030 + 7% - $40).
But if the markets have a bad year and you actually lose 10% in year one you are left with $860 ($1,000 - 10% - $40). In order to get to the same end position the year two return would need to be about 28.1%. This is possible but unlikely. Given that you will be spending principal each year, it is more likely that the hole will just keep getting bigger and the portfolio will fail.
Playing around with a spreadsheet will show that missing the average return by even a small amount a year for a couple of years will kill the portfolio.
Of course, the reverse is also true. A few above average years at the start of the portfolio's life will extend the duration of the portfolio significantly if you can resist the urge not to inflate your lifestyle.
There is a very good explanation here:
The Flaw of Averages
We're also facing a potentially very long period of retirement (40-50 years) and have decided that there is no such thing as a SWR for that length of time. We're also planning on me working for 1-2 years more after we hit the number to provide a buffer against the problem of below average returns in the early years.