Wednesday, January 29, 2014

Stocks for the Long Run

Jeremy Siegel’s book Stocks for the Long Run is in its fifth edition and has so much fascinating information for investors that keeping this review to a reasonable length is a challenge. A major theme running throughout the book is how different stocks look in a long-term view versus a short-term view. Despite Siegel’s deep analysis, this book is very accessible for readers with an interest in investing.

In the Foreword, Peter Bernstein asserts that “stocks must remain ‘the best investment for all those seeking steady, long-term gains’ or our system will come to an end, and with a bang, not a whimper.” Early in the first chapter, Siegel shows a remarkable chart of U.S. asset class returns adjusted for inflation over 210 years. A dollar in stocks grew to over $700,000 from 1802 to 2012 (even after subtracting out inflation). In contrast, bonds grew to only $1800 and gold to a measly $4.52.

The swings in stocks feel large when you live through them, but on a 210-year chart, stock gains look steady and relentless. “Since World War II ... the longest it has ever taken an investor to recover an original investment in the stock market (including reinvested dividends) was the five-year, eight-month period from August 2000 through April 2006.”

The conventional wisdom is that stocks are riskier than bonds. This is true in the short-term, but Siegel shows that for holding periods of 20 years or longer, real stock returns have been less volatile than bond returns! And unlike bonds, “stocks have never given investors a negative real return over a 20-year horizon.”

It’s normal for risks to steady out somewhat over time, but this happens for stocks faster than random chance would predict “because of the mean reversion of equity returns.” In contrast, bond returns steady our more slowly than random chance predicts, which means that bonds show “mean aversion.” So, over long holding periods, stocks offer higher expected returns and lower volatility than bonds, which makes a strong case for owning stocks.

For the rest of this review, I’ll discuss some of the many interesting parts of the book.

On geographic diversification

“Some argue that the increased correlation between the world’s stock markets reduces or even eliminates the incentive to diversify one’s portfolio.” However, “correlations are usually calculated over relatively short periods of time,” and long-term correlations are “significantly lower.” If you’re more interested in the state of your portfolio in a couple of decades than next month or next year, geographic diversification looks valuable.

Retirement age

“Who will produce the goods and services that the [longer-living] retirees will consume”? “If the developed world must rely solely on its own population to produce these goods, then the age that people will be able to retire must increase significantly.”

How can stocks rise faster than the economy?

“The reason that total return grows faster than stock wealth is because investors consume most of the dividends paid by stocks.”

Stock returns of other countries

Some say that the rise of the U.S. to dominate the world is the reason for the impressive history of U.S. stock returns, and that we cannot expect the same in the future for the U.S. or most other countries. In fact, since 1900, U.S. real stock returns have trailed those of Australia and South Africa. There are a total of 12 countries with over a century of real returns over 4% per year, and the world average is over 5% per year.

Efficient frontiers

If you don’t care about expected returns and just want to choose an asset allocation between stocks and bonds that minimizes risk, the result depends greatly on the holding period you consider. For 1-year returns, you get the lowest risk with 13% in stocks. However, for 30-year returns you get the lowest risk with 68% stocks.

Biased P/E ratios

“The traditional way of calculating the P/E of an index or a portfolio is by adding the earnings of each firm in the index and dividing this sum into the total market value of the index.” When some firms report large losses, we can get a “very distorted view of the index’s valuation.” Stock prices can’t go below zero. “Losses in one firm do not cancel the profits of another firm. Equity holders have unique rights to the profits of firms, unsullied by the losses in others.” So, beware of P/E measures of the entire stock market.

Value premium

With charts of stock prices from 1957 to 2012 broken out by quintiles of P/E ratio levels, Siegel shows the existence of a value premium for stocks. However, John Bogle found no evidence for a value premium when looking at returns from 1948 to 2008. Could the slightly different measurement periods be the explanation for the difference of opinion, or is there something else to this?

Investing based on economic growth

“There is a negative correlation between economic growth and stock returns, and this finding extends not only to those countries in the developed world but also to those in the developing world.” The chart showing poor returns in China is striking.

Inflation protection with stocks

“Since stocks are claims on the earnings of real assets—assets whose value is intrinsically related to the price of the goods and services they produce—one should expect that their long-term returns will not be harmed by inflation.”

Republicans vs. Democrats

“The stock market has actually done better under Democrats than Republicans.”

Technical analysis (charting)

A quote from Benjamin Graham: “A moment’s thought will show that there can be no such thing as a scientific prediction of economic events under human control. The very ‘dependability’ of such a prediction will cause human actions which will invalidate it. Hence, thoughtful chartists admit that continued success is dependent upon keeping the successful method known to only a few people.”

The January effect

It’s well known that for some reason, small-cap stocks have performed amazingly well in January for many decades. “Perhaps all the publicity about the January effect has motivated investors and traders to take advantage of this calendar anomaly, since the effect has largely disappeared since 1994.”

Berkshire Hathaway

Berkshire has beaten the S&P 500 by an average of “more than 10 percentage points per year” from 1972 to 2012. “The probability that this return was achieved by chance is less than one in a billion.”

Hindsight bias

Selective memory causes us to regret not acting on past hunches. “Hindsight plays tricks in our minds. We forget the doubts we had when we made the decision not to buy. Hindsight can distort our past experiences and affect our judgment, encouraging us to play hunches and try to outsmart other investors.”


This book has remarkably few typos. One small error occurs in an example of futures markets on page 277 where the figure 1410 should be 1710.


This book is very clearly written and offers powerful evidence for the advantages of investing in stocks. I highly recommend it to investors.


  1. Am I correct in thinking when he says "The longest it has ever taken an investor to recover an original investment in the stock market (including reinvested dividends) was the five-year, eight-month period from August 2000 through April 2006." he's talking about someone who "buys the entire market?" Because there were lots of people who put money in dot coms who never got any of their investment back because the companies poofed into hyperspace.

    So this might be another argument strongly in favour of full-market-index-investing?

    1. @Bet Crooks: Yes, he's talking about buying the entire U.S. stock market. There have been plenty of stocks that have gone to zero. Many speculators have lost money permanently. I agree that this is another argument in favour of low-cost broad indexing.

    2. That seems really short too. Was it using real returns?

    3. @Richard: Yes, it's using real returns. The question Siegel is answering is

      1. What is the longest time it has taken for real stock prices to come back up to where they are at a given starting point?

      This is quite different from asking

      2. What is the longest time it has taken real stock prices to permanently rise above the value they held at a given point in time?

      We can see the difference with an example. Suppose stocks are at 1000 in year X and start dropping. They return to 1000 in year X+4. However, they later drop below 1000 again in year x+7. It's not until year X+9 that prices rise above 1000 and never dip below 1000 again. The 4-year spread is relevant to Q1, but the 9-year spread is relevant to Q2.

    4. Aren't real stock prices below what they were in 2000?

    5. @Richard: Not when you factor in dividends. SPY (S&P 500 fund) trades right now at $178.08. According to Yahoo's historical prices adjusted for dividends, the highest dividend-adjusted SPY level in 2000 was $119.11. The current quote is 50.3% higher than the 2000 max. Inflation over that period has only been about 35%. Dividends matter.

    6. Ah, right thanks. If only everyone reported this consistently :)

  2. Very interesting. I have heard it said that it took 25 years for the stock market to recover after the 1929 crash. But that must mean nominal prices, without dividends reinvested, and ignores the point in the mid thirties when, because of deflation there was a temporary return to real new high. This is a reminder, for those in the withdrawal phase, who are using a "bucket" approach, to reinvest dividends when the market crashes so their equities get back to full value before their 5 years of cash/short term bonds runs out and needs replenishing. Mind you, one should never assume the unlikely (stocks taking longer than about 5 years to recover) is impossible.

    1. @Grant: The 5 years and 8 months maximum was just since World War II. I just poked through Robert Shiller's online monthly data. The peak in 1929 was in September. Counting inflation (deflation, actually) and dividends, purchasing power recovered in November 1936 (but soon plunged again). Without inflation, nominal prices with dividends recovered in December 1944. Ignoring both inflation and dividends, prices recovered in September 1954.

  3. Interestingly, using an online calculator that uses Schiller's data, I found that the real peak with dividends in November 1936 was not reached again until January 1945, so the period of 8 years and 2 months stands as the longest period that real stock prices plus dividends took to regain full value following a crash including the Depression years.