Also, the chances are high that you can find some time grid for which "every single decade" will show whatever you want it to show. For example, between 2000 and 2010, US aggregate stock market performance is... roundabout zero?
As one of my mentors once told me, "Optimize for the case where we aren't fucked, because if we are, there's nothing we can do about it." He was talking about software architecture and startups, but it applies to finance as well.
Survivorship bias basically is a sampling error and, in this case, leads you to underestimate the risk of your investments, both of which are never a good thing. Moreover, in almost all cases, it's a tad fatalistic to say "there's nothing we can do about it". I mean, maybe if the world goes under, but anything short of that, I'd like to think there's always sth I can do about it (even if that's just trading one bias for another... ;)
The reason for this is that if your home country undergoes political turmoil, you'll need to special-case that anyway. It's very likely that your assets will be seized by force; it's fairly likely that you will be too, and potentially lose your life. Your priority is staying alive and getting out of the country, not preserving $X million in wealth. It's highly unlikely that the country you emigrate to will be able to recognize any claims you might have had on assets in a country in turmoil. The best you can do is to get out and have sufficient friends & skills to rebuild your life.
That's like saying "I'll buy only stocks that go up".
You didn't know, in 1900, that the 20th century would be America's.
I'm going to guess that you'll live for at least another decade, and tell you to go right ahead and consider that the long run. If you've any multiples of decades, you're even more set.
But that is factually incorrect. "stocks have been the best asset class investment in each decade for the last two hundred years" That is a totally incorrect statement.
To be proper I should post specific statistics rebutting this, but I don't have them at this time. If you are genuinely interested you can look it up yourself.
I suppose Asia might buck the trend but a portfolio highly skewed in that direction now is a fairly speculative investment.
They may not own many shares, but they are invested in stocks.
http://en.wikipedia.org/wiki/Business_cycle
> This is extrapolating from a pretty small data set
I wouldn't call the last century of the US stock market a small data set. Perhaps what you mean is that the author's choice of decade-long windows is too arbitrary?
But regardless, the business cycle is a broadly accepted and easily observable phenomenon. Exactly what causes these cyclic fluctuations is hotly debated (at least it was ten years in my macro-econ classes.)
> it's quite possible average returns in the 21st century will not be as good as average returns in the 20th century
The article wasn't arguing that one should expect those exact returns, but rather that positive returns are cyclical-- good times generally follow bad times.
No, it wouldn't. It probably wouldn't even beat inflation.
But nonetheless, you can still do better analysis than these kinds of eyeball-the-chart analyses. Does this prediction of cyclic returns hold up under some attempt to retroactively test it, e.g. via cross-validation?
And of course that's ignoring the bigger problem, that it's quite possible average returns in the 21st century will not be as good as average returns in the 20th century were, making any extrapolation too optimistic.