According to founder and CEO Jeff Bezos, the long-term future for Amazon is that it will fail one day. In one sense, this statement is not remarkable.  Nothing is forever.  Eventually, every company will fail (except, maybe, too-big-to-fail banks).  But I wonder what this says about the buy and hold investment strategy.  After all, if Amazon continues to pay no dividend and eventually goes bankrupt, then all investors, even those who bought early and currently enjoy enormous mark-to-market gains, will have realized total returns of minus 100%!

Does this mean that buy and hold strategies are doomed to failure?  Not necessarily.  Rapidly growing companies typically pay no dividends, but as they mature and the rate of growth slows, they tend to begin paying out cash to shareholders in the form of dividends and/or stock buybacks.  Even if the value of the stock eventually reaches zero, the rate of return for the poor fellow holding on the bitter end may not be disastrous.    For example, suppose Amazon is able to continue growing operating cash flows at a rapid rate (25% per year) for the next 30 years, at which point the company fails and the stock price goes to zero.  Further, suppose that 10 years from now Amazon begins paying out half of cash flow each year in dividends and continues doing so for the subsequent 20 years.  The internal rate of return to an investor buying the stock today at $1,600 per share would be nearly 15%.  That means that the investor would enjoy a doubling in value roughly every five years.

Perhaps this is an extreme example.  Rapidly growing companies generally do not pay out dividends or buy back stock because internal investment offer higher returns.  It is only after internal opportunities begin to wane, and growth begins to slow down, that companies typically begin paying out cash.  Thus, the assumption that Amazon begins paying out 50% of cash flow while growth remains robust is a bit of a stretch.  Still, the point is that returns can be substantial even for companies that eventually fail.

Generally, failure rates are greater for young companies with potentially greater growth rates.  Thus, if you seek to participate in the monster potential gains of young startups, you’ll have to contend with a very high failure rate.  If you go for large mature companies, the failure rate is smaller but likewise is potential growth.  What is an investor to do?

Some insight into these issues is provided by Hendrik Bessembinder, a professor of finance at Arizona State University.  The professor has compiled the monthly returns for every stock that has traded on US exchanges since 1926.  He has used this database to calculate lifetime returns for each stock.  His overall conclusion is while that most individual stocks do not generate returns exceeding Treasury bills, there is sufficient positive skew in the return distribution such that broadly diversified portfolios do generate excess return (that is, return over Treasury bills).  The lottery-like feature (high probability of failure of individual stocks) is much greater for small stocks – specifically, the probability that a small stock (first decile of market capitalization) will outperform T-Bills is less than half the probability that a large stock (top decile of market capitalization) will do so.

It appears that a key lesson is this:  diversify!  While important for any sector, the importance of extensive diversification is much greater for small growth companies than for large dividend paying companies (unless, of course, you like to play the lottery).  One way to achieve extreme diversification is to invest in one or more broad passively managed index funds.

While index funds provide an excellent way to obtain market returns at low cost, I prefer what has been termed the “coffee can” approach.  The coffee can idea is to purchase a diversified portfolio of individual stocks (not funds) and then hold on forever.  What are the advantages of the coffee can?  No fees, possible tax benefits (the ability to sell individual stocks to secure tax losses as needed), and the possibility of giant gains on one or more of the individual securities (such gains might never be realized in a fund due to rebalancing).  But the coffee can approach seems particularly vulnerable to the phenomenon noted by Bezos.  If you reinvest the cash flows from this portfolio into the same stocks, and they all eventually fail, then you will have secured the ignominious result of minus 100% return.  To avoid this outcome, it may be necessary to use cash flows from the original coffee can portfolio to build additional such portfolios.

To examine the viability of coffee can portfolios, I set up a simple experiment wherein I simulate portfolio returns for two sectors:  large, dividend paying companies and small growth companies.   For each sector, I assume a dividend payout (that is currently being paid for large companies and is deferred for small companies), a growth rate of cash flow (modest for large companies and quite high for small companies) and an annual probability of failure (low for large companies and high for small companies).  I calibrate the assumptions so as to be consistent with the historical individual stock returns reported by Bessembinder.  Given the basic assumptions I simulate lifetime returns for portfolios of 1, 10 and 100 stocks (the simulation proceeds by randomly drawing, for each company in each year, from a binomial distribution representing probability of default; once a company “defaults” all cash flows cease).

As expected, the returns on 1-stock portfolios are hugely variable, for each sector.  The same is true for 10-stock portfolios.  However, for the large company sector, 100-stock portfolios exhibit small variation in lifetime returns.  Thus, a 100-stock large cap coffee can be viable.  This is not the case for the small company sector.  Even portfolios consisting of 100 stocks show lottery-like returns.  This suggests that even in up markets, your 100-stock portfolio is likely to exhibit significant underperformance.  If small caps or startups is your sector of choice, you should seek the greatest diversification you can find.  One implication is that so-called “Angel Investors” – that is, affluent people that support a small number of startups – are likely to experience extreme variation in results.  To me, this is yet another reason for all of us to applaud the rich people who are willing to take on the huge risk of investing in new companies.  A great example of this is Bezos himself.