stock market bubbles 1929 and 1999
DESCRIPTION
TRANSCRIPT
Bubbles and Busts:
From the 1920s to the 1990s
1999!
“History is continually repudiated.”
--Glassman and Hassett The Dow 36,000 (1999)
•What can we learn from history?•Booms and busts have common elements•Evidence weak that booms are solely fundamentals driven•Evidence weak that booms are mania-driven•Role of Federal Reserve is very limited and should not be pre-emptive
KEY ISSUES
• Boom– Fundamentals– Bubble Formation– Fraud
• Bust– Precipitating Factors– Financial Consequences– Real Consequences
• Policy Lessons – Monetary Policy – Financial Supervision
What’s a Stock Market Boom?
• “Booms” are relatively rare long upward swings that dominate any brief retreat.
• Annual data is the most appropriate frequency to identify a boom.
• An arbitrary criterion that picks out the popularly-identified booms: three consecutive years of returns over 10 percent.
What’s a Stock Market Crash?• October 1929 and October 1987,
universally agreed to be crashes, are used as benchmarks. In both cases, the market fell over 20 percent in one and two days’ time.
• The fall in the market, or the depth, is one characteristic of a crash. There was no sudden decline for 2000. Speed is another feature.
• Crashes are identified with windows of a day, a week, a month and a year to capture other declines.
Identifying a boom by (about) three years of annual returns
over 10%
• 1921-1928: 20, 26, 2, 23, 19, 13, 32, and 39%.
• 1942-1945: 11, 18, 15 and 30%.
• 1949-1956 18, 22, 15, 13, 2, 39, 25%.
• 1963-1965 17, 13, and 9%.
• 1982-1986: 22, 14, 4, 19 and 26%.
• 1995-1999: 27, 21, 22, 25 and 12%.
Crashes: Where the Dow Jones, S&P500 or Nasdaq
declined more than 20%
• 1903 • 1907 • 1917• 1920• 1929• 1930-1933• 1937• 1940
•1946•1962•1969-1970•1973-1974•1987•1990•2000.
Matching booms with crashes:
• Booms:
• 1924-1929
• 1942-1945
• 1982-1987
• 1995-2000
• Crashes:
• 1929/1930-1933
• 1946
• 1987
• 2000
Recovery after 1946 and 1987 was quick, so the closest parallel to the 1990s is the 1920s
Boom and Bust 1920-1933
0
20
40
60
80
100
120
1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933
Pea
ks =
100
Dow Jones Cowles Equally Weighted
Do Booms and Busts Share Common Features?
Do Booms and Busts Share Common Features?
Boom and Bust 1980-1990
0
20
40
60
80
100
120
1980 1982 1984 1986 1988 1990
Serie
s Pe
aks
= 10
0
Dow Jones S&P500 Nasdaq
Boom and Bust 1990-2003
0
20
40
60
80
100
120
1990 1991 1992 1993 1994 1995 1986 1997 1998 1999 2000 2001 2002 2003
Ser
ies
Pea
ks =
100
Dow Jones S&P500 Nasdaq
Do Booms and Busts Share Common Features?
Table 1: Characteristics of Booms and Busts
Peak Trough Drop Peak to Trough
(months)
Recovery to Peak
Date
1920s
Dow Jones Aug-29 Jun-32 -0.822 34 Nov-54
Cowles Sep-29 Jun-32 -0.849 33 Nov-53
Equally-Weighted Feb-29 May-32 -0.896 39 Sep-45
1980s
Dow Jones Aug-87 Nov-87 -0.302 3 Jul-89
S&P 500 Aug-87 Nov-87 -0.311 3 Jul-89
Nasdaq Composite Aug-87 Dec-87 -0.299 4 Jun-89
1990s
Dow Jones Dec-99 Sep-02 -0.339 34 ?
S&P 500 Aug-00 Sep-02 -0.463 26 ?
Nasdaq Composite Mar-00 Oct-02 -0.741 32 ?
What causes stock market booms and abrupt reversals?
• Probabilities of long positive runs are small
• Probabilities of crashes are small
• Can fundamentals really shift that fast?
• Or is the crowd just mad?
Fundamentalists and Maniacs in the 1920s• Professor John B. Williams of Harvard (1938) wrote:
“Like a ghost in a haunted house, the notion of a soul possessing the market and sending it up or down with a shrewdness uncanny and superhuman, keeps ever reappearing….Let us define the investment value of a stock as the present worth of all the dividends to be paid upon it.”
John Maynard Keynes (1936) chose to differ:“A conventional valuation which is established as the outcome of the mass psychology of a large number of ignorant individuals is liable to change violently as the result of a sudden fluctuation of opinion which do not really make much difference to the prospective yield…..the market will be subject to waves of optimistic and pessimistic sentiment, which are unreasoning.”
Fundamentalists and Maniacs in the 1990s
• Robert Shiller (1991) observed:“I present here evidence that while some of the implications of the efficient markets hypothesis are substantiated by data, investor attitudes are of great importance in determining the course of prices of speculative assets. Prices change in substantial measure because the investing public en masse capriciously changes its mind.”
• John Cochrane (1991) expounded:“We can still argue over what name to attach to residual discount-rate movement. Is it variation in real investment opportunities not captured by current discount model? Or is it “fads?” I argue that residual discount-rate variation is small (in a precise sense), and tantalizingly suggestive of economic explanation. I argue that “fads are just a catchy name for the residual.”
Fundamentals•Fundamentals require that stock prices equal the present discounted value of expected future dividends. •The simplest approximation to this fundamentals based stock market valuation is the Gordon growth model. •Dividends (D) grow at a constant rate g and investors command a constant return of r, composed of a risk free rate and an equity premium. A constant fraction of earnings are paid out as dividends, so that D = (1-b)E•For the aggregate price level of the stock market, P, the model is:
(1) P = (1+g)D/(r-g)•The Gordon growth model may also be written as the price-dividend ratio or the dividend yield:
(2) P/D = (1+g)/(r-g) or (D/P)(1+g) = r-g
The Gordon model captures all explanations for asset price movements, including booms and busts:
(a) *Technological change increasing productivity and leading to higher dividend growth
(b) Changes in the payout rate
(c) Changes in the risk free rate
(d) *Changes in the equity premium
Frustrating!•Explaining actual stock price movements with fundamentals has proved frustratingly difficult. •If expectations are rational, stock prices should embody the realized dividends in the future appropriately discounted. •In a classic article, Shiller (1981) found that stock prices moved far more than was warranted by the movement in dividends, where the ex post rational price was equal to the discounted value of the future stream of realized dividends. •Even if there were deviations in was expected from what was realized, the fit should have been good over 1871-1979, yet the variation of prices exceeded the variation in fundamental prices violating any reasonable test.
Campbell and Shiller (1988)
Empirical Regularities•While it has proved difficult to explain the behavior of prices in terms of the movements of future dividends, a different literature has found empirical regularities
•Empirically the behavior of current prices is explained (R2=10%) in terms of past fundamentals.
•This predictability is surprising, given that prices should be forward looking, but it provides an instrument for analyzing stock prices.
Lamont (1998)
Empirical Regularities—Fama & French-Lamont--Campbell &Shiller etc.
Changes in the lagged dividend yield (D/P), the earnings yield (E/P), and payout ratio (D/E) have explanatory power for stock returns.
Higher D/P predict higher future returns because dividends measure the permanent component of stock prices.
Higher E/P or D/E forecasts lower returns because the level of earnings is a measure of current business conditions, and reflects the transitory component
The 1920s Real
Dividends Dividend Yield
Real Earnings
Earnings To Price
Payout Ratio
Real Price
1900-1909 7.6 4.6 12.7 7.6 60.4 173.1 1910-1919 8.0 5.9 13.5 10.1 62.4 149.6 1920-1924 5.3 6.3 7.6 8.8 81.9 83.5
1925 6.1 5.7 12.7 11.8 48.0 110.9 1926 7.1 5.5 12.8 9.8 55.6 128.1 1927 8.1 5.7 11.6 8.3 69.4 138.8 1928 9.0 4.8 14.6 7.9 61.6 183.7 1929 10.3 3.9 17.1 6.5 60.2 263.6 1930 11.2 4.5 11.1 4.5 101.0 230.2 1931 10.4 5.1 7.7 3.8 134.4 182.2 1932 7.0 6.0 5.8 4.9 122.0 105.2 1933 6.0 6.2 6.0 6.2 100.0 99.6
The 1990s Real
Dividends Dividend Yield
Real Earnings
Earnings To Price
Payout Ratio
Real Price
1970-1979 13.2 4.1 29.4 9.4 45.5 360.9 1980-1989 13.5 4.6 28.1 9.5 48.6 321.5 1990-1994 15.9 3.2 27.6 5.4 59.8 521.7
1995 16.2 3.0 39.9 7.3 40.6 561.2 1996 17.0 2.4 44.1 6.3 38.5 721.5 1997 17.4 2.0 44.6 5.2 39.0 873.1 1998 17.9 1.7 41.6 3.9 43.0 1080.8 1999 17.9 1.3 51.7 3.9 34.6 1378.0 2000 16.8 1.1 51.8 3.5 32.5 1531.2 2001 16.1 1.2 25.3 1.9 63.8 1378.1 2002 15.8 1.4 30.3 2.7 52.1 1167.3
Two Booms1920s v. 1990s
• Zero inflation, 4% unemployment, balanced budget
• Real Earnings jump• Real Dividends jump• Payout Ratio constant• Soaring Prices• Collapse in D/P and E/P• Empirical regularities: big
fall in D/P forecasts lower future returns, mitigated by falling E/P.
• Low inflation, low unemployment, balanced budget
• Real Earnings jump • Real Dividends don’t• Payout Ratio drops• Soaring Prices• Collapse in D/P and E/P• Empirical regularities: big
fall in D/P forecasts lower future returns, mitigated by falling E/P.
One Forecast
•Lamont (1998) forecasted the cumulative return of buying stocks on December 31, 1994---hold five years---December 31, 1999
•For sample (1947-1994), the mean excess return was 33%.
•VAR Forecasting out-of-sample forecast is 1% below total Treasury bills returns! (potential total forecast error of 21%)
•His conclusion: in the mid-1990s, stock prices were very high relative to any benchmark
Explanation 1: Technological Change or the “New Economy”
• In the 1920s and 1990s bull markets, technological innovations were viewed as improving the marginal product of capital, increasing earnings and hence dividend growth.
•A wave of innovations, sometimes characterized as a new general purpose technology was believed to have placed the economy on a higher growth path.
The New Economy of the 1920s• Irving Fisher: the stock market boom
justified by the rise in earnings, driven by the systematic application of science and invention in industry and the acceptance of the new industrial management methods.
• High Tech Industries: Automobiles, Radio, Aircraft, Movies, Electric Utilities, Finance
• But some executives (A.P. Giannini of Bank of Italy) feel prices are too high and say they will not pay higher dividends.
The New Economy of the 1990s
• General purpose technology of 1990s greater impact than in 1920s.
• “Moore’s Law” number of transistors per integrated circuit doubles every 18 months, helps drive price declines.
• Estimated average annual price declines for electricity and automobiles: 2%, but computer prices collapsed at a rate of 24%.
• Faster rate of change and price decline in the 1990s, promised higher levels of growth and consumption.
The New Economy of the 1990s• Gordon’s (2000) estimates of annual multi-factor
productivity growth:• 1870-1891: 0.39% • 1890-1913: 1.14%. • 1913-1928: 1.42% • 1928-1950: 1.90% Golden Age of GP Tech• 1950-1964: 1.47% • 1964-1972: 0.89%• 1972-1979: 0.16%• 1979-1988: 0.59%• 1988-1996: 0.79% • 1995-1999: 1.35% IT revolution of a 0.54 unsustainable cyclical
effect and 0.81% trend growth which he attributes wholly to the computer-IT sector.
1920s Fundamentals? Yes?• Sirkin(1975) applied Gordon model to Dow-Jones stocks
in the 1920s to see if P/E ratios could have been justified by a temporarily higher growth of earnings.
• Mean and median at the peak in 1929 were 24 and 20. In his best case, assuming r=9%, if the higher growth rate of 8.9% typical of 1925-1929 had been sustained for ten years, the price-earnings ratio of 20 would have been justified.
• Non-nested calibrations. No other variables. Sensitive to time choice. Growth rate for 1927-1929 justifies or over-justifies all P/E ratios.
• Results reflect econometric fact that earnings are highly variable, compared with the permanent component of dividends and represent the transitory component of stock prices—long-term projections are hazardous.
1920s Fundamentals? Yes?• Donaldson and Kamstra (1996) estimated a
non-linear ARMA-ARCH model for discounted dividend growth for the 1920s
• Out-of-sample forecasts produce a fundamental price series with a similar time pattern to the actual S&P index.
• Non-nested. No significant variation in the equity premium. Fundamentals peak follows the actual peak, suggesting that the fit may partly reflect the highly persistent behavior of dividends.
Donaldson & Kamstra (1996)
1990s Fundamentals?
• The differences in productivity growth in the late 1990s between IT industries and the rest of the economy look like a potential good explanation for the Nasdaq’s behavior
• But the boom outside of the new economy would seem surprising without a major increase in productivity growth.
1990s Fundamentals?
• Heaton and Lucas’(1999) applied a Gordon model to S&P500 data. They calculate the growth rates that would be needed to justify the peak P/D.
• For (1872-1998), average P/D was 28 and real g was 1.4%, implying an r of 5%.
• To match the 1998 high ratio of 48 with r of 5% and 7% would require g of 2.9% and 4.9%---huge historical leaps--a doubling of productivity growth
Explanation 1?For both the 1920s, the conclusion for the 1990s is fairly clear: expected dividend growth was not a major factor driving the boom. The surge in earnings was part of a robust business cycle but did not have a sufficient permanent component to raise stock prices
Explanation 3: Changes in the Discount Rate
• The return required or stock yield (r) has become the favored factor behind stock market booms.
• R = risk free rate and an equity premium. It is believed to be moved primarily by the latter, as the risk free rate is held to be relatively constant.
• In the Gordon model (constant g), the stock yield is
• (3) rt = E(Dt+1 /Pt ) + g
• The equity premium is then calculated as the difference between the stock yield and a measure of the risk free rate.
Figure 6Stock Yield and Equity Premium
1871-2003
-6
-4
-2
0
2
4
6
8
10
12
14
Pe
rce
nt
Equity Premium Stock Yield
The Equity Premium• For 1871-2003:
D/P = 4.5%
Real g = 1.7%
Real 10 year bond yield = 3.2%
Equity premium = 3.1%
Trends: 19th century 4%, surged during Great Depression, slow decline, very small now.
Booms: Below 2% 1929, near zero 1990s
Bullish on the Equity Premium• “A Paradigm Shift” Glassman and Hassett, The
Dow 36,000 (1999) diversified portfolio of stocks no more risky than U.S. government bonds
• “Stocks should be priced two to four times higher---today. But it is impossible to predict how long it will take for the market to recognize that Dow 36,000 is perfectly reasonable. It could take ten years or ten weeks. Our own guess is somewhere between three and five years, which means that returns will continue to average about 25% p.a.
• Rationale: premium can fall from 2.8 to 0.8%: real long term bond rate of 2% and g = 2.3% permits a D/P of 1.5% to fall to 0.5% with a tripling of P.
Why has the equity premium fallen?
• 1920s:• Lower transaction costs
—securities affiliates improve services, better ticker and wire services
• Increased participation—surge of new middle class investors
• Increased diversification by investors—investment trusts make it easier to diversify
• 1990s• Lower transaction costs
—mutual funds, internet trading, computerization
• Increased participation—many more households buy stocks
• Increased diversification—mutual funds, 401(k) make it easier to diversify
But..many are skeptical of explanations
• Households switch from individual stocks to mutual funds, but no huge shift in shares of wealth holding
• New participants do not own much of total stock• Diversification still low. Many hold
disproportionate share of own company stock• Lucas and Heaton (1999) OLG model: no effect
from participation increase, very modest from diversification
• Campbell and Vuolteenaho (2004) find evidence of mispricing—bondholders quickly adjust to inflation but not stockholders
Bottom Line:• Fundamentals don’t explain booms well, neither rapid
productivity growth nor a falling equity premium seems to offer a good explanation
• Flood and Hodrick (1990): any test for a bubble is troubled by the problem that the dynamics of asset prices with a bubble will not appear to different from the dynamics when there is an omitted factor. Studies which purport to find a bubble can be attacked for failing to find some missing fundamental, while results where the conclusion is that there is no bubble are highly sensitive to the choice of parameters (White, 2006)
De Long and Shleifer (JEH 1991)• To avoid the problem
of mis-identifying fundamentals, De Long and Shleifer (1991) examined the prices of closed-end mutual funds, where fundamental value of a fund is simply the current market value of the securities in the fund’s portfolio.
• Usually, there is a small discount for closed-end mutual funds.
Huge growth in these funds: “irrational exuberance”??
Median seasoned fund sold for a premium of 37 percent in the first quarter of 1929, rising to 47 percent in the 3rd then in the 4th.
Instead of buying a fund that was above its fundamentals’ price, investors could
simply have been purchased a portfolio of the underlying stocks or entrepreneurs could have created new funds with the same stocks. T
The only consistent explanation is that investors were excessively optimistic, suggesting the existence of a bubble.
Rappoport and White (1994)
• Evidence in the market for brokers’ loans that lenders were very skeptical of the height that the market had attained in late 1929.
• Extraordinary interest premia and margin demanded on these loans suggest that lenders felt they needed this protection against a potentially huge decline in the market.
Brokers’ Loans as Options—to repay
Extracted the volatility implied by the price of these loans as options, revealing the potential for a crash on
the order of 25 to 50 percent well in advance of
October 1929
Costs of a Bubble: Distorted Decisions
• Keynes (1936): “Speculators may do no harm on a steady stream of enterprise. But the position is serious when enterprise becomes the bubble on a whirlpool of speculation. When the capital development of a country becomes a by-product of the activities of a casino, the job is likely to be ill-done.”
• A bubble will (1) raise household wealth causing higher than optimal consumption, (2) produce overinvestment it will raise market value to book value in Tobin’s q, and (3) induce more firms to borrow because of higher value of collateral and firms switch to equity finance if there is a lower equity premium. Crash will create credit crunch.
Should the Fed intervene? What’s the optimal policy
• Powerful traditional lesson---drawn from experience of 1928-1933---is keep monetary policy focused on inflation and growth, not the stock market
• Update: In a small calibrated model, Bernanke and Gertler (2000) found that an inflation-targeting rule stabilizes inflation and output when asset prices are volatile, driven by a bubble or technology shock.
• No additional gain from responding directly to asset prices because a response to stock prices lowers the variability of the output gap, but increases the variability of inflation. Argue it is more difficult to identify the fundamental component of stock prices than it is the output gap.
Exception
• Inflation targeting is “flexible.” Intervention is appropriate for a central bank as lender of last resort to intervene temporarily in a payments crisis or financial intermediation crisis and then withdraw injected liquidity (as in 1929 and 1987).
Some argument for pre-emptive action
• Cecchetti, Genberg, Lipsky, and Wadhwani (2000) Fed should respond to forecasts of future inflation, the output gap, and asset prices.
• Argue that it is no more difficult to measure stock price misalignments than it is the output gap. Find warranted premium in 2000 was 4.3%.
• Their model suggests that in 1997, the Federal Funds rate should have a Fed funds rate of 10% not 5.5% to keep inflation less than 3%, a very small output gap and a risk premium of just under 3%.
A Closer Look:The Role of the Fed
• How did policy makers in the twenties and nineties confront the booming markets?
• Did their policies hinder or aggravate the booms?
• The general economic conditions and the behavior of the Federal Reserve in the 1920s and 1990s are similar and offer instructive comparisons.
• Three potential lessons
Backdrop to the Booms• 1920s: long post-World
War I economic boom, preceded by high inflation, hard recession and many bank failures.
• 1922-1929, GNP grew at 4.7%, unemployment averaged 3.7%. (2 brief recessions), and no trend inflation
• Fed accommodated seasonal demands for credit and attempted to smooth economic fluctuations.
• 1990s: inflation of the 1970s, bank failures of 1980s, a sharp recession in 1990-1991.
• 1991-2001 longest expansion GNP grew at 3.3% and unemployment averaged 5.5%,with inflation averaging 2.5%.
• Fed focuses on interest rate targeting to smooth economic fluctuations.
Fed blamed for fueling the boom
• Expansionary monetary policy begun in the spring of 1927 to ease pressure on the British balance of payments following secret central bank meeting.
• Critics assert policy too easy, and allows boom to ignite.
• Fed tightens policy in 1928 and there is little increase in total money or credit for 1928-1929.
• Following the Asian crisis, the Fed eases credit and cuts the Fed funds rate to reassure market after September 1998 collapse of Long-Term Credit Management (LTCM).
• Critics asset policy too loose, allowing boom in the stock market to take off in its final phase.
• They argue that it was too late when the Fed finally began to raise interest rates in June 1999.
One key difference: “real bills doctrine”
• In the 1920s, most of Fed believed in real bills doctrine and saw the boom in the market as diverting credit from productive to speculative uses. Believed that if banks would restrict lending to “real bills” it would stabilize the economy.
• Hence, the Board was obsessed with the stock market and the credit used to fund holding stocks (brokers’ loans.)
Jawboning, and then…..• In February 1929, Board
chairman Young spoke out against excessive speculation. “Direct pressure” on member banks to limit "speculative loans.“ NY Fed opposes and wants to raise rates
• Continued boom, credit to market from other sources. No further response from stalemated Fed.
• August 1929 when the discount rate was raise from 5 to 6%, just as the economy reaches its cyclical peak.
• Crash October 2000
• In 1996, Alan Greenspan warns that the stock market was possessed by “irrational exuberance.”
• Continued boom, no response, appears Fed believes in New Economy
• Fed begins to increase the Fed funds rate in June 1999.
• Market begins collapse (Peaks: Nasdaq 3/2000, S&P500 8/2000
Lesson 1? Mistaken Monetary Ease
• Complaint: Loose Fed policy in 1927 fed boom.
• But M1 growth modest at 1% and cycle peak in October 1926. Indicators and simple Taylor rule suggest easier policy.
• No sign of boom in market, June 1927 level with June 1925. Equity premium at 4.7% its historic level
• No signal for intervention from Bordo & Jeanne rule of 3 years rapid growth or Cecchetti’s deviation from equity premium.
• Complaint: Loose Fed policy in 1998 after LTCM—some question whether a lender of last resort operation was needed.
• August-October 1998 market flat or in retreat. Equity premium low but little different than in previous years.
• Hard to identify signal for intervention
Lesson 2? Should the Fed also focus on the stock market?
• 1928-1932 Fed obsessed with the stock market
• Even in 1930, market still much higher than 1927 and equity premium still low.
• Continued worry over stock market leads to excessively tight policy after stock market crash when still worried about speculative excess.
• 1998-2003 Consensus (exception Cecchetti) that Fed has not focused on the stock market, but on inflation and output gap
• Yet, worry that low rates keep market high
Lesson 3 Intervention in a Crash?
• NY Fed injects liquidity into market in response to crash of October 1929
• Stressed brokers and customers need credit as margin calls are made.
• Danger of collapse of securities firms and clearing and settlements system.
• Interest rate spreads widen then close as crisis abates. Board criticizes NY Fed, credit tightened even though country sliding into depression.
• Fed injects liquidity into market in response to crash of October 1987.
• Stressed brokers and customers need credit as margin calls are made and specialists and traders find it difficult to obtain credit.
• Danger of collapse of securities firms and clearing and settlements system.
• Interest rate spreads widen then close as crisis abates. and Fed withdraws liquidity
Conclusion• Measures of stock market boom
fundamental and bubble components are fragile at best.
• Although there may be a bubble in the market, central banks should not intervene but focus on inflation and growth
• Central banks proper role is as a lender of last resort if a stock market crash threatens the payment system or intermediation.