The Buy and Hold Apocalypse

A Time to Measure

Buy and Hold

The rising chorus of doubts about the "buy and hold approach" has recently raised concerns about the viability of long-term investing. Various piles of statistics are being peddled in the wake of the recent market correction to support the notion that stocks are quite dangerous and that investors need to reconsider the buy and hold hypothesis. The Introduction of the Buy & Hold Apocalypse? explored what the term "buy and hold" really means, concluding the bashers have failed to clearly define what they were critiquing before issuing blanket condemnations.

Before we can really understand what the "buy-and-hold approach" is all about, we have to talk about the relative safety of investments in the stock market over various periods of time. The presupposition of the buy-and-hold approach is that over "long" periods of time, stocks are the single best place to have your savings. While this is an attractive hypothesis, it is important to prove this is the case before we either prove or disprove the validity of the "buy-and-hold approach."

The first step in proving that the market is the best place to have your money over long periods of time is to decide what period of time to examine. When should we look at the aggregate returns of stocks, bonds and inflation in order to construct a comprehensive and realistic argument that you should buy stocks for the long haul? After we figure out where to get our data we need to then look at it in a comprehensive and logical fashion in order to determine whether the evidence supports our claim. This involves pinning down a more exact definition of the "long" in "long-term" in order to figure out what the minimum holding threshold of the "buy-and-hold approach" should be.

When should we begin to measure stock returns? Better put, when do the returns of stocks count for the purposes of our analysis? In Stocks for the Long Run, Jeremy Siegel pushes our data about the aggregate returns of equities back in time to 1802, a fascinatingly remote moment in the history of capitalism. The Ibbotson Associates 1996 Yearbook correctly points out, however, that "the stock market [only] became a liquid secondary market in the Twenties." As we are dealing with a liquid secondary market today, this suggests that although this pre-Twenties data should be examined, it should not be viewed as authoritatively as the post-Twenties data. The procedure we will follow is to look at each set of data consecutively in order to determine the least amount of time you have to "hold" in the "buy-and-hold approach" in order to guarantee outperforming all other asset classes.

The data in Stocks for the Long Run is impressively comprehensive and is the first stop in our analytical tour. Using the 1871-to-1992 time period to construct his data, Siegel looks at the returns of stocks, bonds and T-bills relative to one another. The reason why he includes T-bills is because the lower rate of return in T-Bills serves in effect as a proxy for the returns of a money market fund. First, Siegel found that between 1871 and 1992, stocks outperformed bonds 100% of the time, despite the fact that the market was defined as illiquid for more than a third of the time period he was measuring. Over twenty-year periods, stocks outperformed bonds a remarkable 94.17% of the time. Over five-year periods, stocks outperformed bonds 71.19% of the time. Over one-year periods, stocks have outperformed bonds more than 59.02% of the time.

Since stocks have never failed to outperform bonds over periods of thirty years or more, I will initially conclude that the longest you have to hold for the buy-and-hold approach to work is thirty years. Also, we have seen that the odds of outperforming bonds in a period as short as one year are substantial, although we cannot derive any sense of what the total return might be, including all of the down years. What this does suggest, however, is that on an annual basis the stock market goes up a little less than two-thirds of the time -- not exactly bad odds. Over five year periods you have the market up three out of four times. Further out, the odds just keep getting better. The historical data still confirms the validity of the buy and hold hypothesis, with longer and longer holding periods improving the chance of beating a similar return in bonds.

Now we gotta nail down what the "long" in "long term" really means. What kind of total returns can we expect out of stocks over these time periods, even if they are positive? I take up the first part of this question and continue to trim down the minimum holding time in the next installment of Buy & Hold Apocalypse.