Mark Twain popularized the phrase “there are lies, damned lies, and statistics.” There is some validity to that statement when it comes to investing. All data reported in the financial media and promoted by investment salespeople is of necessity historical in nature. After all, if you’re trying to recommend an investment, the only data that you can point to short of pure speculation is its past performance. The problem comes in when selecting the time frame of the data to be used.

Case in point: In May Jeff Sommer, the markets, finance and economy columnist for The New York Times, reported that long duration bonds over the past 20 years had a better average annual return than stocks. More specifically, Treasury bonds with a duration greater than 10 years returned over 8% on average each year since 2000 while the S&P 500 return was only a bit over 5%. Even long corporate bonds gained more than stocks. His conclusion was that such a “reversal of the customary bond-stock performance is deeply troubling. It is a sign … of how risky the markets have become and of how difficult it is to invest sensibly for the future.” While I don’t disagree that investing is certainly difficult in today’s environment, if he had looked at the data over a different time frame he might have reached a very different conclusion.

Ben Carlson at Ritholtz Wealth Management confirmed Sommers results but compared them to the 18 year returns starting in June of 2002, only two years later. He found that during the latter time frame the S&P 500 outgained long Treasury bonds 395% (in aggregate) to 275%. By comparison he calculated the aggregate returns for Sommer’s 20-year period as 194% for the S&P 500 and 428% for long Treasuries.

Carlson speculated that the reason for the poor stock performance since 2000 is simply that the valuation of the S&P 500 that year was the highest in its previous history. At the same time long government bonds were yielding 6.5% and rates were declining to their lowest levels in history (which is very positive for bond returns). Whatever the reason, the point is that a minor difference in time frames resulted in a major difference in outcomes.

Sommers’ conclusion from his analysis is that this is a “strange time” for investors but you should still invest in bonds, stocks, and cash because the future is unpredictable. Carlson concluded that based on the frequency with which long-term bonds outperform stocks – only 9% of the time over every twenty-year period since 1926 – stocks are still the best bet for growth over the long-term (just not every long-term) and that diversification remains of paramount importance.

Excepting Sommer’s view that the investment landscape is “strange” today, I totally agree that stocks remain likely to provide greater growth than bonds (albeit with greater concomitant volatility) and that diversification remains critical to investment success. But I would add that when comparing investment choices, be very careful to understand the time frame and context from which any historical data is used to favor one over another. Data may not lie, but it can deceive!