Recently, I posted an article called What rate of return should you assume for your retirement plan. My conclusion was to use multiple returns because it is near impossible to accurately predict future returns. Despite that conclusion, a number of people went on to email, tweet or comment giving me their predictions on what return was appropriate. Regardless of the fact that we cannot predict future returns, people still try to do it. Maybe it’s fun so here’s my attempt with some data from Morningstar’s Paltrak program.
Annualized returns can be misleading
Everywhere you look, you can find the 1, 2, 5, 10 year numbers but this annualized return data can be misleading because it looks at performance as a ‘snapshot’ in time. Think of it as a click of your camera – one click equals one snapshot or a single moment in time. Take another picture a few minutes, hours, days or weeks later and the picture represents a different moment in time.
Here’s a snapshot of returns as of the end of June 2011:
|1 Mth||3 Mth||1 Yr||3 Yr||5 Yr||10 Yr||15 Yr||20 Yr|
|DJ Industrial Average TR CAD||-1.6||0.6||18.5||4.2||2.0||-0.4||5.2||9.1|
|MSCI World GR CAD||-1.9||-0.1||19.1||-0.6||0.0||-0.1||3.4||6.6|
|NASDAQ 100 PR CAD||-2.5||-1.4||21.6||6.2||5.0||-2.1||6.1||10.8|
|Russell 2000 C$||-2.8||-2.4||24.9||5.8||1.1||1.5||4.9||8.9|
|S&P/TSX Composite TR||-3.3||-5.1||20.9||0.2||5.7||8.0||8.9||9.4|
If you look at the 1 year numbers, the stock markets have done pretty well for that 12 month period. But look at the 3, 5, and 10 year numbers and it can be pretty depressing. In case you didn’t know, any stock market is a risk and has ups and downs.
If you look at this same chart at a different point in time (different snapshot) all the numbers will look different.
Trailing returns is better data
Since snapshot returns can be misleading, the better way to look at performance data is to look at trailing returns. Essentially this is like looking at performance using a video camera instead of a still camera. Here’s an example of the TSX showing trailing 10 year periods:
What you will see here is all of the 10 year returns on the TSX since Jun 1986. The first 10 year period shown is the first bar which represents Jun 1986 to Jun 1996. The next bar shows Jul 1986 to Jul 1996. The next bar shows Aug 1986 to Aug 1996. this continues every month until the most recent 10 year period from Jun 2001 to Jun 2011.
The best 10 year period was from Aug 1990 to Aug 2000 which produced a 10 year annual compound return of 15.59%. The worst 10 year period was quite recent from Aug 2000 to Aug 2010 where the 10 year annual compound return was only 2.84% which was lower than a risk free GIC return. The average 10 year return was 9.23%.
Can we use this data to forecast future returns?
Even on a 10 year basis, the stock market produces unpredictable results. In fact other market indices produced more volatile results:
- best 10 year return = 17.28%
- worst 10 year return = -3.79%
- average 10 year return = 7.24%
- best 10 year return = 22.25%
- worst 10 year return = -2.92%
- average 10 year return = 11.12%
In the end, you have to have some belief that the stock markets will produced some returns over the long term otherwise you won’t be in the market. However, I’m not sure being able to predict a number with any degree of accuracy is possible. It may be fun but not possible.