The reason for this is all very simple, which is that people have an immutable faith in the central tendency of a long-running time series. The problem is that stock market returns don't follow anything close to a normal distribution (CLICK ON PICTURE FOR LARGER IMAGE):
First of all, I would like to point out that there is a big difference between the compounded annual growth rate (CAGR) and the "average". This is because averages almost completely ignore the effect of significant down years. To explain it in a clear example, if you drop 50% in the first year and rise 50% in the next year, the two year average is 0%, but you're actually down 25%. The CAGR captures this, but the average does not. As such, the long term typical rate of return is about 140-150 basis points lower, depending on how you draw your time horizon. The difference is notable, by the way. At 7.9% per year over 40 years, $1,000 turns into $20,932. At 6.5% per year over 40 years, it's $12,416. In other words, please don't base any projections you are doing for your retirement on the average rate of return.
By my count, there is only one year that was +1% or -1% from the long-term CAGR, which was 1993. Hell, make that a 3% band and you still only get about 7 or 8 years, depending on how you round off trailing digits.
As you can see, this is not a normal distribution by any stretch of the imagination.
As such, when you see the annual forecasts come out toward the end of this year, feel free to snicker when you see people project that we will see the long-term average rate of return.