Bubble Investing: Learning From History
William N. Goetzmann
Yale School of Management – International Center for Finance; National Bureau of Economic Research (NBER)
January 11, 2016
History is important to the study of financial bubbles precisely because they are extremely rare events, but history can be misleading. The rarity of bubbles in the historical record makes the sample size for inference small. Restricting attention to crashes that followed a large increase in market level makes negative historical outcomes salient. In this paper I examine the frequency of large, sudden increases in market value in a broad panel data of world equity markets extending from the beginning of the 20th century. I find the probability of a crash conditional on a boom is only slightly higher than the unconditional probability. The chances that a market gave back it gains following a doubling in value are about 10%. In simple terms, bubbles are booms that went bad. Not all booms are bad.
Bubble Investing: Learning From History – Introduction
The broad awareness of financial history seems to correlate to extreme market events. For1 example, the closest comparison to the Dot-Com bubble of the 1990’s was the run-up in U.S. stock prices in the 1920’s. During the 2008 financial crisis, the financial press frequently referenced past bubbles– periods of market euphoria followed by sharp price declines. In this paper I argue that using past crashes in this way is misleading to both investors and policy-makers. Particularly during periods of market booms, focusing attention on a few salient crashes in financial history ignores the base rate for bubbles. In simple terms, bubbles are booms that went bad but not all booms are bad.
To illustrate this last point, I present empirical evidence drawn from more than a century of global stock market data. I define a bubble as a large price decline after a large price increase or, a crash after a boom. I find that the frequency of bubbles is quite small. The unconditional frequency of bubbles in the data is 0.3% to 1.4% depending on the definition of a bubble. Not only are bubbles rare but conditional upon a market boom (i.e. increasing by 100% in a one to three year period). Crashes that gave back prior gains happened only 10% of the time. Market prices were more likely to double again following a 100% price boom.
Prior to the empirical analysis, I present evidence about bubbles (as well as the lack of them) in very early equity investments. Thus, the next section discusses some of the early bubbles in financial history. Section III describes the databases used in the study and the empirical analysis. Section IV discusses the implications of the results for investors and regulators.
Data on Markets and Bubbles
The first bubbles precede the development of organized stock exchanges. Stuart Jenks reports evidence of a bubble in speculative German mining shares, kuxe, at the end of the 15th century.1 Fractional equity interest in individual silver mines in the Hartz mountain district were evidently freely traded, purchased on credit and occasionally had option-like features.
Transactions were settled at financial fairs during which share prices could fluctuate dramatically. These were famously condemned by Martin Luther in 1554: “Ich will kein kuks haben! Es ist spiegelt, und es will nicht wudeln [gedeihen] dasselbige gelt.” “I will have nothing to do with kuxen. They are play money and will not generate hard cash.”
In 1502, on the eve of sailing on his final voyage, Christopher Columbus expressed a desire that his son use his inheritance to purchase shares in the Casa di San Giorgio in Genoa which he observed would generate “6 % interest and constitute a very safe investment.”2 The firm was a financial institution that owned and managed government contracts and ultimately became a bank. Its board regularly declared dividends and these, as well as the shares themselves, were actively traded.
Shares in Genoa’s Casa di San Giorgio fluctuated considerably in the 16th century. Figure 1 shows an index of prices and yields. The dramatic doubling of prices in 1602 looks like a bubble to the modern eye because yields declined from 3% to 1 ½%. This bubble sustained itself for a long time however, prices did not drop back to their former level until 1683. Likewise, a peak in 1622 looks ex post like a bubble, although the fortunes of Genoa as a financial power in the early 17th century also fluctuated considerably. The variation on both occasions might be due to rational speculation on events of the time. Nevertheless, they appear to fit a price-based definition of a bubble. This bubble pattern is not ubiquitous in the early history of equity shares. In Le Bris et al. (2014) we found no evidence of a bubble in the trading history of an even older corporation. Stock prices for the Bazacle milling company of Toulouse, over an extended period from the 1530’s to 1946, moved fairly closely with dividends.
The first discussions in England of a stock market bubble centered on the speculation in shares for start-up companies during the 1690’s. McCleod (1986) argued that intellectual property rights were more likely the excuse for stock market speculation rather than the basis for real valuation in this first English market bubble.
The first great stock market bubble began in France, with the creation of the Mississippi Company by John Law which was an ingenious financial innovation that merged a bank empowered to issue currency with companies chartered for overseas trade– hence the name Mississippi Company. The price of shares grew by more than 10 times during 1719 and 1720.
The Mississippi Bubble burst in the spring of 1720 when shares were made exchangeable with paper currency at a fixed rate, resulting in a massive government commitment to propping up share prices by printing money.4 The Mississippi Bubble was followed shortly by the South Sea Bubble in London and a smaller but significant bubble for shares in the Netherlands. The British and Dutch bubbles subsequently burst in late 1720, and by the end of the year, the boom in stock market speculation was effectively over.
My co-authors and I have worked to understand the basis for this remarkable sequence of international stock bubbles from 1719-1720.5 We found empirical and archival evidence that regulatory enforcement following the Bubble Act in London triggered a crash in the prices of insurance company stocks and that this ultimately spread to the large trading companies and banks in the UK and then overseas to the Dutch West Indies Company and a number of recently launched companies in the Netherlands.
Figure 2 illustrates the parallel growth in share prices for selected companies in London and Amsterdam in this period. The three London companies are Royal Exchange Assurance, London Assurance and the South Sea Company. The two Dutch companies are the Dutch West Indies Company and Stad Rotterdam– an insurance company whose successor firm still exists today. The figure shows the scale of the London and Amsterdam bubbles. The South Sea Company rose by a factor of 7.5 over the year leading to the eponymous “South Sea Bubble.” The two marine insurance companies grew much more by multiples of more than 10 and 13. Only the Dutch West Indies Company grew at a comparable scale in Amsterdam by a factor of 7. Stad Rotterdam did not quite double before declining in price. The graph also shows how inter-connected the Dutch and British bubbles were. Although they rose at different times in the year 1720, the crash in the prices of the London insurance firms and the Dutch West Indies Company occurred at about the same time (a few days lag is consistent with travel times between the two financial centers).
See full PDF below.