The book’s premises are false and its forecasts have been wildly inaccurate. By exposing its flaws I reconfirm some basics of investment.
This year marks the 25th anniversary of James Glassman’s and Kevin Hassett’s book Dow 36,000: the New Strategy for Profiting from the Coming Rise in the Stock Market (Random House, 1999). It didn’t merely contend that stocks were grossly undervalued: it insisted that a four-fold increase of the Dow Jones Industrial Average (DJIA), from approximately 9,000 (its average early in 1999) to 36,000, was in the offing.
Specifically, Glassman and Hassett (hereafter “G&H”) contended (p. 19): “stocks should be priced two to four times higher – today.” “But,” they cautioned, “it is impossible to predict how long it will take for the market to recognise that Dow 36,000 is perfectly reasonable. It could take ten years or ten weeks.” In their very next sentence, however, they threw caution to the winds: “our own guess is … between three and five years …”
During the five years after the book’s publication, the attacks of 11 September 2001 occurred and the Dot Com Bubble burst. Did these shocks delay the realisation of G&H’s bold prediction? Did an even bigger and more unexpected (to the bulls) cataclysm, the Global Financial Crisis, further cloud and impede their prescience? Or had they been completely wrong all along?
Three things are certain. First, on 14 January 2000, a few months after the book’s publication, the DJIA reached an all-time, intra-day high of 11,723. Second, in 2002-2004 it plummeted as much as 35% (to as low as 7,592) and rose no higher than 10,783. Clearly, that was a far cry from 36,000! Although it subsequently recovered (indeed, it scaled a new record high of 14,165 in October 2007), by 9 March 2009 – a decade after the book’s publication – it had crashed 54% to 6,547.
Thirdly, not until November 2021 (that is, more than 20 years after G&H predicted it) did the DJIA briefly breach the 36,000 threshold; and not until December 2023 – almost a quarter-century after the book’s debut – has it remained continuously above it. On 16 May 2024 it reached 40,000.
Was Dow 36,000 a product of its era – that is, of the irrationally exuberant bull market and Dot Com Bubble of the mid-1990s to early-2000s? Or, despite various trials and tribulations, was it actually – as Kenneth Rogoff contends (“Why the Dow 36000 Forecast Was Right,” The Wall Street Journal, 9 September 2021) – far ahead of its time?
Others’ Assessments
Appraisals of the book vary enormously. Hamilton Nolan (“World’s Wrongest Investment Guru Still Thinks His Big Prediction Might Come True,” Gawker Media, 7 March 2013), mocked it as “the most hilariously wrong investment book of all time.”
In response to the DJIA’s substantial decline during the decade and more after the book’s publication, Nolan asked sarcastically: “what did the authors … do? Flee the country? Retire in shame? Give all their remaining assets to those investors unfortunate enough to have followed their advice? Ha, no. James Glassman is still writing op-eds and waiting for his prediction to come true” (see, for example, “Dow 36,000 Is Still Attainable Again,” Bloomberg View, 8 March 2013).
Hassett, too, remains prominent. Moreover, he’s potentially very powerful. Before 1999, he was an economist at the Fed. Since 1999, he’s been a Senior Fellow and Director of Economic Policy studies at the American Enterprise Institute, and subsequently served as economic advisor to the presidential campaigns of John McCain (2000 and 2008), George W. Bush (2004) and Mitt Romney (2012). He was also Chairman of the Council of Economic Advisers in the Trump administration from 2017 to 2019. According to The Wall Street Journal (“Trump Economic Advisers Float Three Names for Fed Chair,” 18 March 2024), Hassett has been shortlisted as a possible head of the Federal Reserve if Donald Trump wins the presidential election in November.
Paul Krugman, Distinguished Professor of Economics at the Graduate Center of the City University of New York (formerly at MIT and Princeton University), winner in 2008 of the Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel and long-time columnist in The New York Times, was also scathing: “when all is said and done, Dow 36,000 is a very silly book; and the attempts of Glassman, at least, to deny the patent silliness are borderline dishonest. But I am prepared to view Hassett’s role as youthful indiscretion” (see “Dow 36,000: How Silly Is It? Undated).
Other well-known and powerful people, however, lauded it. John Bogle, who in 1999 was “senior chairman” of The Vanguard Group and is now acknowledged as the father of index investing, said in a blurb on its dust jacket: “while there will be bumps – maybe big ones – along the way and the road may be surprisingly long, Dow 36,000 offers superb advice. With an eminently readable style, the authors present sound and simple wisdom about investment principles, mutual fund selection, index funds, and asset allocation. I am impressed!”
David Malpass, who was a senior official in the U.S. Treasury under presidents Reagan, George H.W. Bush and Trump, chief economist at Bear Stearns from 1993 until its collapse in 2008 and President of the World Bank from 2019 to 2023, also extolled Dow 36,000. In 1999 he gushed: G&H’s “ideas are timely and thought-provoking. Either we are in a bubble with inefficient financial markets, or else past theories on stock prices and price-earnings multiples have to be revised. In every one of my meetings with mutual funds these days, I have to address the issue of whether stocks are overvalued. G&H’s theories make the solid case that, on average, they are not.”
Some people loved G&H’s book; others loathed it. To my knowledge, however, nobody has subjected its key assumptions, claims, inferences and predictions to detailed scrutiny. That’s what I do in this article.
A Preview of My Evaluation
Dow 36,000 makes a few sound (indeed, astute) points. Yet its foundation is rotten: its claim that stocks are no more risky than bonds is mostly false. Indeed, in three of the four respects that I consider, equities are riskier than bonds.
Not surprisingly, given its weak base, G&H’s predictions – whether about individual stocks or the DJIA – have been spectacularly wrong. Most fundamentally, and ironically, although they insisted that they were calculating rationally and promoting sensible investment, they actually thought and acted like many people do during booms, bubbles, bull markets, manias, etc.: they abandoned common sense and embraced wishful “New Era” thinking.
G&H spurned “outdated” (p. 4) measures and standards of valuation and attempted to rationalise the mania of the late-1990s. Above all, they extrapolated into the indefinite future the extraordinary returns of the recent past. Dow 36,000 wasn’t ahead of its time; it was an irrationally exuberant product of its time.
The Case for Dow 36,000
On pp. 17-18, G&H summarise their argument. They present it as ten distinct propositions; I’ve condensed it into three sets of contentions, each of which is based upon a key concept.
Risk
- “Over the long term, a diversified portfolio of stocks is no more risky, in real terms, than an investment in bonds issued by the U.S. Treasury.” (Indeed, on p. 4 they assert that equities are less risky than bonds.)
Equity Risk Premium
- “Stocks have historically paid shareholders a large premium – about seven percentage points (sic) more than bonds …
- “This equity (risk) premium (ERP), based on the erroneous assumption that the (stock) market is so risky that anyone who invests in it should get a higher return as compensation, gives investors a delightful unearned dividend.
- “Evidence abounds that investors are catching on, realizing that the equity (risk) premium is unnecessary (that is, should before long fall to zero) …”
Valuation
- “If there is no ERP, then, over time, stocks and bonds should put about the same amount of money into the pockets of the people who buy them.
- “Therefore, the correct valuation for stocks – the perfectly reasonable price (PRP) – is the one that equalizes the total flow of cash from stocks and bonds in the long run.
- “Several complementary approaches show that the price to earnings (“P/E”) ratio that would equalize (stocks’ and bonds’) cash flows is about 100.
- “The Dow Jones Industrial Average was at 9,000 when we began writing this book, and its P/E was about 25. So, in order for stocks to be correctly priced, the Dow should rise by a factor of four – to 36,000.”
Prospective Returns
- “The Dow should rise to 36,000 immediately, but to be realistic, we believe the rise will take some time, perhaps three to five years …
- “In other words, stocks are an exceptional investment. They are (no more risky than Treasury) bonds over long periods of time, and their (prospective) returns – at least for now – are exceptional.”
My Assessment
Risk
“In the long run,” G&H emphasise (on p. 24 and elsewhere throughout the book), “stocks are not very risky.” The authors acknowledge that “people are naturally cautious, especially with their own money, and the return on stocks is volatile from day to day.” However, they omit to mention that equities’ returns can also fluctuate greatly from week to week, month to month, year to year and even over multiple years.
G&H then ask (p. 25): “what is risk, anyway? … In simple terms, think of financial risk as the volatility of returns – the severity of the ups and downs of a stock’s (or a bond’s, etc.) performance … The most common way to quantify risk (p. 26) is through an analysis of ‘standard deviation,’ which tells how much a stock’s return has varied from its own average” (see also pp. 95-96).
“All risk indicators,” they rightly add, “have two drawbacks. First, they can only describe what has happened in the past, not what will happen in the future … Second, … risk indicators don’t account for cataclysmic events … But despite its inadequacies, history is the best source of information we have when analyzing risk, and the history of stocks is clear and consistent: in the short run, stocks are very risky; in the long run, they are no more risky than Treasury bonds.”
That last bit is demonstrably false: by the very measure which quantifies G&H’s definition of risk, and over medium-term (five-year), long-term (ten-year) and very long-term (20-year) periods, stocks’ returns are more volatile – that is, stocks are riskier – than bonds’.
Using monthly data compiled by Robert Shiller (which G&H cite at several points), I calculated the five-year, ten-year and 20-year compound annual growth rates (CAGRs) of the S&P 500 Index’s (hereafter “stocks”) and the 10-year Treasury bond’s (hereafter “bonds”) total (that is, distribution or dividend as well as capital gain or loss) real (adjusted for the Consumer Price Index) returns.
I then computed the standard deviation of each of these six CAGRs. As a third step, I separated the data into two eras: (a) all months before December 1996 (when Alan Greenspan delivered his “irrational exuberance” speech) and (b) all months since January 1997; finally, I computed the standard deviations of stocks’ and bonds’ real total returns during each era. The results appear in Table 1.
Table 1: Standard Deviations, CAGRs of the S&P 500 and 10-Year Treasury Bond, Medium, Long and Very Long Terms
In one crucial respect, G&H are correct (p. 27): “the longer you hold on to stocks, the less volatile are your returns …” That’s true for the entire (153-year) period under consideration, as well as the longer (125-year) and shorter (28-year) sub-periods: reading down each column, as time increases the returns’ standard deviation (risk) falls. Equally, however, and just as importantly, this pattern also applies to bonds.
Yet G&H’s core contention is clearly false: comparing the two “Entire Period” columns, the two “Until Dec 1996” columns, etc., there are no exceptions: over medium, long and very long terms, and whether the intervals are until December 1996 or since January 1997, the bonds’ standard deviations are lower – that is, the bonds’ returns fluctuate less and are thus, by this definition, less risky – than are the stocks’.
Moreover, since 1996 the standard deviations of stocks’ returns haven’t decreased (that is, stocks haven’t become less risky), but bonds’ clearly have. Over the past 30-odd years, in other words, bonds have become significantly less risky relative to stocks.
Figure 1: Standard Deviations of 20-Year CAGRs, S&P 500 Index and 10-Year Treasury Bond, January 1911-March 2024
Figure 1 elaborates this result. It plots the standard deviations of stocks’ and bonds’ 20-year total, CPI-adjusted total returns (expressed as CAGRs). Since 1911, stocks’ long-term return has averaged 2.0% and bonds’ 1.5%. Their returns fluctuate one-third more than bonds’ returns (2.0% ÷ 1.5% = 1.33); alternatively, bonds’ returns vary just three-quarters as much as stocks’.
Each series has tended to regress towards its mean. As a result, although bonds’ long-term returns usually fluctuate less than stocks’, over two intervals the opposite has been true. The first lasted from 1911 to 1921, and the second from 1990 to 2009. For a decade before and a decade after G&H wrote Dow 36,000, bonds were riskier – that is, their returns fluctuated more – than stocks.
But since 2009, the usual pattern, whereby stocks’ returns fluctuate more than bonds’, has reappeared. This is the product of two developments: first, the standard deviation of stocks’ 20-year CAGRs has zoomed; secondly, the standard deviation of bonds’ 20-year CAGRs has plunged.
As a result, since the GFC stocks haven’t merely become much more risky relative to bonds: they’re presently riskier than at any time in the past.
It’s not an explicit part of G&H’s definition of risk, but it’s nonetheless sensible (indeed, it’s Warren Buffett’s definition) and they resort to it repeatedly: the greater is the probability that an investment generates a loss, the riskier it becomes.
Figure 2 plots stocks’ and bonds’ long-term, CPI-adjusted total returns. It shows not only that stocks’ CAGR (7.2%) is much greater than bonds’ (2.1%); in barely 1% of the 240-month intervals since 1891 have stocks generated a return less than 0% – and when they have, the return has been barely negative. Bonds, in sharp contrast, have generated negative long-term returns almost one-quarter of the time. Indeed, between 1951 and 1985, their long-term returns were almost continuously and significantly negative. Today, they’re on the verge of dipping below 0%.
Figure 2: CPI-Adjusted, 20-Year Total CAGRs, S&P 500 Index and 10-Year Treasury Bond, January 1891-March 2024
In this key sense, stocks as a whole are much less risky than bonds. As a result, who in his right mind would ever buy and hold bonds?
For each month since 1871 I computed stocks’ and bonds’ CPI-adjusted one-month total return. I then annualised these monthly returns and divided them into two portions: (1) those months which, according to the National Bureau of Economic Research (which is the semi-official arbiter of the business cycle in the U.S.), occurred within a recession, and (2) the months between recessions. I then computed returns in each subset of observations. Figure 3 summarises the results.
Figure 3: Total CPI-Adjusted Returns, Annualised, S&P 500 Index and 10-Year Treasury Bond, During and Outside Recessions, February 1871-March 2024
During months outside recessions, stocks (13.6% annualised return) greatly outperform bonds (0.4%). During months within recessions, on the other hand, the polar opposite occurs: stocks generate annualised losses of -8.8%, whereas bonds gain 7.8%.
Why buy and hold bonds and related instruments such as bills and deposits? Among other reasons, they greatly mitigate equities’ losses during recessions (see also Recessions usually crush shares – but investors can always reduce their ravages, 31 October 2022, How we’ve prepared for the next bust, 28 November 2022 and How we prepare for – and profit from – recessions, 18 August 2023).
Figure 2 buttresses what G&H dub the “Dow 36,000 Theory,” and a graphic like it underpins Jeremy Siegel’s Stocks for the Long Run: The Definitive Guide to Financial Market Returns & Long-Term Investment Strategies (6th ed., McGraw Hill, 2022).
These results aren’t false, but without careful elaboration they’re misleading. Over the long term, the returns of stocks as a whole exceed those of bonds; yet most individual stocks underperform bonds. That’s because stocks’ overall long-term outperformance – and thus, say G&H, lower risk – is attributable to just a handful of companies.
In “Do Stocks Outperform Treasury Bills?” Hendrik Bessembinder demonstrates that most don’t: “58% of common stocks have (long-term) holding period returns less than those on one-month Treasuries … the entire gain in the U.S. stock market since 1926 is attributable to the best-performing 4% of listed stocks.” And in “Shareholder Wealth Enhancement, 1926 to 2022” he added: “the degree to which wealth enhancement is concentrated in relatively few stocks has increased over time: for example, the number of high-performing firms that explain half of the net wealth creation since 1926 decreased from 90 as of 2016 to 83 as of 2019 and to 72 as of 2022.”
Got that? The total return of the S&P 500 Index since 1926 results from the gains of just 20 companies. The vast majority at a given point in time (480) don’t merely underperform the Index: they underperform Treasury bills! Moreover, this sharp bifurcation has increased as time has passed. Why buy and hold Treasuries? Over the long term they outperform most stocks.
Figure 2 shows that in rolling 240-month intervals since 1891, a specific group of 500 stocks (i.e., those comprising the Index during each interval) virtually always outperform 10-year Treasury bonds (Siegel’s analysis extends this point to include long- and short-term bonds, gold and commodities). Unfortunately, this interpretation has been ubiquitously – and invalidly – extended to all portfolios, including those containing far fewer than 500 stocks.
This inference has contributed mightily to the universal but false belief that funds managers’ portfolios of, say, 150 stocks, as well as individual investors’ portfolio of 10-20, whatever their strategy and as long as they wait long enough, will outperform bonds. It has also obscured a fundamental truth: as I demonstrated in How the 60/40 portfolio outperforms (17 October 2022), a portfolio that prudently combines stocks and bonds often outperforms both.
On p. 91, G&H ask: “How risky is the stock market?” In the short run, referring to its mostly-random fluctuations, they correctly state that it’s very risky. In the long run, they assert, “it is no more risky than the market for Treasury bonds. That undeniable, historical fact forms the foundation of the Dow 36,000 Theory.”
My assessment is very different: the riskiness of stocks relative to bonds depends crucially upon your choice of stocks and conception of risk. The key strands include volatility of returns, probability of long-term loss, vulnerability to short-term factors such as recession – and choice of stocks.
By three of these four criteria, over more than 150 years in the U.S., stocks have been riskier than stocks. G&H’s inadequate characterisation and negligible assessment of risk provides a weak basis for the Dow 36,000 Theory.
Equity Risk Premium
“Investors in recent years,” G&H allege, “have begun to understand this secret about the riskiness of stocks” (namely that in the long run they’re supposedly no more risky than bonds). From this basic error stem three others (p. 91; see also p. 39):
- “As a result, they have bid up (stocks’) prices – a process that will continue until stocks reach their PRP (perfectly reasonable price).
- “Another way to describe what’s happening is that investors, acting rationally at last, are requiring a lower ‘risk premium’ – a smaller bonus – in return for taking the chances involved in investing in stocks.
- “Not only are investors becoming more enlightened about the riskiness of stocks, (their) real-life riskiness is actually going down …”
To ascertain the return you can expect from a stock or market index, G&H say (p. 96), you add its dividend yield in a given month to the percentage rate of dividend growth you anticipate during the next 12 months. “That number,” they state, “has typically been higher than the yield on a (10-year) Treasury bond …”
In January 1999, when G&H were writing Dow 36,000, the S&P 500 Index’s dividend yield was 1.3%. They assumed that during the next 12 months dividends would grow 6.0%. Hence the Index’s “expected cash return” for the year to come was 1.3% + 6.0% = 7.3%. Meanwhile, in January 1999, 10-year Treasury bonds yielded 4.72%.
The difference between G&H’s expected cash return and the bond’s yield – that is, the ERP – was 7.3% – 4.72% = 2.58%.
That number was “the expected extra cash flow that investors (demanded) to compensate them for the extra risk of owning stocks instead of Treasury bonds.”
That’s sensible: stocks’ long-term returns usually fluctuate more than bonds’; moreover, and unlike bonds, stocks usually fall – often heavily – during recessions; finally, most individual stocks underperform the Index As compensation for these risks, investors demand an equity risk premium.
According to G&H, on the other hand, investors perversely “perceive additional risk (from stocks) even if it doesn’t exist, so they demand higher returns. But what happens if these investors wake up to reality? What happens if their … perceptions shift as they learn the truth about stocks? To put it simply: stocks’ prices will move toward the perfectly reasonable price (PRP) …”
“The (equity) risk premium,” they elaborate, “is the number of percentage points you have to add to the Treasury bond interest rate in order to make it equal to the dividend yield for stocks plus the dividend growth rate” (p. 99).
As G&H note, there’s “no official risk premium figure.” Given a key assumption, it’s easy to construct one. It’s hard, however, to ascertain if that assumption is reasonable: “it’s easy to find the bond rate and the dividend yield, but we have to guess what people expected the growth rate of dividends would be each year.” For simplicity, G&H assume that the expected growth rate has been constant (6% per year) since 1871.
Since the Second World War, G&H’s assumption has been reasonable, but until then it significantly overestimated dividends’ actual rate of growth. For each 12-month period since January 1871 I calculated the actual rate of the S&P 500 Index’s dividends’ growth. I then calculated these percentages’ average over rolling 120-month intervals. For January 1871-January 1872, for example, dividends grew 1.3%. And the 12-month periods to January 1882 (i.e., February 1872-February 1873, …, January 1881-January 1882) averaged 2.7%, and so on for each rolling 120-month period to March 2024. Figure 4 plots the results.
Figure 4: Two Assumptions about Dividends’ Rate of Growth, S&P 500 Index, January 1871-March 2024
Given the two assumptions in Figure 1, it’s easy to compute each month’s ERP. In January 1871, for example, the sum of the dividend yield (5.86%) and G&H’s estimate of dividend growth (6.0%) was 11.86%; subtracting the 10-year Treasury bond’s yield during that month (5.32%) gives their estimate of the ERP during that month: 11.86% – 5.32% = 6.54%, and so on for each month to March 2024. Figure 5 plots the results.
Figure 5: Two Estimates of the S&P 500 Index’s Equity Risk Premium, January 1871-March 2024
Before the Second World War, my estimated ERPs are lower and much more variable than G&H’s; since 1945, however, the two series are strongly correlated (R2 = 0.71).
Given my assumptions – but, it’s worth noting, not G&H’s – the ERP attained its all-time maximum (15.9%) in June 1932. At the nadir of the Great Depression, shell-shocked investors perceived that stocks, which since September 1929 had crashed 77% (net of CPI and including dividends), were extremely risky relative to Treasury bonds (which during the same interval returned, net of CPI, 38% in interest and capital gains).
Given both G&H’s and my assumptions, the ERP reached its all-time low in August 1981. That, too, makes sense. In that month, the S&P 500’s dividend yield was 5.0% and dividends’ estimated rate of growth during the next year was 6.0% (G&H) and 7.47% (Leithner). The problem was that the 10-year Treasury’s yield was very close (14.94%) to the all-time high (15.32%) it scaled in September of that year. Hence the ERP was 5.0% + 6.0% – 14.94% = -3.94% (G&H) and -2.72% (Leithner).
At that juncture, investors regarded stocks as safe relative to bonds because they regarded bonds as extremely risky (volatile and loss-making): during the preceding years, their CPI-adjusted total loss was 39.4%, versus stocks’ gain was 30.1%).
As G&H acknowledge, these results generalise: “stock prices are inversely related to the (equity) risk premium” (p. 98). As a rule, returns rise as the ERP falls; and as it rises, returns fall. Does a change of the ERP systematically affect stocks’ returns? Although it’s central to their case, G&H don’t substantiate it.
In each month since January 1876, I therefore (1) adopted G&H’s assumptions to compute the ERP. Next, I computed (2) its change and (3) the Index’s CAGR over the previous 60-months. I then (4) sorted the data according to the change of ERP, (5) divided the dataset into quintiles and (6) computed each quintile’s mean CAGR. Finally, I repeated steps 1-6 over 120-month and 240-month intervals. Table 2 summarises the results.
Table 2: the Effect of Changes of ERP upon Stocks’ Return, Three Intervals, 1876-2024
It shows very clearly that when ERP falls and the faster it falls (quintiles 1 and 2), stocks’ returns wax; and when it increases and the more rapidly it rises (quintiles 3-5), returns wane. These relationships hold over all three intervals.
In several key respects, however, the results in Table 2 and the evolution of both series in Figure 5 disconfirm the Dow 36,000 Theory – which stated in 1999 that the ERP would continue to decrease and stocks’ returns to increase.
The risk premium declined steadily from 1945 (when G&H’s estimate exceeded 8% and mine 7%) to 1981 (when G&H’s estimate approached -4.0% and mine exceeded -2.5%). From then until 2000, G&H’s estimate averaged 0.95% (versus an average of 5.8% since 1871) and mine 1.7% (versus its average of 4.0% since 1881). In other words, at the turn of this century equities’ perceived risks were little greater than bonds’. “We see this decline in the risk premium as reasonable and long lasting,” averred G&H (p. 101), “not as insane and transitory.”
Moreover, “to believe that the market is overvalued (in 1999), you have to believe that the ERP, once so irrationally large and (more recently) getting rationally small, will move back to that irrationally large state again. It is our strong belief – and the linchpin of our theory – that the risk premium will continue to shrink, and for good reasons. The best reason is (that) … stocks are no more risky, in the aggregate and over the long term, than bonds” (p. 102).
G&H were flatly wrong: over the quarter-century from 1999 to 2024, the ERP – calculated according to their definition – didn’t continue to drop. Quite the contrary: it rose – and, not surprisingly, given the results in Table 2, medium- and long-term returns sagged (see Figure 10a in the penultimate section).
“This is the Dow 36,000 Theory’s case for stocks,” declare G&H (p. 98): “they should rise in price as investors squeeze out the equity risk premium. If stocks and bonds are equally risky, as history shows they are (actually, and as I’ve shown, it demonstrates unambiguously that in two respects they aren’t, but never mind), then the risk premium … should be zero, and stocks will increase until they reach the PRP.”
The subsequent quarter-century has emphatically refuted these expectations: using G&H’s assumptions, the ERP rose as high as 7.5% in April 2020, and from January 2001 to March 2024 averaged 4.6%; by my estimate, it zoomed to a maximum of 11.2% in April 2020 and over these years averaged 4.6%.
“We have a very short response to concerns over a rising risk premium,” said G&H: “since stocks and bonds are equally risky over the long run, it is reasonable for the returns of stocks and bonds to be equal over the long run. The idea that stocks should return more than bonds is based on a fallacy – that stocks are more risky than they are” (p. 137).
Reality mocks G&H’s expectations: each $100 invested in the S&P 500 Index on 1 January 2000 – net of CPI, including dividends and ignoring tax – grew to almost $307 in March 2024. That’s a CAGR of 4.8% per year. Over the same interval, an equivalent investment in 10-year Treasuries grew to $140. That’s a CAGR of 1.4% per year.
“But let’s assume,” G&H muse, “that the risk premium goes back up – reverts to the mean.” That, of course, is exactly what’s occurred. What are its consequences? “Once more, investors will demand much higher returns from stocks than from Treasury bonds … Dividend yields will rise and P/E ratios will fall. The market could lose more than half its value” (p. 134).
Only one of these three things has transpired. In January 2000, the S&P 500’s dividend yield was 1.2%. Since then it’s fluctuated without trend; in March of this year, it was 1.3%. The market’s P/E ratio has fallen marginally, from 29 in January 2000 to 25 in March 2024. Above all, the market has soared more than 250% rather than plunged 50%.
Finally, G&H allege: “over the past two decades (1979 to 1999) investors have become more and more aware of this fallacy (that the ERP should exceed zero), so they are doing the reasonable thing – bidding up the prices of stocks. The risk premium is falling to zero because that is where it reasonably belongs. For investors to demand a higher risk premium is to behave unreasonably. It could happen, certainly, but we prefer to place our bets on reason rather than on unreason” (p. 137).
It’s quite amusing: when the future mocks their predictions, G&H won’t humbly admit they were just plain wrong; they arrogantly insist everybody else has been unreasonable and even irrational!
Implications for Prospective Returns: Individual Stocks
On pp. 41-42, G&H are sensible and orthodox. “The (net present) value of a business is the (net present) value of the stream of current and future dividends that it generates. Dividends are the only thing that matter. Period.” They cite a key passage from John Burr Williams’ seminal book The Theory of Investment Value (1938):
“Earnings are only a means to an end, and the means should not be mistaken for the end. Therefore we must say that a stock derives its value from its dividends, not its earnings. In short, a stock is worth only what you can get out of it” (see also Dividends aren’t a bane – they’re a boon, 20 November 2023).
“Remember,” G&H say (p. 68), “that we want (a) stock to provide the same flow of cash as a … long-run Treasury bond. The stock will do so if the sum of its dividend yield and the growth rate of its dividends is equal to the interest rate on that normal Treasury bond. If the sum is greater than the Treasury rate, then the stock is (underpriced).”
In Chapter 6 of Dow 36,000 (which carries the ironic title “A Conservative Look at How High the Stock Market Can Go”), G&H apply their ideas about risk and the ERP to concrete examples. They quickly cease to be sensible: “when we look back, the great bull market that began in 1982 seems only logical, considering the bushels of cash that stocks delivered to shareholders through the astonishing growth of dividends. Up was the only direction that stocks could go (p. 56).”
To anybody who bothers to look, the truth is clearly very much otherwise: but let’s not let facts spoil a good story! The S&P 500 Index’s dividends did indeed increase – from $6.81 in June 1982 to $16.45 in June 1999. That’s a 17-year CAGR of 5.3% per year. Adjusted for CPI, they rose from $21.89 in June 1982 to $30.86 in June 1999. That’s a 17-year “real” CAGR of 2.0% per year.
G&H provide several examples of dividends’ supposedly astonishing growth. The first, Wells Fargo & Co., paid an annual dividend (that is, the sum of four quarterly payouts) of $0.75 in 1999. “Between 1993 and 1998” (p. 61), it “increased its dividend per share at a rate of 16.5% annually (I make it 18.0%, but never mind); between 1988 and 1998, 14.5%. Those growth rates are solid, but Wells Fargo’s story is not unusual. There are hundreds of firms that have (done much the same).”
Again, G&H greatly overegg their omelette: unadjusted for CPI, the S&P 500’s dividends did indeed increase from $19.34 in January 1993 to $39.66 in January 1998. That’s a CAGR of 15.4% per year. Adjusted for CPI, however, payouts lifted from $27.14 in January 1993 to $30.00 in January 1998. That’s a CAGR of just 2.0% per year. Dividends rose mostly as a result of inflation – not executives’ brilliance.
No matter: G&H project their exuberance into the long-term future. “If Wells Fargo can sustain similar growth in the future,” they say, “then its dividend payments will become very big, very fast. Growing at 16.5% per year, that $0.75 dividend becomes a $1.61 dividend in the fifth year (2004). In the tenth year (2009), it becomes $3.45. In the twentieth year (2019), it is $15.91. In the thirtieth year, it rises to $73 …”
“But of course,” they grudgingly concede, “growth at 16.5% (per year) cannot go on forever.” Indeed, it’s “probably unrealistic, but if we are willing to make some assumptions about the growth that the company will be able to post in the future, we can predict the amount of cash it will generate in dividends – and from that figure, we can compute its (net present value discounted by the 10-year Treasury bond rate) today.”
Given various assumptions about the growth of its dividends (p. 62), G&H estimated that the NPV of Wells’ shares in 1999 was at least $117 and as much as $214. “Which is the most likely scenario? … Our own guess would be a P/E ratio of 150 (and thus a NPV of $180 per share). Is Wells Fargo special? Not at all! The stock universe is filled with companies that have stories that are at least as compelling” (p. 63).
With the advantage of hindsight, it’s obvious that G&H grossly overestimated the NPV of Wells’ shares (Figure 6). Unadjusted for dividends, etc., they attained their all-time high ($65.78) in January 2018; that’s just 37% of the NPV which G&H’s estimated in 1999. Alternatively, G&H’s NPV was almost 2.75 times its actual all-time high.
Figure 6: Share Price, Wells Fargo & Co., Nominal and Adjusted for Dividends, Buybacks, Etc., Monthly, January 1985-March 2024
Why did G&H miss so badly? They didn’t overestimate the growth of Wells’ dividend per share (DPS) since 1999. Quite the contrary: they predicted it very accurately. Figure 7 plots them on an annualised basis since 1972. DPS has grown from $0.38 in 1999 to $1.35 in 2024; that’s a 25-year CAGR of 15.4% per year. Since 1972, DPS has grown from $0.03; that, however, is a 52-year CAGR of just 7.6% per year. And since 2009, DPS has plunged, soared, plummeted again and zoomed again; as a result, its 15-year CAGR is effectively 0% per year.
Figure 7: Annualised Dividend per Share, Wells Fargo & Co., 1972-2024
G&H bungled Wells Fargo’s valuation so badly because they believe, ardently but largely incorrectly, that stocks as a whole are no more risky than Treasury bonds – and should be priced accordingly. But that’s hardly the only example that they botch. Fannie Mae is surely the worst.
The Federal National Mortgage Association, commonly known as “Fannie Mae,” was created as a government institution in 1938, became a private (albeit implicitly government-backed) corporation in 1968 and commenced trading on the New York Stock Exchange in 1970. According to G&H, it’s been “a great dividend-paying stock over the past ten years … In May 1999, Fannie was trading at $67 per share and paying an annual dividend of $1.08. Between 1988 and 1998, …, (it) increased its dividend at an astounding rate of 47% (I make it 40%, but never mind); between 1993 and 1998, at 30%” (I calculate 17%, but never mind).
Unable to restrain their enthusiasm, G&H “look at what would happen if the 30% (growth) rate continued for a little while … The $1.08 dividend would become $4 in five years, $14.89 in ten years, $205 in 20 years and $537,764 in fifty years – for one share! If the dividend grows at 30% for the next five years and then at 0.5% less than the real GDP rate, we calculate that Fannie Mae’s perfectly reasonable price would be $315 … If Fannie can manage its recent growth rate for ten years, the PRP will hit $903, or more than ten times the current (1999) price, for a P/E of 269.”
Of course, every word of the preceding paragraph is total rubbish. Fannie and the Federal Home Loan Mortgage Corp. (“Freddie Mac”) buttressed the U.S. mortgage market: in 2008, they owned or guaranteed half of America’s $12 trillion mortgage market – and the Global Financial Crisis, at whose heart lay bad mortgages, caused them to implode.
Figure 8: Share Price, Fannie Mae, Nominal and Adjusted for Dividends, Returns of Capital, Etc., Monthly, January 1985-March 2024
On 7 September 2008, the Federal Housing Finance Agency (FHFA) announced that it had placed Fannie and Freddie under its conservatorship. This action, deadpanned FHFA’s director, was “one of the most sweeping government interventions in private financial markets in decades.” FHFA sacked Fannie’s and Freddie’s CEOs and boards of directors, and forced them to issue new senior preferred stock and common stock warrants to the Treasury. These shares and warrants rendered worthless the common and preferred stock owned by pre-conservatorship holders (Figure 8); FHFA also suspended indefinitely the payment of dividends on previously outstanding stock (Figure 9).
Figure 9: Annualised Dividend per Share, Fannie Mae, 1983-2024
In 2008, FHFA stated that it didn’t intend to liquidate Fannie, and since then it hasn’t. It’s little-known but startling: today, with more than $4.3 trillion of assets, Fannie Mae is the largest company in the U.S. and the fourth largest in the world.
Fannie’s and Freddie’s pre-conservatorship shares now trade “over the counter” (OTC), meaning that you can’t buy them on a major stock exchange. Fannie’s shares’ price fell to an all-time low of $0.20 in December 2011. During the first quarter of 2024, they sold at an average price of $1.43. That’s a seven-fold rise from their all-time low. On the other hand, that’s a long way – to put it mildly – from $903!
The second column of Table 3 summarises G&H’s valuations in Dow 36,000. I’ve added these shares’ minimum and maximum prices since 2000; I’ve also included the shortfall between their maximum price after 1999 and G&H’s NPVs in 1999. The mid-point of Campbell Soup’s NPV, for example, is $141, and ($52.13 – $141) ÷ $141 = -63%. Finally, I’ve calculated these shares’ actual total return (expressed as 24-year CAGRs). In five of six cases, the all-time maximum share price doesn’t breach the lower bound of G&H’s NPV; hence the shortfall of NPV to maximum actual price is massive (average of -81%).
We all make mistakes, and over the past 25 years Leithner & Co. has made a few (by striving to minimise sins of commission, ours have mostly been sins of omission). But Table 3 summarises more than merely isolated and random errors: it demonstrates that G&H got things systematically – and egregiously – wrong: specifically, they grossly overestimated these companies’ NPVs.
Table 3: A Summary of G&H’s Valuations – and Their Results
Most importantly, the actual returns, expressed as CAGRs over 24 years, on average are woeful (1.8% per year). That’s little different from bonds’ return since 2000 (CAGR of 1.4%). Stocks’ weak returns strongly imply that, contrary to G&H’s emphatic and repeated assertions, in 1999 stocks as a whole weren’t dirt cheap. Quite the contrary: they were prohibitively dear.
Implications for Prospective Returns: Market Indexes
Stripped of all complexities, why, fundamentally, did G&H err so badly? “Our theory is rooted in the idea that investors have become more rational,” Glassman wrote in “Dow 36,000: The Case for a New Stock Market Model” (The Washington Post, 11 September 1999). “While many market analysts think that investors are currently off their rockers, we think that, to the contrary, they are finally becoming rational – recognizing the true value of stocks for the long term.”
The irony is exquisite: although G&H insisted that they were calculating and acting sensibly, the truth is that, as many investors do during every boom, bubble, bull market, mania, etc., they’d abandoned common sense and succumbed to euphoria.
They thereby spurned traditional approaches and embraced “New Era” thinking that attempted to rationalise the mania; above all, they extrapolated into the indefinite future the extraordinarily high returns of the previous several years. Figure 10a plots the S&P 500 Index’s CPI-adjusted total returns as rolling 60-month, 120-month and 240-month CAGRs since January 1982. Figure 10b plots them since January 1872.
In December 1999, three months after the publication of Dow 36,000, the five-year CAGR reached a level (25%) which it had rarely attained – one that had most recently occurred in 1987, and previously in 1937 and 1929. Similarly, in 2000 the 10-year CAGR matched the all-time record (16%) it set in 1928-1930. Finally, in April 2000 the Index’s 20-year CAGR reached 14%. Although it came close in 1961-62, that’d never occurred previously.
Figure 10a: Five-Year, Ten-Year and Twenty-Year CAGRs, S&P 500 Index, January 1982-March 2024
The crucial point – which G&H could have known in 1999, had they kept their wits and analysed data such as these – is that these three series regress strongly towards their long term means (which respectively, are 7.2%, 6.9% and 6.2%; Figures 10a and 10b plot the mean of these means, i.e., 6.9%).
Figure 10b: Five-Year, Ten-Year and Twenty-Year CAGRs, S&P 500 Index, January 1876-March 2024
The higher a CAGR rises above its long-term at one point in time, the greater is the likelihood that subsequent returns fall (“regress”) towards it. Conversely, the higher a CAGR falls below its long-term mean at one point in time, the greater is the likelihood that subsequent returns will rise towards it.
Accordingly, on an historical basis the exceptional returns of the late-1990s counselled caution – not exuberant rationalisations like Dow 36,000.
In “Why I Was Wrong About Dow 36,000,” The Wall Street Journal, 24 February 2011), James Glassman continued to flout logic and evidence. By that month, the S&P 500’s five- and ten-year CAGRs had sagged well below their long-term means; indeed, each had fallen into negative territory. Moreover, the five-year CAGR had mostly been below 0% since 2002. Before then, though, it hadn’t occupied negative territory since the bear market of the early-1980s – which preceded the bull market of the 1990s.
Given all that had occurred over the previous decade, psychologically 2011 was a difficult time to buy shares. Logically, however, it was an excellent time. But just as Glassman had succumbed to irrational exuberance in 1999, in 2011 he submitted to irrational despondency.
“Today,” he wrote, “the Dow Jones Industrial Average is just 20% higher than it was when Dow 36,000 was published in September 1999 … Even with dividends, annual returns over the past 11 years have been a few piddling percentage points. What happened? The world changed.”
The first major change, he reckoned, was that the relative economic standing of the U.S. was declining. The second one involved risk – “the kind that we can’t really measure or expect, (such as) the murder of 3,000 Americans by terrorists in a single day, the Dow losing 1,000 points within minutes in a ‘flash crash,’ or home values in the U.S. suddenly plummeting. These discontinuous risks – or ‘uncertainties,’ as the University of Chicago economist Frank Knight called them – are multiplying …”
“Perhaps I’m wrong about the world being a riskier place, but even if I am, this is still a time for investors to proceed with caution. They can protect themselves against the worst by ratcheting down the proportion of stocks they own compared with bonds, and by buying hedges such as ‘bear funds,’ whose prices go up if the market goes down.”
“This strategy,” he continued, “not only fits the reality and danger of our times, it also fits the psychology of investors. In theory, historical averages show that stocks are a good buy if you can hang on through the miserable periods. But most investors find that excruciatingly difficult to do – a fact that I never fully appreciated in my 30 years of writing about investing.”
“Fear,” Glassman concluded in 2011, “or simply a need for cash, triumphs, and people sell before stocks bounce back. I’ve gotten tired of telling investors to buckle up and hang on. Instead, I am urging them to adopt a more cautious strategy than the conventional financial wisdom – or Dow 36,000 – would dictate.” Just two years later, however, the five-year CAGR was zooming (in 2014 it reached 20%) and Glassman greatly changed his tune. Indeed, he expected his prediction in 1999 to vindicate him (see “Dow 36,000 Is Still Attainable Again,” Bloomberg View, 8 March 2013).
In 1999, Glassman was virtually certain that equities’ future returns would remain very high when, logically and empirically, he should have been very doubtful; conversely, in 2011 he was decidedly downbeat about stocks’ prospects when he should have been cautiously optimistic.
Conclusions and Implications
Since December 2023, the Dow Jones Industrial Average has exceeded 36,000. Reaching that threshold took 24 years, rather than D&H’s “guess” of 3-5. Moreover, what G&H confidently predicted would cause this rise – a continued fall of the equity risk premium – simply hasn’t happened. Indeed, quite the opposite has occurred.
Kenneth Rogoff (“Why the Dow 36000 Forecast Was Right,” The Wall Street Journal, 9 September 2021) is wrong: Dow 36,000’s foundation is unsound and its predictions have been wildly inaccurate – that is, uniformly and grossly overoptimistic. Yet clarifying its fatal flaws and thereby debunking it thoroughly is a useful exercise which reaffirms some fundamental truths. Most importantly, stocks are generally riskier than bonds.
This year marks the 75th anniversary of Benjamin Graham’s The Intelligent Investor: A Book of Practical Counsel (Harper & Row, 1949). In crucial respects, it’s the polar opposite of Dow 36,000. Unlike G&H, who chastise others as unreasonable or even irrational when reality mocks their predictions, Graham’s view of human nature is nuanced, realistic and based upon classical philosophy rather than contemporary mainstream economics and finance: “the investor’s chief problem – and even his worst enemy – is likely to be himself” (p. 8; ironically, G&H site this very passage on p. 152 of Dow 36,000).
However, and as Jason Zweig added in his comments to The Intelligent Investor’s 2009 edition, “by developing your discipline and courage, you can refuse to let other people’s mood swings govern your financial destiny. In the end, how your investments behave is much less important than how you behave” (p. xiii; see also Investing lessons from Benjamin Graham, 1 November 2020 and Successful investors are stoics, 23 October 2020).
Unlike G&H, who erred systematically on the side of overconfidence (recall Table 3, and see also Why you’re probably overconfident – and what you can do about it, 14 February 2022), Graham counseled prudence: “you must deliberately protect yourself against serious losses; you must aspire to ‘adequate,’ not extraordinary, performance” (p. 35). Furthermore, “while enthusiasm may be necessary for great accomplishments elsewhere, on Wall Street it almost invariably leads to disaster” (p. 5).
And unlike G&H, who disparage traditional wisdom when it contradicts their newfangled ideas, Graham observed: “no statement is more true and better applicable to Wall Street than the famous warning of (the Spanish-American philosopher and poet, George) Santayana: “those who do not remember the past are condemned to repeat it” (p. 1).
In his comments, Jason Zweig emphasised what Graham well understood but utterly eluded G&H: “the market is a pendulum that forever swings between unsustainable optimism (which makes stocks too expensive) and unjustified pessimism (which makes them too cheap). The intelligent investor is a realist who sells to optimists and buys from pessimists” (p. vii). Furthermore, and in Graham’s words, “speculative stock movements are carried too far in both directions, frequently in the general market and at all times in at least some of the individual issues” (p. 31).
The implication is crucial: the intelligent investor understands that stocks generally become more risky, not less (that is, their prospective returns sag), as their prices rise – and less risky, not more, as their prices fall.
Graham’s intelligent investor, writes Zweig, thus “dreads a bull market, since it makes stocks more costly to buy. And conversely (so long as you keep enough cash on hand to meet your spending needs), you should welcome a bear market, since it puts stocks back on sale” (p. 17).
James Glassman and Kevin Hassett weren’t ahead of their time. Instead, Dow 36,000 was a product of its era – the irrationally exuberant bull market and Dot Com Bubble of the mid-1990s to early-2000s.
Given its rotten foundation and wildly overoptimistic predictions, Dow 36,000 has rightly been mostly forgotten. But investors worthy of the label will long continue to study and apply The Intelligent Investor.