The Buy and Hold Apocalypse
No Moss on a Rolling Return
Buy and Hold
The "buy and hold" approach has received the harshest attacks recently from investors preoccupied by skewed data which suggests that in recent history buy-and-hold candidates have caused investors ruin. These critics love to pull out all sorts of statistics proving that investors who invested at the height of the market at various points in history took exceedingly long periods of time to break even again. Much of this silliness is calculated off the cuff using straight index values for the Dow Jones Industrial Average or S&P 500 -- a process we know is flawed because indices do not measure total return -- they just measure price-weighted or market capitalization-weighted capital appreciation. (See the Evening News from July 29th and July 30th for more details on this claim.)
What is the best way to disprove this notion? To examine the actual aggregate returns of stocks since the advent of the liquid secondary market and to find out the actual returns, while isolating a class of securities that can be used as a proxy for the typical "buy and hold" stock. (Readers who have been following this series will remember that we defined a "buy and hold stock" in our Introduction to the Buy & Hold Apocalypse as one whose current market value was low, relative to its underlying economic value.)
We can look at the aggregate returns of all stocks on a rolling basis since 1926 by simply opening up the Ibbotson Associates1996 Yearbook, a compendium of market facts available in the business section of many libraries. The Yearbook looks at the returns of "large company" and "small company" stocks, long-term bonds, Treasury Bills (our proxy for a money market) and inflation over various rolling periods from 1926 to 1995. A "rolling" period is where you run consecutive time series data that overlaps. For instance, if you were measuring ten-year returns you would go from 1971 to 1980, 1972 to 1981, 1973 to 1982, etc., in order to get rolling returns. This process enriches the data you have to work with when you are dealing with a relatively short period of time.
In yesterday's Evening News, we established through Jeremy Siegel's data that you were guaranteed to have a positive return that was superior to that of bonds if you held stocks for thirty years or more between 1871 and 1992. The rolling return data in the Ibbotson Yearbook over a more recent and relevant time period shows that "large company" stocks produced positive returns in 56 out of 56 overlapping periods, while "small company" stocks did the same in 53 out of 56 periods. Small company stocks were the highest-returning asset in 44 out of the 56 periods and large company stocks were the highest or second highest returning asset in 52 of the 56 periods. Putting this information together makes a pretty convincing case that our "holding" period can be whittled down to 15 years from 30 years to outperform bonds or other financial assets -- a great boon to many baby boomers who are starting now to save for a retirement 15 or 20 years away.
Although not quite as overwhelming, the ten-year returns for large company stocks and small company stocks are heartening. Both had positive returns in 59 out of 61 overlapping ten year periods, with the *single* worst ten-year period for large company stocks measuring a total -0.89% return and small company stocks getting crunched for -5.70% in its worst ten-year period. It would stink to be flat for ten years, but this is hardly the crisis scenario that market timers love to paint. I mean, if the worst ten years in history has you basically even (not including inflation), it is hardly a catastrophe if you have continued to put new money into the market in subsequent periods, enjoying the high rate of return that often follows market underperformance (i.e. late 1987, 1975 and 1976, etc.).
For the universe of all stocks with five-year rolling returns, things get a little less sure. "Large company" stocks have positive returns in 59 out of 66 rolling five-year periods and "small company" stocks have positive returns in 57 out of 66 similar chunks of time. If the general trend is the further out, the less risk you take, this makes sense. It is interesting to note that the return is positive in both situations about 90% of the time. Either "large" or "small company" stocks outperformed all other asset classes for 55 out of the 66 periods measured, implying that the compounded total return for stocks was better than all other asset classes almost 83% of the time over five year periods -- not exactly terrible odds. And these odds can be improved greatly if one recognizes the common causes of market corrections and adjusts one's fundamental screening criteria for those time periods (which we will discuss in the last episode of this series on Friday, Market Timing Is Bunk.)
If we tighten up our stock selection requirements and enter the realm of "buy and hold stocks" as defined previously, the results become even more impressive. Recently I wrote that an investor who bought "value", i.e. "buy and hold", stocks and held them for five or ten years over the past forty years would have done well. A commentator on the message boards called this proposition ludicrous, beginning to give the bull and bear market talk that clearly identified him as a market timer. He first disputed my claim by stepping outside the time period I circumscribed, referencing the returns from the top at 1929 to 1950. He then further claimed that from January 1965 to December 1975 stocks averaged -1.0% a year, from November 1968 to October 1978 stocks averaged 2.30% per year and finally that from December 1972 to July of 1982 stocks averaged 3.90%.
Fortunately, I have compiled the stock by stock returns for the thirty Dow Jones Industrial Average stocks between 1961 and 1996 on an annual basis and have some hard data to dispute these claims. (This 35-year spreadsheet is available in FoolMart for a measly twelve bucks, by the way.) Measuring the thirty Dow stocks as a portfolio, assuming equal amounts invested like real investors do, yields significantly different results than what were quoted. For the January 1965 to December 1975 period (actually eleven years, incidentally) the return was more like 5.1%, not the -1.0% quoted. And, for 1965 to 1975 the two rolling ten-year returns are 1.83% annualized and 3.97% annualized, with inflation at 5.2% and 5.7% respectively, on an annual basis for the same period.
It is important not to take these returns out of context, however. Given that T-Bills (our proxy for money markets) returned 5.7% and 5.6% over the same period, the possibilities of cash were not all that staggering either. One of the great fallacies promulgated by market timers is that corrections in the stock market happen completely outside of the entire monetary and economic background, something any true student of the data recognizes is patently untrue. The important fact to focus on here is that despite the tumultuous 1973-1974 period, a ten-year holder would have at worst been about even and would have had average annual returns within two to four percent of a money market. Assuming any ability at all to select stocks, the returns would have been much higher. We will take the Dow Dividend Approach 10 as a proxy for adequate stock selection, as a four-year could pick these. (For more information on the Dow Dividend Approach, check out our Dow Dividend Approach area.)
Frankly, I think the single greatest logical error of the market-timing, scare-mongering crowd is assuming mean performance when simple approaches like the Dow Dividend 10 are out there. Even over the devastating 1965-to-1974 period, a Dow 10 portfolio would have returned an average 5.85% and would have outperformed both T-bills and bonds. Come again? It's true... get the spreadsheet and check for yourself. And furthermore, the whole "long-term bonds are safer 'cause you are guaranteed your principle back" school of thought needs a head check as well. If you are tying up money in long bonds for ten to thirty years in order to guarantee the return of principle, you might as well be in stocks. And you should definitely not pretend that selling your bonds before maturing will not cause a loss -- the same economic factors that cause stocks to fall in value don't help bonds all that much either most of the time, as we will see in Market Timing Is Bunk.
For you aggressive investors who want to focus on five-year returns, the worst five-year period since 1961 for a Dow 30 portfolio was 1970 to 1974, where you would have had a -0.87% average annual return. Not great, but no disaster. Particularly when you consider the Dow 10 over the same period would have logged a 7.09% performance. In fact, looking at the rolling return data (which you would have to construct using the data on the spreadsheet) that is the only negative five-year period in the past 35 years, buttressing my initial claim that a buy-and-holder in buy-and-hold stocks would have done fine during the most devastating market decline in this half of the century.
NOTE: It is important to realize that the returns for large and small company stocks that I have been using for the past two days from the Ibbotson data are something I view with a little skepticism. The reason for this is a phenomenon called extinction. Some argue that the returns for small company stocks have been pumped up by researchers looking back over historical databases to calculate them because of the fact that bankrupt companies are routinely erased from database records. This means that the returns for the companies which totally blew it and lost all of their shareholders' money have been erased, biasing the returns. As a disproportionate amount of bankruptcies occur among small stocks, this pushes the historical returns of small stocks up a notch or two. I personally believe that the total return of small and large stocks has actually been about the same since 1926 after you factor all bankruptcies back in -- which would be around 10.0%.