The time was August 1982 and the stock market indexes as measured by the Dow-Jones Industrials and the S&P 500 were marking new lows at 777 and 102, respectively. A busy day on the NYSE would see 60 million shares traded (versus two billion 20 years later). The back-to-back recessions of the early 1980s were just ending; Fed Chairman Paul Volcker had in the previous two years used double-digit interest rates to wring inflation down from an average annual rate above 13% in 1980 over 1979 to around 3% for 1983 over 1982.
The bond market had already bottomed after sliding for more than three decades.
In addition, the personal computer, introduced circa 1975 in the form of a kit (the Altair), had finally been developed to the point where ordinary people,
0.8 1.2 1.6 2.0 2.4 2.8 3.2 3.6 4.0
84 85 86 87 88 89 90 91
Nikkei 225
S&P 500 DJIA
Fig. 2.3 Nikkei 225 versus S&P 500 and Dow-Jones Industrial Average normalized weekly prices, 1984–1990 (see also Ziemba and Schwartz (1991, p. 186) for a similar chart going back further and Loeys and Panigirtzoglou (2006), who illustrate five-year real price returns for Japan’s Topix and for the S&P 500. The five-year real price return on the S&P 500 from 1905 to 2006 had never exceeded 30%, whereas the Topix slightly exceeded this return in the late 1980s)
0 40 80 120 160 200
70 75 80 85 90 95 00
U.S.
Japan Market cap % of GDP
Fig. 2.4 Total stock market value (capitalization) as a percent of respective GDP, United States (NYSE, AMEX, and NASDAQ) and Japan (TSE, all section listings) 1970–2000. (Data sources: US Department of Commerce, World Stock Exchange Fact Book, 2004, Bank of Japan, University of Hong Kong. See also Hall (2001). The Wilshire 5000 calculation of Fig. 1.4 shows the same timing and directional progression but a lower peak)
0 20 40 60 80 100 120
10,000 15,000 20,000 25,000 30,000 35,000 40,000
80 82 84 86 88 90 92 94 96 98 00
Nikkei 225
Nikkei Index Land prices, Nationwide
Land price Index
Fig. 2.5 Japanese real estate price indices, 1990 = 100, all national, 1980–2001 and semi-annual average for Nikkei 225 index. Tokyo prices (not shown) roughly tripled between 1985 and 1988 (see also Ziemba and Schwartz (1991, p. 189) for 1955–1990 commercial land price index). (Sources: Siebert (1999, p. 9); also, Japan Real Estate Institute, Bloomberg Index series, JPNLPTALL and JNLPTRES, and Datastream)
not solely hobbyists, began to find uses for it in playing primitive games and adopting the first VisiCalc spreadsheets. Just the year before, in 1981, it had been a major news event when IBM introduced its first personal computer product, effectively endorsing the technology. But although the technical the- ories and implementations of large-scale communications networks had already been developing for 20 years, nothing resembling the Internet as we know it today was then conceivable.41
The earlier mini-boom in tech stocks—ignited by the introduction of per- sonal computer products from IBM and Apple as based on new microprocessor and memory-storage chips developed by Intel—soon fizzled. By 1984—the year of the famous anti-Big Brother (i.e., anti-IBM) television commercial that had been produced by Apple and shown at that year’s Super Bowl broadcast—
investors and speculators alike had begun to shun tech shares even though unit sales of personal computers had already multiplied by 20-fold in five years to 15 million annually. Overcapacity in chip production had led to severe price cuts and rapidly declining revenues and profits.
However, with the bond market spurting to new highs, interest in stocks had by 1986 generally returned. The US long-bond (8%) futures easily broke through par (100) and reached prices not seen in more than a generation. By 1987 confidence had risen to the point that the S&P 500 had more than tri- pled from its 1982 closing low of 102.42 to 336.77 (closing on August 25, 1987). The Dow-Jones Industrials had meanwhile topped at 2722.42 (closing price on August 25, 1987) as compared to the 776.92 closing low just five years before on August 12, 1982.
For institutional investors, confidence was further bolstered by the wide- spread use of what came to be known as “portfolio insurance,” peddled by advisory firms as a way to insulate portfolios from serious loss using a strategy of selling (shorting) proportionately more and more futures positions as the markets declined. Individuals too were again being attracted to the technology sector as Microsoft’s highly successful launch as a public company just the year before heralded the start of an era in which millions of personal computers would be sold as far into the future as anyone could see.
But for reasons not yet fully understood the market couldn’t hold onto its gains. The ensuing crash of 1987 culminated on October 19—on a then all- time record trading volume of 604.3 million shares—with a one-day decline in the Dow-Jones Industrials of 508.32 points (22.6%) that percentagewise was almost twice the 11.7% drop of October 29, 1929. Declines of comparable magnitudes were also seen around the same time in other important stock mar- kets: As measured in local currency units, drops ranged from 11.4% in Austria to 45.8% in Hong Kong, where the market was thereafter shut for a week. The FTSE-100 Index lost 10.8% on that day and another 12.2% the following day.42
Jacobs (1999, pp. 183–202) explains the event as follows:
[S]ynthetic portfolio insurance in 1987 performed a role similar to that played by margin speculation in the crash of 1929. Just as margin buying elevated the bull
market of the 1920s, portfolio insurance increased demand for stocks in the 1982–7 period. And just as the automatic, trend-following stock sales of portfolio insurers exacerbated the 1987 crash, so the trend-following stock sales forced by margin calls accentuated the decline in stock prices in 1929…In the wake of the crash, portfolio insurance vendors blamed… – everything but the strategy itself…During the crash, transaction costs skyrocketed. Liquidity dried up…Portfolio insurance failed just when it was most needed.
And Henriques (2017, p. 3) characterizes the event as one in which Wall Street players:
…became far more homogenized, subscribing confidently to academic theories that led giant herds of investors to pursue the same strategies at the same time with vast amounts of money…It is not an overstatement to say that Black Monday was the first modern crash, the first to spotlight…fundamentally new risks.
The losses in the Dow-Jones Industrials and the FTSE-100 are illustrated in Fig. 2.6.
Yet much to the surprise of almost everyone, the market’s sudden plunge did not foreshadow or lead to any serious problems in the real economy: The DJIA, crash and all, nevertheless actually ended 1987 slightly higher than where that year had begun (with real GDP gaining 3.4% in 1987 and another 4.1% in 1988).
The market turbulence of 1987 thus only briefly interrupted the rise in tech sector share prices. With the 1989 fall of the Berlin Wall and the presumed end of the Cold War in sight, investors began to talk of a “peace dividend” in which defense spending would significantly decline as a percentage of GDP and thereby liberate and redirect capital for uses in the civilian sectors.
1,600 1,800 2,000 2,200 2,400 2,600 2,800
Jul 87 Aug 87 Sep 87 Oct 87 Nov 87 Dec 87 Jan 88 DJIA
1,400 1,600 1,800 2,000 2,200 2,400 2,600
Jul 87 Aug 87 Sep 87 Oct 87 Nov 87 Dec 87 Jan 88 FTSE-100
Fig. 2.6 The October 1987 crash illustrated for DJIA (left) and FTSE-100, July 1987 through January 1988, daily
In retrospect, the Gulf War period of 1990–1991 was merely another pause in the climb to higher valuations: The first mid-January 1991 night (January 14) that the United States began to attack Iraqi command and control centers was the first night that a massive rally in stocks and bonds began. By the end of 1991, the DJIA was pushing toward new all-time highs near 3200. Only two years later, it had reached 4000. And, thanks to the invention of browser soft- ware, the Internet was starting to become commercially useful and to fire up the imaginations of investors everywhere. The following summarizes key tech- nology events.43
1975 First PCs on sale 1993 Mosaic browser released
1978 Apple II introduced 1995 Netscape IPO
1981 IBM unveils its first PC 1997 Amazon.com IPO
1986 Microsoft IPO 1999 Priceline.com IPO
1991 World Wide Web created 2000 NASDAQ tops 5000
Still, there can be no doubt that passage by the US Congress of the Telecom Act of 1996, which was presumed at the time to be highly deregulatory, was important: It unleashed, and seemed to justify, the spending of countless hun- dreds of billions of dollars on telecom and cable investments, many later seen to be of dubious merit. And it led to unprecedented spending on Internet- related advertising in all media. The Standard, now defunct but at the time one of the top magazines of this genre, in the year 2000 sold 7400-plus ad pages, more than any magazine in US history; a typical issue had the heft and size of a small-city phone directory.
Fund management companies meanwhile began to merge into much larger units, each time diminishing the diversity not only of opinion but also of port- folio and fund management structures.44 Emphasis within brokerages also turned more toward rewarding sales, promotional, marketing, political, conference- organizing, and investment-banking skills than toward provision of insightful, penetrating analysis that might disagree with mainstream views.45
Fund managers, moreover, generally failed to recognize that earnings growth projections in support of higher future prices were largely a function of the bubble itself. And analysts would hardly ever look beyond their customary comparisons of stocks and bonds on the basis of relative (e.g., yield spreads) rather than absolute values. A stock would be labeled as “undervalued” if its various metrics fell below those of a similar company; for instance, a price-to- earnings ratio of 50 would appear inexpensive compared to one of 70.
Meanwhile, the whole equity market’s valuation was itself aligned with and derived from comparisons to a bond market then in the midst of its own liquidity- driven bubble.
In March 1999, AOL “with a stock market capitalization of US$140 billion was worth more than Walt Disney, Viacom and CBS combined, and well over twice as much as General Motors.” The ten largest market cap stocks of the S&P 500 were in early 2000 trading at a p/e ratio—more accurately, a price- tofantasy ratio – of 62.6.46
Such inflated values relative to historical norms appeared in other major markets too. In London, just around the peak, several high-admission-price conferences and courses promised to instruct investors on how to account for the differences between conventional and new valuation-model results. Terra Networks, a spin-off of Spain’s Telefonica, sported a capitalization of €25 bil- lion despite an absence of profits. Share prices would soar by just adding dot- com (or an e or I) to the name of a tired, old company or by reporting financially irrelevant data (e.g., number of web-page views).47 Hong Kong police were required to control crowds applying for the vastly (many hundreds of times) oversubscribed IPO of an Internet company with almost no revenues, no prof- its, and a website still under construction.48
At the time it was indeed not unusual for companies with virtually no reve- nues to be valued at billions of dollars. Unlike the situation in the classical model of perfect competition in which abnormally high profits attract new entrants, in the TMT (telecom, media, technology) bubble new entrants appeared even though there were no profits to begin with.49 Here, the catch phrase leading to a successful fund-raising campaign was “burn rate.” This sup- posed indicator of potential future growth suggested the speed at which a new enterprise was burning through its initial capital and depleting cash—the higher the rate, the higher the valuation—rather than as a metric of time-to- bankruptcy, the normal and correct interpretation. Such a flight from quality in the bubble phase is, of course, the ironically opposite extreme of the frantic flight toward quality (and liquidity) that’s always seen in crashes.
By the peak, the bull market in tech stocks had taken the NASDAQ to an unprecedented price-earnings multiple of 245 as compared to a range of 15–30 for most of the NASDAQ’s existence since 1971. In the aggregate, Internet shares traded at an average of around 35 times revenues.50 The S&P 500 mean- while rose above 35 times earnings—more than twice its long-run average.
Yet rather than throttling back the frenzy with tighter credit policy, the Fed – in response to widespread “Y2K” concerns that the nation’s older computer system codes would catastrophically malfunction once the calendar turned into the new century year of 2000 – actually moved in the opposite direction. The Fed expanded the monetary base by 8.7% (a 17.4% annual rate) in the half year between the end of June 1999 and the start of January 2000.
All of this allowed telecom companies to raise trillions of dollars, with which they rushed to build expansive and expensive networks composed of millions of miles of fiber-optic cables buried under city streets and seas. The resulting capacity glut, wherein under 3% of the installed fiber was used as of 2002, quickly sent bandwidth prices down by an average of more than 65% and even- tually led to the bankruptcies and massive scandals of companies such as WorldCom, Enron, Global Crossing, and Qwest.51
The heights achieved by high-tech and telecom stocks as reflected primar- ily in the NASDAQ 100 Index as compared to the overall NASDAQ Composite Index and the S&P 500 Index are displayed in Fig. 2.7. As the three lines are all indexed (first week of 1995 = 1.0), it can be seen that at the
subsequent peak, the NASDAQ 100 rose by nearly twelve times, the NASDAQ Composite by around six times, and the S&P 500 by approximately three times.
This chart also helps to illustrate an intensity measure that broadly enables comparisons of all bubbles and corresponds to the crash intensity metric that is later developed (Sect. 3.3). As shown by Dent and Panchioli (2017, pp. 182- 192), the origin of the bubble is the point in time at which a linear upward trend begins to accelerate onto an exponential trajectory, For the NASDAQ 100 that would be mid-1996 when trading around 635 and rising to comple- tion after 3.75 years to a March 2000 peak at 4,398. Here, the peak-to-origin index ratio – i. e., its intensity – was approximately 6.9, which compares to the later housing bubble’s ratio of 4.3 as measured from mid-2002 at around 300 to an early 2005 peak at 1,300 (Fig. 2.9) and to a 1920s DJIA ratio of around 4.0. Subsequent crashes then almost always return prices close to the bubble origin’s linearlly extrapolated trendline, which is at price levels near to where the bubble began. Although many such events play out over a period of six or seven years, the downside phase will typically last around half the time that was expended in the preceding rise and end in collapse.
The severe pressure for professional investors to participate is reflected in the sector concentration displayed in Fig. 2.8. A comparison of the infotech and later housing bubbles then later appears in Fig. 2.9.
Evidently, the degree of the market portfolio’s diversification is not stable over time. And because of this, in a bubble episode, the market portfolio’s abil- ity to reduce exposure to firm-specific risk is markedly diminished. “The great- est benefits in risk reduction,” as Merton (1992, p. 31) has observed, “come from adding a security to the portfolio whose realized return tends to be higher
0 2 4 6 8 10 12
95 96 97 98 99 00 01 02
Nasdaq 100
Nasdaq Comp
S & P 500
Fig. 2.7 NASDAQ 100, NASDAQ Composite, and S&P 500 indexed weekly (first week of 1995 = 1.0), January 1995 to December 2002. (Source data: Yahoo Finance)
when the return on the rest of the portfolio is lower.” That does not apparently happen in the real world of a major index like the S&P where—particularly in crashes—directional correlations of index components tend toward 1.0 (Fig. 3.3). Often there is more risk for the same expected return or, equiva- lently, a lower return for taking the same risk.
As the TMT bubble illustrates, at a top most participants will actively ignore or ridicule “prudent” professionals proffering historical knowledge and wis- dom. Cautionary advice is considered to be worthless and detrimental to cur- rent-period performance: Naysayers are not at all heeded, let alone tolerated.
And in going against the grain of a bubble by moving to cash or short- selling,
“[E]arly looks a lot like wrong.”52
Most participants must thus by necessity either be oblivious to a bubble’s existence or be unable to resist participation as bubble-riding comes increas- ingly to be seen as a viable and rational investing strategy.53 As a leading tech- nology investor noted at the time, “You either participate in this mania, or you go out of business. It’s a matter of self-preservation.”54
By comparison to the previous periods, 1995–2000 was remarkable in terms of price gains, dividend yield declines, and total returns. Indeed, none of the other listed periods came close to what—even with major financial crises included—occurred in the last six years of the 1990s. Stock market capitaliza- tion as a percentage of nominal GDP, as displayed in Fig. 1.4, clearly showed that prior to 1997 (and even going back to 1929), this ratio had not been above 100%.
0 5 10 15 20 25 30
92 96 00 04 08 12 16
Info Tech
Financials
Health care
Energy
% of S&P 500
Fig. 2.8 Industry sector concentration (year-end) in the S&P 500, sector percent of total market value of S&P 500, 1989–2017. Note that an 11th sector for real estate was carved out from financials in 2016. (Source data: Standard and Poor’s)
2.7 housing, crEdiT, and coMModiTiEs, 2002–2008
Housing and Credit
Housing markets are usually analyzed somewhat differently than those for readily tradable shares or bonds. That is because housing transactions involve relatively large individual transactions, low liquidity and/or infrequency of trading, and absence of portability (i.e., immobility) of the underlying assets.
However, given that housing is a major asset class in which a worldwide bubble ultimately involved trillions of dollars more than even in preceding the tech stock episode, it is instructive to briefly review its history. (The IMF estimated that in the aftermath global GDP declined 1.5% in 2009.)
The basic underlying forces here (i.e., expansive credit followed by contrac- tion) were the same as always. Only with housing there was a much greater apparent impact on overall consumption patterns and prices: Swarms of unqual- ified but credit-crazed borrowers ended up buying vastly overvalued homes.55
The centrality of real estate-related lending to macroeconomic activity and the inherent systemic fragility that this entails is why, when housing bubbles burst, the ensuing damage is so great. Indeed, mortgage lending against the collateral value of a private home or commercial property is especially attractive to banks and other lenders because the value will normally be much more easily appraised than in most other asset classes (including capital investments for business enterprises).
Moreover, because asset values in economic expansions practically by defini- tion tend to rise, the outstanding mortgage loans usually become much more serviceable and collateralizable and thus of diminishing risk to lenders. In expansions, any defaults on real property debts also will then be much more readily handled via refinancings, foreclosures, and resales.56
The starting date for the main upleg of this event was late 2001, just after the tragic terrorist attacks of September 11 of that year. The collapse of the Internet stock bubble and the economic sluggishness that followed led the Federal Reserve Board of Governors under Chairman Greenspan to drop the target Fed funds rate to a low of 1% (June 2003) from a high plateau of 6.5%
(May 2000 to January 2001).57 The banking system responded in turn by aggressively marketing and extending on relatively easy terms loans of all types, including those for automobiles and credit cards.
Some of the greatest efforts were focused on providing mortgages to the previously untapped and far-from-creditworthy borrowers in the subprime market—that is, to those with no income, no jobs, and no assets (which were known as NINA, NINJA, or no-doc loans).58 Support for this came implicitly from politicians of all types and stripes and also from the American Dream Downpayment Act passed by the US Congress in 2003 for the purpose of creat- ing a program that would make it easier for the poor to secure first mortgages.
As a result, nearly half of mortgage originations in 2004–06 were of the default- prone adjustable-rate (ARM) type.59 For at least over the short term, the lend- ing industry found such loans to initially be quite profitable.