Blackwell Publishing Ltd.

Oxford, UK and Malden, USAEHRThe Economic History Review0013-0117Economic History Society 20052005LVIII2233271ArticlesIRRATIONAL BEHAVIOUR DURING THE SOUTH
SEA BUBBLERICHARD S. DALE, JOHNNIE E. V. JOHNSON, AND LEILEI TANG

Economic History Review, LVIII, 2 (2005), pp. 233–271

Financial markets can go mad:

evidence of irrational behaviour
during the South Sea Bubble1
By RICHARD S. DALE, JOHNNIE E. V. JOHNSON, and
LEILEI TANG

T he volatility of world stock markets in recent years has been the subject
of considerable debate. At one extreme there are those who argue that
these fluctuations have resulted from shifts in fundamental values, others
focus on the actions of rational investors confronted with new, fast-changing
information and, at the other extreme, there are those who blame investor
irrationality. Thaler suggests that an important conclusion to emerge from
the behavioural finance literature is that the resolution of these debates is
more likely ‘if we learn more about the behavior of people who operate in
these markets’.2 To this end, this paper explores whether investors can be
prone to bouts of irrational speculative mania.
Investor behaviour is examined during a period when market prices
systematically deviate from fundamental values; that is, during a bubble.
Specifically, the paper focuses on one of the most famous bubbles, that
associated with the rise and fall of the South Sea Company during 1720.
The causes of this bubble have attracted considerable academic debate,
since it formed an important part of the first major speculative boom and
bust in European stock markets, and its aftermath has had a considerable
impact on financial markets. Much of the earlier literature characterizes this
bubble as an episode of irrational speculative activity, but more recent
explorations have sought to explain it within a rational expectations frame-
work. In this article new evidence is presented, which calls into question
these latter contributions and refocuses attention on the central conclusion
to emerge from Adam Anderson’s original history of the Bubble; namely
that it represented an ‘unaccountable frenzy’, which should serve ‘as a
warning to after ages’.3 These findings underline the important role that
irrationality may play in stock markets and suggest that today’s policymakers
may need to counteract the potentially damaging effects of irrational pricing
of financial assets.

1
We are grateful to the anonymous referees for their valuable comments on an earlier draft of this
paper.
2
Thaler, Behavioral finance, p. xxi.
3
Anderson, Origin of commerce, p. 123.
© Economic History Society 2005. Published by Blackwell Publishing, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street,
Malden, MA 02148, USA.
 14680289, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1468-0289.2005.00304.x by Swinburne University Of, Wiley Online Library on [18/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
234 RICHARD S. DALE, JOHNNIE E. V. JOHNSON, AND LEILEI TANG

The article is arranged as follows. Section I outlines the events surround-

ing the rise and fall of the South Sea Company in 1720. Section II briefly
reviews the literature on the origins of bubbles in general, and section III
provides a summary of the debate concerning the causes of the South Sea
Bubble. Section IV provides details of the data, and the methodology
employed in this study is discussed in section V. Results are presented
in section VI and discussed in section VII. Some conclusions follow in
section VIII.

I
During 1719, John Law’s financial experiment to convert France’s debt into
stock in the Compagnie d’Occident (the ‘Mississippi company’) proved a
great success, and extraordinary profits were being made by Mississippi
investors.4 The government in England was already exploring means of
managing its outstanding debt, and events in France prompted it to explore
a scheme similar to Law’s Mississippi operation.5 Following a bidding
process between the Bank of England and the South Sea Company, the
latter ‘won’ the right to buy in all outstanding government long annuities,
short annuities, and redeemable debts (~£31.5 m) in exchange for its own
stock. The government agreed to credit the Company with an increase of
£31.5 m in its nominal share capital if all subscribable debts were
exchanged, and to pay interest on the debt partly at 4 per cent and partly
at 5 per cent until 1727, when a single rate of 4 per cent would be paid. In
exchange for these privileges, the South Sea company agreed to pay the
government a maximum sum of £7.6 m, assuming full conversion of the
government debt.
The South Sea Company refused to set in advance the nominal amount
of stock exchangeable for debt, and any rise in the share price above the
nominal value (£100) would therefore reduce the amount of stock needed
to convert the outstanding government debt. For example, if the share price
were £300, the government debt of £31.5 m could be bought in by issuing
only £10.5 m (nominal) of stock. The remaining £21 m (nominal) of issu-
able stock could therefore provide a potential cash flow of £63 m at the
prevailing market price (£300), some of which could be used to meet the
potential £7.6 m payable to the government.6 This, together with the desire
of South Sea projectors to make a capital gain on their own share holdings,
gave the Company an incentive to secure an appreciation in the share price.
The largely redundant commercial interests of the South Sea Company
offered no outlet for any additional capital raised. However, an appreciation
in the share price did provide the Company with the means of raising
dividends—albeit, unsustainably—and consequently stimulating even fur-

4
Murphy, John Law economic theorist, pp. 156–65.
5
Dickson, Financial revolution in England, pp. 90–121.
6
Scott, Constitution and finance, p. 307.
© Economic History Society 2005
 14680289, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1468-0289.2005.00304.x by Swinburne University Of, Wiley Online Library on [18/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
IRRATIONAL BEHAVIOUR DURING THE SOUTH SEA BUBBLE 235
Table 1. Schedule of payments for South Sea money subscriptions (£)
First Subscription Second Subscription Third Subscription Fourth Subscription
(£s) (£s) (£s) (£s)

14 April 60 (29 April) 40 16 June 100 24 Aug. 200

1720 1720 1720 1720
14 June 30 14 Sept. 40 2 Jan. 100 24 Feb. 200
(24 Aug.) 1721 1721
14 Aug. 30 14 Jan. 1721 40 2 July 100 24 Aug. 200
(24 Dec.)
14 Oct. 30 14 May 40 2 Jan. 100 24 Feb. 200
(24 April) 1722 1722
14 Dec. 30 14 Sept. 40 2 July 100 24 Aug. 200
(24 Aug.)
14 Feb. 1721 30 14 Dec. 40 2 Jan. 100
(24 Dec.) 1723
14 April 30 14 March 1722 40 2 July 100
(24 April)
14 June 30 14 June 40 2 Jan. 100
(24 Aug.) 1724
14 Aug. 30 14 Sept. 40 2 July 100
(24 Dec.)
14 Dec. 40 2 Jan. 100
(24 April) 1723 1725
Total 300 Total 400 Total 1,000 Total 1,000

Notes: The sources of data and the method of compilation are discussed in appendix I.

ther rises in the share price. In addition, it emerged later that in their bid
to secure the scheme’s passage through Parliament, the Company had
effectively given ‘call options’ to key government ministers, to 40–50 mem-
bers of Parliament, and to those closely connected with the Crown, thereby
ensuring that this powerful group also had a strong desire to see the share
price rise.
In their implementation of the debt conversion scheme, the Company
engaged in what Hutcheson, a sceptical Member of Parliament at the time,
described as ‘artful management of the spirit of gaming’. 7 A number of
devices were employed to issue more stock, while maximizing the South
Sea share price. First, four successive share issues were launched on gener-
ous subscription terms involving small down payments and extended calls
(see table 1 for details of the four subscription issues). Second, the partly
paid (and therefore highly leveraged) scrip was designed to be popular with
speculators, particularly as it could be transferred using a simple legal
assignment. Third, to increase investor liquidity and, consequently, to
encourage further demand for its stock, the South Sea Company encour-
aged investors to borrow from the Company against the security of its shares
or subscription receipts. A total of more than £11 m was eventually lent in
this way. Fourth, the South Sea Company supported its share price by

7
Hutcheson, Collection of calculations, p. 64.
© Economic History Society 2005
 14680289, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1468-0289.2005.00304.x by Swinburne University Of, Wiley Online Library on [18/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
236 RICHARD S. DALE, JOHNNIE E. V. JOHNSON, AND LEILEI TANG

buying up its own stock. Fifth, the Company raised investors’ expectations
by carefully orchestrated dividend announcements, it being announced, for
example, on the day before the first money subscription that the midsum-
mer dividend would be increased to 10 per cent (from 3 per cent). Lastly,
the Company delayed the issue of shares to those who had converted their
annuities, and failed to issue receipts for the third and fourth subscriptions.
The intention here was to make it more difficult for converting annuitants
and subscribers to sell their shares and partly-paid scrip, thereby giving
further support to the share price. However, this latter ploy was, at best,
only partially successful, since investors could and did trade in subscription
receipts before they were delivered out by the company, and there is con-
siderable evidence that such trading was widespread. For example, it was
reported that one-sixth of all subscription three receipts were resold within
days of the issue, prior to receipts being delivered by the company. 8
Against this background, South Sea shares, which had already advanced
from £130 in February 1720 to over £300 in early April, rose further to
£400 (20 May), then to £500 (28 May) before touching £600 on 31 May.
This vertical ascent continued into the following month, the share price
breaching £700 on 1 June and £800 on 4 June. On 23 June the Company
closed its books for two months in order to process the midsummer divi-
dend, so that quoted prices during this period are in fact forward prices ‘for
the opening of the books’. The highest (forward) price was £1,050, recorded
on 25 June, but when the books were reopened the spot price had fallen to
£820 (cum dividend). Thereafter South Sea stock weakened dramatically—
to £520 in mid-September, £290 at the beginning of October and a low of
£170 on 14 October 1720. The bubble had burst.

II
An important distinction is made in the literature between bubbles that
result from rational as opposed to irrational behaviour.9 It has been sug-
gested that, broadly, the former arise from three causes, self-fulfilling
expectations (‘rational bubbles’), mispricing of fundamentals (‘intrinsic
bubbles’), and the endowment of irrelevant exogenous variables with asset
pricing value (‘extrinsic bubbles’). The South Sea Bubble has been attrib-
uted to each of these causes.
Rational bubbles occur when asset prices continue to rise because inves-
tors believe that they will be able to sell the overvalued asset at a higher
price in the future.10 Investors are aware that a point will be reached when
the bubble will burst and, consequently, they require compensation for the
risk in the form of higher returns. As the bubble grows, the probability of
a price collapse increases and investors require ever increasing compensa-

8
Dickson, Financial revolution in England, p. 129.
9
Katsaris, Collapsing speculative bubbles, pp. 9–10.
10
Flood and Hodrick, ‘Speculative bubbles’, p. 86.
© Economic History Society 2005
 14680289, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1468-0289.2005.00304.x by Swinburne University Of, Wiley Online Library on [18/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
IRRATIONAL BEHAVIOUR DURING THE SOUTH SEA BUBBLE 237
tion (returns). This behaviour is associated with spiralling prices, which
result, eventually, in the bubble bursting.
Intrinsic rational bubbles occur when investors systematically and persis-
tently misvalue fundamentals.11 This might arise, for example, during a
period of rapid innovation, when even rational investors find it difficult to
determine an asset’s fundamental value. Zeira demonstrates that intrinsic
bubbles can result in prices rising significantly for long periods, and then
crashing as a result of informational dynamics.12 Froot and Obstfeld suggest
that empirical evidence for intrinsic bubbles is based on the persistent
overreaction that occurs following news of dividends.13
It is argued that extrinsic rational bubbles, or ‘sunspots’, occur when
otherwise rational investors, face uncertainty (cf. risk) concerning their
environment.14 In the face of this uncertainty they erroneously endow asset-
pricing value to exogenously determined factors which have no impact on
fundamental values.15 When these beliefs are widely shared it may result in
asset prices deviating from fundamental values. Support for the existence
of extrinsic bubbles comes, for example, from adjustments of asset prices
to forecasts from investors who do not hold superior information.
In contrast to the above explanations for bubbles developed within a
rational expectations framework, some authors have argued that these
bubbles may result from investors being driven by irrationally optimistic
expectations, fashion, or fads.16 Under such conditions investors may
develop totally unrealistic expectations about a company’s future profitabil-
ity; as a result the relationship between fundamental values and prices
breaks down.17 These bubbles have been associated with herd behaviour,18
and it has been suggested that investors may be driven by psychological
factors unconnected to fundamental values.19 For example, when investors
are uncertain about the quality of information they hold, they may revert
to a simple heuristic of following market trends.20 If investors use the market
in this manner to improve their information set, it can result in a form of
herd behaviour or mimetic contagion21 that displays the typical features of
a bubble.
The distinction between bubbles developed from rational or irrational
behaviour may not be as clear-cut as the above analysis suggests. For
example, investors may mistakenly be convinced that market valuations are

11
Froot and Obstfeld, ‘Intrinsic bubbles’, p. 1190.
12
Zeira, ‘Information overshooting’, pp. 239–50, 253.
13
Froot and Obstfeld, ‘Intrinsic bubbles’, p. 1208.
14
Azariades, ‘Self-fulfilling prophecies’, pp. 380–1.
15
Ibid., p. 395.
16
Shiller, Irrational exuberance, pp. 17–168.
17
Ofek and Richardson, ‘DotCom mania’, p. 1134; Perkins and Perkins, The Internet bubble,
pp. 197–201.
18
Devenow and Welch, ‘Rational herding’, p. 605.
19
Shiller, Irrational exuberance, pp. 135–67.
20
Avery and Zemsky, ‘Herd behaviour in financial markets’, p. 740.
21
Lux, ‘Herd behaviour’, pp. 893–4.
© Economic History Society 2005
 14680289, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1468-0289.2005.00304.x by Swinburne University Of, Wiley Online Library on [18/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
238 RICHARD S. DALE, JOHNNIE E. V. JOHNSON, AND LEILEI TANG

realistic and as a result become participants in an ‘irrational’ bubble; but

the degree of misjudgement that should be classed as irrational is a matter
of subjective judgement. Similarly, those who fail to sell assets before the
crash of a rational bubble may have misjudged the crash probability. Once
again, the degree of misjudgement permitted for an investor to be termed
rational is debatable.

III
Those who seek to explain the South Sea Bubble in terms of irrational
behaviour rely largely on contemporary anecdotes and quotations. For
example, Carswell quotes John Martin, a banker who subscribed to £500
worth of South Sea stock in June 1720 with the comment, ‘when the rest
of the world are mad, we must imitate them in some measure’. 22 Carswell
suggests that it is difficult to explain within a rational expectations frame-
work how investors could have ‘believed their [the South Sea Company’s]
airy structures would not, in the absence of real assets or even prospects,
collapse in ruins’.23 Kindleberger uses the South Sea Bubble as a classic
example of speculative mania, where speculation spread to members of the
public who had little knowledge of the market.24 Chancellor identifies three
factors which he believes point to irrationality associated with those who
purchased shares towards the peak of the bubble; first, that there was
considerable public information indicating that the shares were overvalued,
second, the risk/reward ratio for these late entrants was very poor, and lastly,
the price volatility during 1720 could not be explained by changes in the
fundamentals of the South Sea Company.25 Dickson suggests that in the
early eighteenth century even the judgement of experienced businessmen
was affected by factors in the social and economic environment, which ‘bred
an appetite for gain’ that ‘could and did become uncontrolled’. 26 In sum-
mary, the above accounts point to the South Sea Bubble arising from
irrational mania.
Recent, more quantitative contributions to the debate have attempted to
explain the bubble within a rational expectations framework. Neal, for
example, presents evidence to suggest that the rise in the South Sea share
price from early February to mid-May was caused by investors re-evaluating
the company’s fundamentals in the light of the unprecedented financial
innovation introduced to convert government debt into stock. 27 However,
in retrospect, there appears to have been some mispricing of the fundamen-
tals at this time and, consequently, this phase displays the features of an

22
Carswell, The South Sea Bubble, p. 163.
23
Ibid., p. 241.
24
Kindleberger, Manias, panics, and crashes, pp. 111–12
25
Chancellor, A history of financial speculation, p. 94.
26
Dickson, Financial revolution in England, p.156.
27
Neal, Rise of financial capitalism, pp. 111–12.
© Economic History Society 2005
 14680289, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1468-0289.2005.00304.x by Swinburne University Of, Wiley Online Library on [18/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
IRRATIONAL BEHAVIOUR DURING THE SOUTH SEA BUBBLE 239
intrinsic bubble. He suggests that the price rises between mid-May and late
June resulted from shrewd foreign investors realizing that the South Sea
directors were engaged in various forms of price rigging. They therefore
believed that they would be able to sell these overvalued shares at even
higher prices in the future, and a rational bubble was created. Neal argues
that a technical convergence of the forward and spot prices occurred from
late June to late August, when the transfer books were closed. 28 Lastly, he
suggests that the subsequent collapse was initially caused by a credit squeeze
that led to an unwinding of speculative positions, and later by the uncer-
tainty surrounding the proposed reorganization schemes for the Company.
Overall, Neal’s view is that the South Sea Bubble can be explained in terms
of rational behaviour in that it ‘appears to be a tale less about the perpetual
folly of mankind and more about the continual difficulties of the adjust-
ments of financial markets to an array of innovations’.29 Carlos, Moyen, and
Hill explore whether movements in the share price of a trading organization,
the Royal African Company, could be attributed solely to responses to
changes in underlying fundamentals during the South Sea Bubble. 30 The
evidence they present suggests that ‘fundamentals cannot fully explain the
market price for the first three quarters of 1720’.31 However, they conclude
that on the whole their results are consistent with investor rationality during
this period.
Garber estimates that the market capitalization for the South Sea Com-
pany at the end of August 1720 was £164 m; that is around five times the
tangible net assets.32 He argues that speculators, ‘working on the basis of
the best economic analysis available’, believed that the fund of credit which
the South Sea Company had accumulated justified the market valuation,
since it could be used for commercial expansion.33 However, it is clear that
the Company engaged in no significant trade other than holding govern-
ment debt, and Garber himself points out that ‘anyone projecting commer-
cial returns high enough to justify the higher prices of South Sea shares was
probably too optimistic’.34 Consequently, if, as Garber argues, rises in the
South Sea share price can be attributed to rational behaviour, then these
can, at best, be attributed to a mispricing of fundamentals (an intrinsic
bubble) or to a view amongst investors that some outside factors might
improve the Company’s prospects (an extrinsic rational bubble). This view
to some extent mirrors Scott’s more qualitative assessment of the history of
the South Sea Company up to the end of May.35 He argues that the share
price was not excessive during this period because investors were not aware

28
Ibid., p. 111.
29
Ibid., p. 90.
30
Carlos, Moyen, and Hill, ‘Royal African Company share prices’.
31
Ibid., p. 80.
32
Garber, ‘Famous first bubbles’, p. 52.
33
Ibid., p. 52.
34
Ibid., p. 52.
35
Scott, Constitution and finance, pp. 308–20.
© Economic History Society 2005
 14680289, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1468-0289.2005.00304.x by Swinburne University Of, Wiley Online Library on [18/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
240 RICHARD S. DALE, JOHNNIE E. V. JOHNSON, AND LEILEI TANG

of the price rigging and other malpractices of the South Sea directors. The
60 per cent excess in share price over tangible assets (goodwill) at the end
of May could, in his view, be accounted for by overoptimism rather than
irrationality (an intrinsic bubble).36 However, his explanation of the rise in
prices from June onwards includes references to ‘a spirit of wild speculation’
and market manipulation37 that suggest either a rational bubble, an extrinsic
bubble (based around a false view that market manipulation might succeed
in producing ever higher prices) or even irrational behaviour.
The extent to which the South Sea Bubble can be explained as a rational
bubble is debatable, for two reasons associated with bubble theory. First,
South Sea trade was effectively dead, and the bulk of the Company’s assets
consisted of loans to the government earning a known rate of interest. It
was therefore essentially an annuity holding company or ‘bond’, subject to
a call option after seven years—when the government was entitled to begin
to repay its debt at face value—that is, an asset with known cash flows and
terminal values. Bubble theory suggests that such assets cannot be the subject
of rational bubbles since, when the ‘bond’ is repaid, investors face a certain
capital loss.38 Second, the supply of South Sea stock increased significantly
with each successive money subscription, and as the price increased, a larger
increase in supply became permissible. The connection between the con-
version price for government debt and the Company’s issuable capital was
well publicized, and it is therefore likely that investors understood that a
greater volume of shares would come onto the market the higher the share
price rose. Bubble theory suggests that in such circumstances a rational
bubble is not possible, since the market will be unable to absorb an ever-
increasing supply of stock at a price higher than its fundamental value. 39
As is evident from this brief overview of the literature, there is a great
deal of debate concerning the degree of investor rationality during the rise
and fall of the share price of the South Sea Company during 1720. On the
whole, those authors who employ quantitative analysis favour a rational
explanation, and those who employ qualitative analysis conclude that irra-
tionality prevailed. The aim here is to employ new data to explore the
behaviour of investors in South Sea stock during 1720, and hence to shed
light on the origins of the South Sea Bubble.

IV
The data employed in this study relate to a key source that to date has not
been employed in interpretations of the South Sea Bubble. This involves
the price performance of South Sea shares in relation to subscription
receipts, and the price performance of the four separate subscriptions in

36
Ibid., p. 313.
37
Ibid., p. 319.
38
Katsaris, Collapsing speculative bubbles, p. 17.
39
Tirole, ‘Asset bubbles’, p. 1521.
© Economic History Society 2005
 14680289, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1468-0289.2005.00304.x by Swinburne University Of, Wiley Online Library on [18/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
IRRATIONAL BEHAVIOUR DURING THE SOUTH SEA BUBBLE 241
relation to each other at each point in time throughout the Bubble period.
The analysis is confined to the four ‘money’ subscriptions, which should be
distinguished from the two conversion offers that allowed holders of gov-
ernment annuities to convert the debt into South Sea stock. The successive
conversion offers changed the terms of the exchange of government annu-
ities into stock, and it would be difficult to compare the prices of the
conversion offers. However, the four ‘money’ subscriptions do not suffer
from this disadvantage, since each subscription simply provided the holder
with a means of acquiring stock through a system of staged payments—
which varied for each subscription. Having appropriately discounted the
staged payments of the four subscriptions, the stock and the ‘money’ sub-
scriptions are converted into directly substitutable instruments, in that they
represent equivalent claims on assets and dividends. Consequently, the
adjusted subscription prices can be compared directly with each other, and
with the stock price.
The schedule of payments for each of the successive money subscriptions
is provided in table 1. Receipts for these subscriptions could be traded
through a simple legal assignment, as was made clear in the subscription
preambles.40 The price data employed for South Sea stock and subscriptions
is from Freke’s Prices of Stocks etc, which was published twice per week, each
issue covering three days’ trading (Wednesday to Friday, and Saturday to
Tuesday), there being no trading on Sundays. John Freke published his
price list from 1714–22. He collected prices in the morning and until
3.00 pm each afternoon, sometimes recording as many as three daily prices,
for both stock and subscription receipts. Where multiple daily prices are
quoted, a simple average is used to provide a single price for the day, since
no trading volume is available to determine a weighted average.
An alternative data source is John Castaing’s The Course of the Exchange,
which began publication in 1697 and eventually became the official stock-
exchange price list. However, Freke’s price list is employed because it is the
most comprehensive, and as such is uniquely placed to provide the data
necessary for a detailed comparison of stock and subscription prices during
the South Sea Bubble. For example, subscription prices are quoted by Freke
for periods not covered by Castaing and, crucially, for the third and fourth
subscriptions Freke, but not Castaing, makes a distinction between ‘for
money’ and forward subscription prices. This allows more direct compari-
son of stock and subscription prices during the period when these two forms
of secondary market for receipts were active, even though the receipts had
not been delivered out by the Company.
Scott, in his seminal work on the South Sea Bubble, chose to use Freke’s
stock prices,41 but both Dickson42 and Neal43 use Castaing’s prices, and

40
House of Lord’s Record Office Parchment Collection, Box 157, 1720.
41
Scott, Constitution and finance, pp. 392, 422.
42
Dickson, Financial revolution in England, p. 139.
43
Neal, Rise of financial capitalism, pp. 183–4.
© Economic History Society 2005
 14680289, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1468-0289.2005.00304.x by Swinburne University Of, Wiley Online Library on [18/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
242 RICHARD S. DALE, JOHNNIE E. V. JOHNSON, AND LEILEI TANG

Neal suggests that Castaing is the more reliable source.44 The grounds for
Neal’s allegation are not made explicit, but it may be because Freke was
known as a supporter of the South Sea Company and its scheme. However,
this fact alone is not sufficient to discount Freke’s prices, since it is not
obvious how a systematic distortion of prices could have helped the South
Sea cause. Equally, as a commercial provider of data, Freke would be well
aware that any perception that he was distorting the data would have
destroyed his business, particularly as he was in close competition with
Castaing. In addition, Freke’s publication was reportedly aimed at a
London readership, and these subscribers would have ready access to
Exchange Alley, where trading took place; differences between Freke’s
reported prices and actual prices would therefore have quickly come to light.
Castaing’s sheet, on the other hand, may have been intended more for non-
London residents, who would have less ready access to the market for
subscription receipts.45 Parsons uses Freke’s prices in his study of the
behaviour of London stock prices.46 He compares Freke’s prices with a
source taken mainly from the Daily Courant and finds ‘reasonable similarity
between the two price series’.47
The most complete series of Freke’s prices is housed in the British Library
(only issue no. 96, covering data for 21–23 May 1720, is missing) and, for
the reasons discussed above, this is used to provide a continuous series of
South Sea stock and subscription prices from 14 May 1720, when the first
listing of a subscription price appears, until 27 September 1720, when the
South Sea Bubble had burst.48 Subscription prices continue to be available
during the rest of 1720 but their value for this study is questionable, since
towards the end of September the South Sea Company began to consider
proposals for remedial action involving retrospective adjustment of the last
two subscription prices. The first detailed reports of this appeared in
Castaing in early October. A cut-off point of 27 September is used to allow
for possible anticipation of such action, which may have led to some
distortion in subscription prices.
The first three partly paid subscriptions may be viewed as equivalent
to fully paid shares, apart from the phasing of the subscription payments,
since they were entitled to the 10 per cent midsummer stock dividend of
1720, and to the cash dividends promised in September. 49 The fourth

44
Ibid., pp. 183–4.
45
Ibid., pp. 169–70.
46
Parsons, Behaviour of prices, p. 60.
47
Ibid., p. 67.
48
One potential point of confusion with Freke’s subscription prices is that he changes his method of
quotation. Until 24 June 1720, his prices exclude down payments and calls, which therefore have to be
added back to arrive at full market prices. After 24 June 1720, Freke makes clear that his prices are
quoted on an inclusive basis (i.e. including down payments and calls) except in those cases—mainly
affecting the fourth subscription—where he expressed the price as ‘premium’ or ‘discount’ on the
amount paid. In contrast, Castaing’s subscription prices are quoted throughout the Bubble period on
an exclusive basis.
49
Dale, The first crash, p. 162.
© Economic History Society 2005
 14680289, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1468-0289.2005.00304.x by Swinburne University Of, Wiley Online Library on [18/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
IRRATIONAL BEHAVIOUR DURING THE SOUTH SEA BUBBLE 243
subscription was not subject to the 10 per cent stock dividend, and therefore
a 10 per cent upward adjustment is made to the sum of quoted prices and
discounted calls for the fourth subscription, to put it on a comparable basis
with the first three subscriptions and the stock price.
The interpretation of subscription prices is complicated by the fact that
the receipts for the subscriptions were not immediately delivered out by the
company. Consequently, where receipts had not been delivered, transac-
tions in subscriptions were in effect forward contracts; the (indeterminate)
settlement date being the date on which delivery was made to the buyer.
Consequently, there were two types of contract in the subscriptions’ sec-
ondary market: ‘forward/forward contracts’, where both payment for and
delivery of receipts were deferred until the receipts were given out, and
‘spot/forward contracts’, where buyers paid cash upfront in anticipation of
being provided with receipts when they became available. This distinction
between contracts for time (forward/forward) and contracts for money
(spot/forward) is clearly described in the contemporary pamphlet litera-
ture,50 as well as in litigation relating to South Sea contracts.51 Contracts
for money were risky, since they involved in effect an unsecured loan to the
seller of receipts, and they also implied loss of interest to the buyer on the
money deposited. Accordingly, the spot/forward (designated by Freke as ‘for
money’) quoted prices are significantly lower than the regular forward/
forward prices quoted by Freke.
To handle these issues during the period when subscription receipts had
not been delivered out by the company, a number of choices and adjust-
ments in relation to the subscription price data are made. First, where
Freke gives separate prices for regular and ‘for money’ transactions—as in
the case of the third and fourth subscriptions—the former are used on the
grounds that settlement risk distorts the latter. Second, where no separate
prices are given for the two types of transaction it is assumed that the prices
quoted relate to forward/forward transactions, since this is consistent with
Freke’s terminology. Third, in order to calculate a present-value or spot-
price equivalent for the subscription receipts, an appropriate discount rate
is applied (see below) to the quoted forward price. For this purpose, an
assumption has to be made about the expected maturity of the forward
transaction; that is, on what date the subscription receipts were expected to
be delivered out by the company. It is known that the subscription receipts
for the first two subscriptions were given out between six weeks and two
months after the issue date.52 Since it is likely that market expectations

50
See, for example, ‘A full confutation of the subscribers pretensions to receipts for the first payment
made upon the third and fourth subscriptions . . .’ London, 1721, p. 11.
51
See, for example, Thomas Paterson, appellant, Richard Graham, respondent, the appellant’s case,
London, 1733, pp. 1–2.
52
Receipts for the first subscription had to be given out by the time of the first call on 14 June (since
the company was prepared to advance the money due on the security of the receipts), but it is known
from the South Sea Company’s Court minutes that the form of these receipts had not been finalized as
of 2 June 1720: Minutes of the Court of Directors of the South Sea Company, 1720, British Library.
© Economic History Society 2005
 14680289, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1468-0289.2005.00304.x by Swinburne University Of, Wiley Online Library on [18/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
244 RICHARD S. DALE, JOHNNIE E. V. JOHNSON, AND LEILEI TANG

would be based on this experience, a two-month interval is assumed

between the issue date of each subscription and the expected delivery of the
corresponding receipts.
Having suitably discounted market prices for subscriptions during the
period when receipts had not been delivered out by the Company, each
subscription price for the period 14 May to 27 September 1720 is converted
into a (fully paid) ‘stock price equivalent’, by adding the present value of
all unpaid calls to the market price. For the second subscription, the
schedule of calls was changed around 22 July 1720. Consequently, to
produce ‘stock price equivalents’, the original schedule is used until 22 July,
and thereafter the revised schedule is employed. As noted above, the fourth
subscription did not qualify for the 10 per cent midsummer dividend; the
discounted market price plus the discounted calls for subscription four are
therefore increased by 10 per cent to produce a valuation that can be
compared directly with the stock price.
The aim is to compare the stock price equivalent of the South Sea
Company’s subscription prices with the Company’s stock prices. How-
ever, a complication occurs with quoted South Sea stock prices, since
from 23 June to 22 August 1720 the transfer books were closed to allow
processing of the midsummer dividend. During this period, shares could
not be transferred and the quoted stock prices are for ‘the opening of the
books’ (i.e. they are forward prices). Consequently, to calculate a present
value equivalent, an appropriate discount rate is applied to the quoted
prices during the period the books were closed. Castaing generally quotes
South Sea stock prices ‘ex-dividend’, whereas Freke’s prices are cum-
dividend (these are therefore approximately 10 per cent higher than
Castaing’s prices). For purposes of comparison with prices prior to clo-
sure of the books, the cum-dividend price is the more appropriate, since
the 10 per cent midsummer dividend amounted in effect to a capitaliza-
tion or bonus share issue, which should not of itself have affected the
stock price.53
From the above discussion, it is clear that an appropriate discount rate
has to be selected in order to: (a) produce an equivalent stock price for
South Sea stock during the period when the transfer books were closed;
(b) produce an equivalent stock price for subscriptions during the period
in which the subscription receipts had not been delivered out by the Com-
pany; and (c) discount the unpaid calls on subscriptions to arrive at a
present value market price. A 5 per cent discount rate was applied through-
out, since a number of factors suggest that a rate close to this figure is
appropriate. First, the South Sea Company and the Bank of England
lent money freely on the security of their stock from the spring of 1720 at
4–5 per cent, while the Bank also typically charged 5 per cent for discount-

53
It is not clear exactly when those entitled to the dividend were credited with the 10 per cent stock
and the prices became ex-dividend, but it is clear from a notice published in Freke’s Prices of Stocks etc.
on 20 September 1720 that the prices he quoted in September continued to be cum-dividend.
© Economic History Society 2005
 14680289, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1468-0289.2005.00304.x by Swinburne University Of, Wiley Online Library on [18/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
IRRATIONAL BEHAVIOUR DURING THE SOUTH SEA BUBBLE 245
ing promissory notes and bills of exchange. Second, the contemporary
benchmark short-term interest rate, the yield on the East India
Company bonds, averaged 5 per cent during the latter half of 1720. 54 Third,
subscribers to the first and second subscriptions who were prepared to
prepay their remaining calls were given a 4 per cent per annum discount by
the South Sea Company. Fourth, during the summer of 1720, some inves-
tors were able to borrow at rates as low as 5 per cent from goldsmith-
bankers. Fifth, Hutcheson, in his own valuations of South Sea stock,
assumes a discount rate of 4 per cent.55 Lastly, Carlos, Moyen, and Hill
provide evidence, based on historical studies, to justify a discount rate of
5 per cent throughout the period from Spring 1720 until January 1725,
when the final calls on the subscriptions were due.56
Neal’s view that a credit crunch may have occurred at the height of the
Bubble57 is disputed by some scholars,58 and there is evidence that the Bank
of England continued to lend at 5 per cent throughout the period covered
by the analysis, even though it applied stricter criteria for lending in the
later stages of the Bubble.59 Hutcheson notes that some individuals were
paying very high rates as early as April 1720,60 but care must be exercised
in distinguishing rates charged to individuals without an established credit
standing who posed a high default risk, from properly secured lending rates
charged to those of good standing. In addition, as noted above, regardless
of their credit standing or the prevailing market conditions, individuals were
able to borrow at between 4 and 5 per cent on the security of their stock
in the Bank of England or the South Sea Company, or even on the security
of subscriptions in the South Sea Company.
In summary, the evidence suggests that 5 per cent is an appropriate
discount rate for the period under investigation, although it is acknowledged
that some loans to certain individuals would have been made at a higher
rate. Consequently, when applying the adjustments indicated above, a dis-
count rate of 5 per cent is employed, but sensitivity analysis is also under-
taken to explore the levels to which rates would need to have risen to
contradict the conclusions of the analysis.
Applying the adjustments indicated above, and using a discount rate of
5 per cent, an ‘equivalent stock price’ is produced for South Sea stock at
each time t (St, hereafter referred to as the South Sea stock price), and a
‘stock price equivalent’ is determined for subscription j’s receipts at time t

54
East India bonds were perceived as high quality instruments almost equivalent to cash in normal
market situations. However, they did of course carry some degree of default risk, and during the credit
scare of the first few days of October 1720, East India bonds fell to a discount: Weiller and Mirowski,
‘Rates of interest’, p. 5.
55
Dale, The first crash, pp. 86,116.
56
Carlos, Moyen, and Hill, ‘Royal African Company share prices’, p. 76.
57
Neal, Rise of financial capitalism, p. 101.
58
Dale, The first crash, pp. 133–5.
59
Bank of England Minutes of the Court of Directors, 22 September 1720, Bank of England MS
collection.
60
Dale, The first crash, p. 134.
© Economic History Society 2005
 14680289, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1468-0289.2005.00304.x by Swinburne University Of, Wiley Online Library on [18/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
246 RICHARD S. DALE, JOHNNIE E. V. JOHNSON, AND LEILEI TANG

1250

1050

850
Value

Stock
650
Subs. 1

Subs. 2
450
Subs. 3

250 Subs. 4
/5

/6

/6

/7

/7

/8

/8

/9
1/

8/
14

15

30

14

28

11

25

23
Date (1720)

Figure 1. South Sea Stock and Subscription Price Series (14/5–27/9, 1720)
Notes: The most complete series of Freke’s Prices of Stocks etc., housed in the British Library, is used to construct the
graph. Freke does not provide prices for subscription 3, during 18–21 June, and for this period, prices are taken from
John Castaing’s Course of the Exchange. Freke’s Prices of Stocks etc., issue number 96, covering data for 21–23 May 1720
is missing. Castaing’s The Course of the Exchange does not supply subscription prices during this period either and,
consequently, the data for this period is omitted from the graph.
To aid comparison between the series, non-trading days (Sundays) are not included on the date axis.

Table 2. Statistical properties of the stock and subscription price series

Normality, test
Series Mean Std Skew Kurtosis statistica (sig.)

stock 762.54 203.12 -0.82 -0.43 1.60 (0.01)

subscription one 765.50 163.56 -1.34 0.58 2.47 (0.00)
subscription two 802.34 172.26 -1.33 0.47 2.53 (0.00)
subscription three 1,095.24 84.17 -0.91 -0.33 1.68 (0.01)
subscription four 1,043.18 47.26 1.35 3.09 1.44 (0.03)

Notes: The sources of data and the method of compilation are discussed in appendix II.
a
The Kolmogorov-Smirnov Z statistic.

© Economic History Society 2005

 14680289, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1468-0289.2005.00304.x by Swinburne University Of, Wiley Online Library on [18/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
IRRATIONAL BEHAVIOUR DURING THE SOUTH SEA BUBBLE 247
( Ctj , hereafter referred to as the price of subscription j, j = 1, 2, 3, 4).61 The
series of stock and subscription prices for the period 14 May to 27 Septem-
ber 1720 are displayed graphically in figure 1, and their statistical properties
are summarized in table 2.

V
The aim is to explore the relationship between the South Sea stock price
series (St) and each of the subscription price series ( Ctj ), and this analysis
is conducted in four stages. First, a static analysis is undertaken, simply
comparing stock and subscription prices on and close to the issue date for
each of the subscriptions. Second, residual analysis compares the differences
between the price series for South Sea stock (St) and the price series for
each subscription ( Ctj ). Third, a comparison is made between the discount
rates that would be required to align the South Sea stock and subscription
prices. Lastly, the relationship between stock and subscription prices is
investigated using co-integration analysis, which captures both the short-
run dynamics of the price series and the long-run equilibrium relationships.
Since the stock and subscription receipts (with prices adjusted in the man-
ner indicated above) represent substitutable assets, a rational expectations
framework would predict that at a given time t:
Ctj = a j S tj + et (1)
j
where a = 1 and et represents a random error term.
A simple OLS regression could be used to test this relationship if the St
and Ct series were stationary processes. However, most financial time series
are not stationary, in that the underlying distribution changes through time,
so that the series’ unconditional variances do not exist. Applying OLS
regression to a non-stationary time series can lead to ‘spurious regression’
in which a significant relationship may be detected among totally unrelated
variables. The non-stationarity could be removed by simply taking the first
difference of each series and using OLS to estimate the following:
Ctj - Ctj-1 = a j (S t - S t -1 ) + f t (2)
where ft represents a random error term. Rationality on the part of investors
would imply that a j = 1.
However, such a formulation, whilst capturing the short-run dynamics of
the time series, fails to incorporate information concerning the long-run
equilibrium relationship. These failings are overcome by employing co-
integration, introduced by Granger,62 and details of the procedure are given
in appendix IV.
61
Throughout the paper, a continuous discounting procedure is employed such that the discounted
i ˆ
value (£y) of a given amount (£x) is calculated as follows: y = xe - Ê t , where i = annual discount
Ë 365 ¯
rate, t = discount period in days.
62
Granger, ‘Econometric model specificating’, p. 127.
© Economic History Society 2005
 14680289, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1468-0289.2005.00304.x by Swinburne University Of, Wiley Online Library on [18/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
248 RICHARD S. DALE, JOHNNIE E. V. JOHNSON, AND LEILEI TANG

Since the South Sea stock and the subscription receipts represent substi-
tutable assets, then rational behaviour on the part of investors should result
in a close agreement between the adjusted subscription ( Ctj ) and stock
prices (St). Investor rationality should, therefore, lead to co-integration
between the stock and each subscription price series. South Sea stock
represents the asset to which subscription receipts relate and, therefore,
stock prices are expected to Granger-cause subscription prices. Using the
methodology indicated above, these propositions are tested.

VI
In the static analysis, stock prices are compared with the price of each
subscription (a) on its issue date, and (b) one week after the issue date. The
market premium on each new subscription issue relative to the issue price
is also estimated (by comparing the market price of each subscription one
week after its issue with the first down payment required on the issue date).
The results are displayed in table 3 and demonstrate that, other than for
the first subscription, the issue price (i.e. the ‘stock price equivalent’) of
subscriptions two, three and four ( Coj ) exceed the prices (i.e. ‘equivalent
stock price’, So) for South Sea stock (by between 9.7 and 27.6 per cent),
the third and fourth subscriptions registering the largest premiums. Each
subscription traded at a considerable market premium relative to its first
down payment, the third and fourth subscriptions trading at 150 per cent
and 53.5 per cent above the initial down payment within one week of being
issued. In addition, one week after each subscription issue, the prices of
subscriptions two, three, and four significantly exceeded the price of South
Sea stock (by 11.2 per cent, 43.3 per cent, and 43.2 per cent respectively).
Consequently, the static analysis reveals that the first subscription was
issued at ‘fair value’ relative to the stock price, but later subscriptions were
issued at substantial premiums on the stock price. Within a week of its issue
each subscription traded at an even larger premium compared to the down
payment made at issue date; paradoxically, the subscription that was issued
at the most substantial premium over the stock price (subscription three),
displayed the largest premium over its first down payment. In addition,
shortly after its issue date, each subscription traded at a premium over the
price for the South Sea stock—a directly substitutable instrument.
Residual analysis is employed to investigate the tendency for subscription
receipts to trade at a premium on the stock price over the longer term. To
achieve this objective, the price series for South Sea stock (St) is compared
with the price series for each subscription ( Ctj ). Figure 1 suggests a reason-
ably close agreement between stock and subscription prices for the first two
subscriptions, but with a large and increasing disparity between the daily
prices of the third and fourth subscriptions and the equivalent stock prices.
Daily prices for subscription one and subscription two exceed stock
prices by 58.4 per cent and 62.8 per cent of the time, respectively, and the
© Economic History Society 2005
 14680289, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1468-0289.2005.00304.x by Swinburne University Of, Wiley Online Library on [18/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
IRRATIONAL BEHAVIOUR DURING THE SOUTH SEA BUBBLE 249
Table 3. Comparison of stock and subscription prices close to subscription issue
date
Subscription

1 2 3 4

Issue Date (1720) April 14 April 29 June 16 August 24

New issue price 300 400 1,000 1,000
First subscription payment 60 40 100 200
SPE of issue price 291 372 894 1,046.1
ESP at subscription issue date 294 339 737 820
Premium of subscription over -3 (-1.0%) 33 (9.7%) 157 (21.3%) 226 (27.6%)
stock price at issue date
Subscription prices one week 66 44 250 307
after issue
Market premium of 6 (10%) 4 (10%) 150 (150%) 107 (53.5%)
subscriptions over first
payment
ESP one week after 344 338 763 812
subscription issue date
SPE one week after their issue 297 376 1,093 1,163
Premium of subscription over -47 (-13.7%) 38 (11.2%) 330 (43.3%) 351 (43.2%)
stock price one week after
issue

Note: The sources of data and the method of compilation are discussed in appendix III.

prices of subscriptions three and four always exceed stock prices. Mean daily
subscription prices exceed stock prices by an increasing margin for succes-
sive subscriptions (subscription one: £2.09; subscription two: £38.36; sub-
scription three: £284.04; and subscription four: £434.82). Whilst these
results are indicative of differences in the daily stock and subscription prices,
each series displays features of a non-normal distribution (see table 2).
Consequently, non-parametric tests are employed to test for differences
between these series. In particular, the Wilcoxian signed-rank test is used,
since this incorporates information about the sign and margin of difference
between two price series. These tests confirm the picture to emerge from
comparing the differences in mean daily prices, namely that stock prices
are not significantly different from subscription one prices (z = -0.623,
sig. = 0.533) but subscription two, three, and four prices significantly exceed
stock prices (z = -5.12, sig. = 0.000; z = -8.05, sig. = 0.000; z = -4.78,
sig. = 0.000; respectively). Figure 1 suggests that the residuals for the third
and fourth subscriptions increased over time with the prices of subscription
three and subscription four exceeding the stock price by, respectively,
£328.3 and £238.6 on 24 August, £313.3 and £344.5 on 8 September,
£549.8 and £610.5 on 22 September, and £605.4 and £650.4 on
27 September 1720. These results indicate that the discount rates required
to equate the present value of the stock and subscription prices rose signif-
icantly in August and September 1720.
© Economic History Society 2005
 14680289, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1468-0289.2005.00304.x by Swinburne University Of, Wiley Online Library on [18/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
250 RICHARD S. DALE, JOHNNIE E. V. JOHNSON, AND LEILEI TANG

The discount rates that would be required to bring the South Sea stock
(St) and the subscription prices ( Ctj ) into line are calculated. The results
are displayed in figure 2, and two things are clear from this analysis. First,
markedly different discount rates must be applied to bring the values of
different subscriptions into line with the price of South Sea stock. This
implies that to bring the market values of each of the subscriptions into line
with each other would require the simultaneous application of significantly
different discount rates. Second, the discount rate required, over the staged-
payment lives of the four subscriptions—16, 36, 54, and 24 months respec-
tively for subscriptions one, two, three, and four—to align the prices of
South Sea stock with the prices for the first, third, and fourth subscriptions
rose to well over 50 per cent late in September and, in the case of subscrip-
tion four, to over 100 per cent. The large negative (in June and July) and
large positive (in September) discount rates required to equalize prices of
subscription one receipts and stock may appear surprising, given the rela-
tively small differences between the price of subscription one receipts and
the price of South Sea stock (shown in figure 1). However, calls for sub-
scription one were required over a much shorter period than for any of the
other subscriptions and, consequently, larger differences in discount rates
are required to bring the equivalent stock and subscription prices into line
for this subscription.
The large discount rates required to equalize stock prices with those of
subscriptions one, three, and four in late August and September are com-
pletely out of line with the 5 per cent discount rate that has shown to be
appropriate for this period. In addition, whilst there is evidence that high
interest rates were charged on some personal borrowing during this period,
these typically applied only to short term loans of around two months. 63
East India bonds were yielding only 5 per cent, and it is inconceivable that
discount rates in excess of 50 per cent over the staged payment lives of the
various subscriptions (16–54 months) were applicable during this period.
The evidence presented above points to considerable divergence between
the South Sea stock and subscription prices, particularly for the third and
fourth subscriptions. To explore this further, co-integration analysis was
conducted, as outlined in appendix IV.
The stationarity of each of the stock and subscription price series was
tested by applying the Augmented Dickey-Fuller (ADF) test. The results
are reported in table 4. The null hypothesis of a unit root for all the price
series cannot be rejected, confirming that all the price series are non-
stationary processes in levels. The power of the ADF test is low when dealing
with small samples64 and, consequently, there is a low probability of reject-
ing the null hypothesis when it is false. This applies particularly to subscrip-
tions three and four (see table 4 for details of the test power and size).

63
Dickson, Financial revolution in England , pp. 191, 470.
64
Campbell and Perron, ‘Unit roots’, p. 21.
© Economic History Society 2005
 14680289, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1468-0289.2005.00304.x by Swinburne University Of, Wiley Online Library on [18/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
IRRATIONAL BEHAVIOUR DURING THE SOUTH SEA BUBBLE 251

2
Discount Rate (%)

0
Subs.1

Subs.2
–1
Subs.3

–2 Subs.4
/5

/6

/6

/7

/7

/8

/8

/9
1/

8/
14

15

30

14

28

11

25

23

Date (1720)

Figure 2. Discount rate to equate South Sea stock and subscription prices
Notes: To aid comparison between the series, non-trading days (Sundays) are not included on the date axis. There were
two dates (28/7 and 9/9 for subscriptions 1 and 2) for which a discount rate which equated the equivalent stock price
of the subscription with the equivalent spot price of stock could not be determined—the discount rate for these dates
is left undefined.
The equalizing discount rate applicable to the first subscription, and to a lesser extent the other subscriptions, falls
dramatically on 23 June 1720, which was when the books were closed. A partial explanation for this disparity is that
the first subscription, having a relatively short repayment period, magnifies changes in the equalizing discount rate as
compared to the later subscriptions, with longer repayment schedules.

However, the consequences of not rejecting the null hypothesis when it

is false are less severe than rejecting it when it is true, since in the latter
situation OLS would be employed, and may well discover a spurious rela-
tionship between the variables.65 In the case where the series is stationary
and it is assumed to be non-stationary, the co-integration procedure can
still find the true relationship between the variables. Consequently, the unit
root tests are taken at face value, the series are assumed to be non-
stationary, and the Johansen and Juselius multivariate co-integration test is

65
Engle and Granger, ‘Co-integration and error correction’, p. 271.
© Economic History Society 2005
 14680289, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1468-0289.2005.00304.x by Swinburne University Of, Wiley Online Library on [18/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
252 RICHARD S. DALE, JOHNNIE E. V. JOHNSON, AND LEILEI TANG

Table 4. Unit-root test statistics for stock and subscription prices

Augmented Dickey-Fuller test statistics Stock Sub. 1 Sub. 2 Sub. 3 Sub. 4

t (without trend)a -0.767 -1.021 -0.830 -0.177 -1.588

Critical value at 5% -2.887 -2.887 -2.887 -2.896 -2.975
Number of observationsb 114 114 114 86 29

Notes: a t represents the test statistic using a lag of one day.

b The ADF unit root test has low power when dealing with small samples and, therefore, for subscriptions three and
four in particular, the probability of rejecting the null hypothesis of a unit root when it is false is low. Simulations
involving 50,000 replications indicate that for the third subscription the test size was 0.047 and the power was 0.07,
0.25, 0.64, and 0.99 for alternative hypotheses 0.99, 0.95, 0.90, and 0.80 respectively; similarly, for the fourth
subscription the test size was 0.049 and the power was 0.06, 0.09, 0.16, 0.41, 0.68, and 0.89 for alternative hypotheses
0.99, 0.95, 0.90, 0.80, 0.70, and 0.60 respectively.

employed to explore whether a stationary linear combination of the stock

and subscription price series exists.66
The co-integration test is sensitive to the order of lags included in vector
auto-regression (VAR) models.67 A general-to-specific approach is therefore
employed to decide on the lag length, starting with a fourth-order VAR
model, and then sequentially reducing the lag structure. 68 The Schwarz
information criterion (SC) is used to choose the lag length. This criterion
suggests one lag in the VAR for stock and subscriptions one, two, and four,
and two lags in the VAR for stock and subscription three, as reported in
part A of table 5.
The results of co-integration estimation are displayed in part B of table 5.
Two hypotheses are tested, first, that no co-integration exists (i.e. rank = 0),
and second that the number of cointegrating vectors is at most one (i.e. rank
£ 1). The null hypothesis of no co-integration is not rejected for subscrip-
tions one, three, and four at the 5 per cent level, on the basis of the
maximum eigenvalue and trace tests.69 The null hypothesis that stock (St)
and subscription prices (Ct) are not co-integrated is rejected at the 5 per cent
level for subscription two using the maximum eigenvalue test, but not using
the trace test. The null hypothesis that the number of co-integrating vectors
is at most one between the stock and the price series for subscription two
cannot be rejected at the 5 per cent level.
To explore the relationship between the prices for South Sea stock and
the prices for the second subscription further, an error-correction model

66
Johansen and Juselius, ‘Inference on cointegration’, p. 170.
67
Gonzalo, ‘Estimating long-run equilibrium’, p. 210.
68
The issue of setting the appropriate lag length is important. While a large value for lag length may
result in over-parameterization, which affects the estimation of co-integration rank, small values may
distort the size of the tests: Cheung and Lai, ‘Finite-sample sizes’, p. 322.
69
Reimers finds that with small samples it is generally easier to find co-integration relationships where
none exist. Consequently, since no co-integration relationship is found for subscriptions one, three, and
four (which involve small samples), this is highly suggestive of no co-integration existing between these
series. In fact, since the Johansen procedure over-rejects when the null hypothesis of no co-integration
is true in small samples, the maximum eigenvalue and trace tests that are applied to all four subscription
series are adjusted for the small sample size: Reimers, ‘Multivariate cointegration’.
© Economic History Society 2005
 14680289, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1468-0289.2005.00304.x by Swinburne University Of, Wiley Online Library on [18/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
IRRATIONAL BEHAVIOUR DURING THE SOUTH SEA BUBBLE 253
Table 5. Results of Johansen’s multivariate co-integration tests for stock and
subscription price series
Part A: Lag order selection using the SC criteriona

Lag order Stock vs. Sub1 Stock vs. Sub2 Stock vs. Sub3 Stock vs. Sub4

0 17.862 18.462 18.125 17.607

1 13.247 13.649 12.718 14.204
2 13.296 13.665 12.601 14.515
3 13.409 13.767 12.797 14.741
4 13.566 13.919 12.983 15.207

Part B: Johansen multivariate co-integration tests

Test statistics

Null hypothesis of rank r lmax b

lmax (95% value)c Trace Trace (95% value)

stock vs. sub1. (114 observations)

r=0 9.92 15.7 10.12 20.0
r£1 0.20 9.2 0.20 9.2
stock vs. sub2. (114 observations)
r=0 15.87* 15.7 16.26 20.0
r£1 0.39 9.2 0.38 9.2
stock vs. sub3. (86 observations)
r=0 6.17 15.7 8.57 20.0
r£1 2.40 9.2 2.40 9.2
stock vs. sub4. (29 observations)
r=0 6.69 15.7 7.06 20.0
r£1 0.37 9.2 0.37 9.2

Notes: a The Schwarz criterion is defined as SC = log| Ŵ| + klog(T )T -1, where -2T -1log| Ŵ| is the estimated log
likelihood, T is the sample size, and k is the number of parameters estimated. The selected lag-length is decided by
the minimum criterion values.
b The test statistics are adjusted to account for the small sample size, according to the procedures recommended by
Reimers ‘Multivariate co-integration’ (1992), as follows: lmax statistic = -(T - nm)ln(1 - lr ), the maximum eigenvalue
n
test statistic (r vs. r + 1); trace statistic = -(T - nm) Â ln(1 - li); where T is the number of observations, n is the number
i = r +1
of variables, m is thelag length, r is the dimension of the co-integration space and li is the ith smallest squared canonical
correlation in Johansen and Juselius ‘Inference on co-integration’ (1990).
c lmax (95% value) and trace (95% value) are the critical values for the lmax and trace tests, respectively.
* = rejection of the null hypothesis at the 5% level.

(ECM) is estimated, with an error-correction term and lagged values of the

first differences of each series explicitly included.70 The OLS estimation for
the error-correction parameters and the lagged variable parameters are
presented in table 6.
The results displayed in table 6 indicate that in the ECM for stock and
subscription two prices the error-correction coefficient as (coefficient =
-0.1065, t = -2.273) is significant at the 5 per cent level, but ac (coefficient

70
Granger first noted the relationship between co-integration and ECM, and Granger proves how
co-integrated series may be represented by the ECM: Granger, ‘Econometric model specificating’,
pp. 128–9; Granger, ‘Cointegrated economic variables’, p. 216.
© Economic History Society 2005
 14680289, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1468-0289.2005.00304.x by Swinburne University Of, Wiley Online Library on [18/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
254 RICHARD S. DALE, JOHNNIE E. V. JOHNSON, AND LEILEI TANG

Table 6. OLS estimation of the error correction

model for stock and subscription two
Estimation of the co-integrating vector:

St = -218.6 + 1.2212Ct2

DSt = a s zˆt -1 + qss,t -1DSt -1 + qsc,2t -1DCt -1 + e st

Variables Coefficients t-values

as -0.1065 -2.273
qss,t -1 0.0752 0.762
qsc,2t -1 0.3029 2.738*

DCt = a c zˆt -1 + qcs2 ,t -1DSt -1 + qcc22y ,t -1DCt -1 + e ct

Variables Coefficients t-values

ac 0.0181 0.380
qcs2 ,t -1 0.2132 2.123*
qcc22,t -1 0.1072 0.951
zˆ t = St + 218.6 - 1.2212Ct2

Note: *significant at the 5% level.

= 0.0181, t = 0.380) is not. This confirms that there is a co-integration

relationship between the prices of stock and subscription two, and suggests
that stock prices respond to the previous period’s deviations from the long-
term equilibrium, while subscription two prices do not; that is, all long-run
adjustments between stock and subscription two prices arise from stock
price changes.
In terms of short-term price movements, there is evidence (see table 6)
that stock prices Granger-cause subscription two prices ( q cs2 ,t -1 = 0.2132,
t = 2.12) and that subscription two prices Granger-cause stock prices
( q sc,2t -1 = 0.3029, t = 2.74). Similarly, short-run Granger causality analysis is
conducted by estimating VAR in first differences for subscriptions one,
three, and four. Table 7 reports the results of this analysis. The results
indicate that the underlying stock prices and subscription one prices
Granger-cause each other in the short-run ( q cs1 ,t -1 = 0.1974, t = 1.96;
q sc,1t -1 = 0.2746, t = 1.99). Stock price changes have no influence on the
change in subscription three prices in the short run ( q cs3 ,t -1 = 0.0970,
t = 1.39; q cs ,t -2 = 0.0516, t = 0.83). However, changes in subscription three
3
prices Granger-cause stock price movements ( q sc,3t -1 = 0.9564, t = 5.28). The
results displayed in table 7 also suggest that there is no short-run relation-
ship between stock prices and subscription four prices ( q cs4 ,t -1 = 0.2003,
t = 0.76; q sc,4t -1 = 0.0343, t = 0.20).
© Economic History Society 2005
 14680289, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1468-0289.2005.00304.x by Swinburne University Of, Wiley Online Library on [18/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
IRRATIONAL BEHAVIOUR DURING THE SOUTH SEA BUBBLE 255
Table 7. OLS Estimation of the VAR models for stock and subscriptions one,
three, and four
DSt = a s + qss,t -1DSt -1 + qss,t -2 DSt -2 + qsc,jt -1DCt -1 + qsc,jt -2 DCt -2 + e st

Stock with Sub.1 Stock with Sub.3 Stock with Sub.4

Variables Coefficients t-values Coefficients t-values Coefficients t-values

as -0.4418 -0.150 -4.467 -1.472 -23.08 -3.546*

qss,t -1 0.0889 0.754 -0.0103 -0.091 -0.2243 -1.100
qss,t -2 — — -0.0386 -0.382 — —
qsc,jt -1 0.2746 1.990* 0.9564 5.283* 0.0343 0.2040
qsc,jt -2 — — 0.1822 0.885 — —

DCt = a c + qcsj ,t -1DSt -1 + qcsj ,t -2 DSt -2 + qccjj ,t -1DCt -1 + qccjj ,t -2 DCt -2 + e ct

Stock with Sub.1 Stock with Sub. 3 Stock with Sub. 4

Variables Coefficients t-values Coefficients t-values Coefficients t-values

ac -0.0945 -0.038 -0.5380 -0.287 -0.5179 -0.062

qcsj ,t -1 0.1974 1.962* 0.0970 1.394 0.2003 0.759
qcsj ,t -2 — — 0.0516 0.828 — —
qccjj ,t -1 0.0987 0.838 0.1378 1.232 -0.0184 -0.084
qccjj ,t -2 — — 0.0338 0.267 — —

Note: *Significant at 5% level

Overall, the co-integration analysis identifies no long-run relationship

between the price series for stock and subscriptions one, three, and four.
For subscription two, a co-integration relationship is found with stock
prices, with all long-run adjustments between these price series arising from
stock price changes. The picture that emerged from the earlier static, resid-
uals, and discount rate analyses revealed an increasing separation between
the prices of receipts for later subscriptions and the underlying stock. This
pattern is confirmed by Granger-causality analysis, the short-run prices of
both stock and subscription receipts Granger-causing each other for the first
two subscriptions, subscription three prices Granger-causing stock prices,
and no short-run relationship existing between stock and subscription four
prices.

VII
In examining the results presented above, it is important to recall that ‘stock
price equivalents’ of subscription prices ( Ctj ) are being compared with
‘equivalent stock prices’ for South Sea stock (St). These are directly com-
parable prices, since the South Sea stock and the subscription receipts are
directly substitutable instruments, representing equivalent claims on assets
© Economic History Society 2005
 14680289, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1468-0289.2005.00304.x by Swinburne University Of, Wiley Online Library on [18/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
256 RICHARD S. DALE, JOHNNIE E. V. JOHNSON, AND LEILEI TANG

and dividends. Furthermore, investors were free to exploit any arbitrage

opportunities either by selling subscription receipts and buying stock (in
the forward market when the books were closed) or by selling ‘expensive’
subscriptions and buying more favourably priced issues.
The results of the static and residual analyses present a consistent picture
of a reasonably close correspondence between stock and subscription
prices for the first subscription, but a significant and widening difference
over time between stock and subscription prices for the third and fourth
subscriptions. The discount rates required in order to equate both subscrip-
tion three and subscription four prices with the stock price rise to incon-
ceivable levels during September. Furthermore, there is no interest rate
that could be applied simultaneously to all subscription series to equate the
subscription price series with each other. During this period, subscription
three and four receipts traded at prices that were significantly more expen-
sive than subscription one and two receipts. This anomaly requires
explanation.
The lack of co-integration between the prices of South Sea stock and
subscriptions one, three, and four indicates a complete absence of a long-
run equilibrium relationship between the prices of these substitutable
instruments. In addition, although a co-integration relationship is found
between the prices of stock and subscription two, it is not subscription two
prices that adjust to deviations from this long-run relationship (as would be
expected in a rational expectations framework), but rather adjustments are
confined to stock prices. In the short run, stock prices appear to influence
the prices of subscription one and two receipts (as might be expected in a
rational market) but, in addition, subscription one and two prices influence
stock prices. For subscription three there is not even a short-run influence
of stock prices on subscription prices but, remarkably, subscription three
prices appear to influence stock prices. Furthermore, there is a complete
separation of the prices of stock and subscription four receipts with no long-
or short-term relationship being discovered between these price series.
In summary, the results taken together pose a number of challenges to a
rational explanation for investor behaviour during the South Sea Bubble.
First, there appears to be a separation of the valuation of subscriptions from
those of the underlying stock: subscription receipts, particularly subscrip-
tions three and four, traded at prices far in excess of those for the directly
substitutable South Sea stock, no long-run equilibrium relationship is
detected between the prices of South Sea stock and the prices of three of
the four subscriptions, and where a long-run equilibrium relationship is
found (between the prices of stock and subscription two), it appears that
stock prices rather than subscription prices are adjusting to deviations from
this equilibrium. Second, the prices for subscriptions three and four
diverged from stock prices increasingly during September 1720, as the
South Sea Bubble burst. Third, the prices of the four subscriptions differed
significantly from each other.

© Economic History Society 2005

 14680289, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1468-0289.2005.00304.x by Swinburne University Of, Wiley Online Library on [18/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
IRRATIONAL BEHAVIOUR DURING THE SOUTH SEA BUBBLE 257
Before concluding that irrationality overtook investors in 1720, it is
important to try to address each of the above challenges within a rational
expectations framework. For example, the first of these challenges might be
tackled by noting that subscription receipts required only a simple legal
assignment compared with the obligation to register share transfers on the
Company’s books. Consequently, these minor differences in administrative
effort required to trade the two assets might be expected to result in
subscription receipts trading at a small premium over South Sea stock.
However, it is highly unlikely that this could account for the large differences
in their market valuations that are identified here. In addition, subscription
receipts were subject to settlement risk because they were deferred contracts
in which the counter-party might fail, whereas stock transactions (other
than for the opening of the books) were based on delivery versus payment.
This settlement risk is likely to have offset any liquidity premium on the
subscription receipts. Equally, if there were a liquidity premium, this would
apply uniformly to all four subscriptions, but significant valuation differ-
ences between the subscriptions are observed.
A second ‘rational’ explanation for the disparity in stock and subscription
prices may involve investors’ perceptions of the enforceability of the sub-
scription contracts. If these were not legally enforceable, purchasers of
subscriptions might simply walk away from their obligations, in effect treat-
ing their contracts as options. However, there can be no doubt about the
legally binding status of subscriptions, since in early 1720 Parliament initi-
ated legislation that was specifically designed to ensure that any subscription
issue made by the South Sea Company would be legally valid, and trans-
ferable by simple legal assignment. Clause 27 of the so-called Bubble Act,
which finally received the royal assent on 11 June 1720, provided that ‘all
such subscriptions [by the South Sea Company] shall be firm and valid,
and all receipts made out and given, or to be made out and given, concern-
ing the same, shall be assignable at law by endorsement made or to be made
thereon’.71 Furthermore, after the Bubble had burst, Parliament reasserted
the binding nature of South Sea contracts by resolving in December 1720
that ‘all the subscriptions of public debts and incumbrances, and other
contracts made with the South Sea Company, by virtue of an Act made last
session, remain in the present state, unless altered for the ease and relief of
the proprietors by a General Court of the South Sea Company, or set aside
by due course of law’.72 The contemporary pamphlet literature also shows
that investors were well aware of the legal enforcement issue. 73
In addition, if subscriptions were treated as options by investors, they
should certainly not sell at a discount relative to stock. Yet subscriptions

71
6 Geo I, ch. 18, c. 27.
72
Cobbett, Parliamentary history, p. 680.
73
See, for example, ‘Further reasons offered and fresh occasions given for making void and annulling
fraudulent and usurious contracts’, London, 1721, pp. 3–6.

© Economic History Society 2005

 14680289, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1468-0289.2005.00304.x by Swinburne University Of, Wiley Online Library on [18/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
258 RICHARD S. DALE, JOHNNIE E. V. JOHNSON, AND LEILEI TANG

one and two traded at a discount on the share price for 41.6 per cent and
37.2 per cent of the period covered by the analysis, respectively, and there
is no evidence of a widening gap between stock valuations and those of
subscription two as the share price fell to levels below the amount of calls
outstanding late in September. These are not trading patterns one would
associate with financial instruments where investor’s viewed their outstand-
ing obligations as unenforceable. The ‘non-enforceability’ argument also
fails to account for the different Granger-causality relationships between
the prices of stock and the receipts for successive subscriptions. In addi-
tion, this explanation does not account for the increasing divergence
between the prices of stock and those of subscriptions three and four
during September, nor does it explain why the prices of the four subscrip-
tion receipts differed from each other. In summary, the evidence points to
an expectation on the part of subscribers that they could not walk away
from their obligations.
More generally, if subscription receipts were merely options, it is difficult
to explain how the South Sea Scheme got off the ground, since few rational
individuals would invest in a company that was known to be issuing call
options on a dramatic and ever-increasing scale. In addition, given that the
company itself would face highly uncertain cash flow from such operations,
it is highly unlikely that it would issue unenforceable subscriptions. Lastly,
it is difficult to understand why the government and Parliament would
specifically endorse the issue of subscriptions if they thought they were
unenforceable.
The evidence, therefore, indicates that subscription contracts created
legally binding obligations that could not be construed as options, and that
investors were fully aware of this. Nevertheless, there is scope for further
research in order to determine the extent to which subscription prices
deviated from modern option-pricing formulae.
A third potential ‘rational’ explanation for the difference between the
price of South Sea stock and subscription receipts may lie in the gearing
effect built into the subscriptions. Clearly, if investors believed that the
stock—and their related subscription receipts—was likely to rise in price, a
larger potential return for a given outlay could be obtained by investing in
subscriptions compared with stock. This may have caused the demand (and
hence the price) of subscriptions to rise beyond that of stock. However, the
gearing effect of subscriptions could be replicated to some extent by inves-
tors in stock who could borrow either from the company or from goldsmith-
bankers against the security of South Sea shares up to around two-thirds of
their market value. In fact, large numbers of investors took this option, and
£9 m was borrowed in this way from the company.74 Stockholders who took
this option entered a legally binding agreement, and the expectation would
have been that they could not walk away from their outstanding obligations

74
Cobbett, Parliamentary history, p. 736.
© Economic History Society 2005
 14680289, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1468-0289.2005.00304.x by Swinburne University Of, Wiley Online Library on [18/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
IRRATIONAL BEHAVIOUR DURING THE SOUTH SEA BUBBLE 259
any more than investors in subscriptions could default on future calls. A
gearing effect explanation of the results also fails to explain why subscription
one and two receipts traded, for substantial periods, at a deficit on stock
prices, and it does not explain why the equivalent stock prices of each
subscription differed significantly from each other.
The second challenge to rational investor behaviour stems from the
increasing difference between the prices of subscriptions three and four and
the underlying South Sea stock as the Bubble began to burst. This might
be explained by the fact that many investors borrowed heavily on the
security of stock, and as stock prices fell, lenders accelerated the process by
distressed selling of pledged stock.75 While borrowing did take place on the
security of subscription receipts, this was largely restricted to loans from
the South Sea Company. Borrowing was thus concentrated on stock, as
subscription receipts did not represent safe security because they had no
value if the calls were unpaid. The concentration on distressed selling of
stock (cf. subscription receipts) might partially account for the increasing
divergence between stock and subscription prices in September 1720. How-
ever, this explanation does not account for the different valuations placed
on the various subscriptions either during September, or earlier in 1720. It
is in fact difficult to develop a rational explanation for such differences in
directly substitutable instruments.
The co-integration analysis suggests that there is no long-run relation-
ship between the prices of stock and those of subscriptions one, three, and
four. While a long-run relationship is shown to exist between stock and
subscription two prices, it appears that investors focused more on the
movements of subscription two prices than on those of stock prices.
Granger-causality analysis confirms a separation of stock and subscription
prices for later subscriptions, with subscription three prices influencing
stock prices in the short run, and no short-run relationship existing
between the prices of stock and subscription four receipts. This suggests
that the South Sea Bubble was not a ‘rational bubble’, to the extent that it
did not have its origin in self-fulfilling expectations. It appears that inves-
tors viewed each new subscription as a distinct asset, with a valuation
increasingly divorced from the underlying stock. The view is reinforced by
the South Sea directors’ claims that they were under constant pressure
from the public to issue new subscriptions.76 The issue prices of successive
subscriptions were set at a progressively higher premium over the equiva-
lent stock prices. However, this appears to have simply spurred investors to
‘irrationally’ expect that the projectors of the scheme (Blunt and his cabal
of South Sea directors) would ensure that the new subscription would be
an unparalleled success in its own right, irrespective of the price of the
underlying stock or previous subscription receipts. This view was presum-

75
Scott, Constitution and finance, p. 326.
76
Dickson, Financial revolution in England, p. 127.
© Economic History Society 2005
 14680289, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1468-0289.2005.00304.x by Swinburne University Of, Wiley Online Library on [18/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
260 RICHARD S. DALE, JOHNNIE E. V. JOHNSON, AND LEILEI TANG

ably based on their understanding that the South Sea scheme could only
be perpetuated if successive subscriptions raised increasing receipts, which
could be redistributed as dividends. The increasing premiums of successive
subscriptions over the stock price at issue date were sustained in the sec-
ondary market. This may have resulted from investors being subject to the
anchoring heuristic whereby individuals bias their subsequent judgements
(e.g. of an asset’s value) on an initial estimate irrespective of how spurious
that initial valuation might be, subsequent adjustments being crude,
imprecise and inadequate.77 Consequently, progressively larger premiums
of subscription prices over the stock price at issue date may have gener-
ated, via the anchoring heuristic, expectations of a still larger premium in
the secondary market.
Many of the accounts that argue that the South Sea Bubble resulted from
irrational behaviour point to the spirit of gambling that pervaded early
eighteenth-century England.78 In this regard it is interesting to note that
Blunt himself helped fuel the nation’s thirst for gambling, since he per-
suaded the Chancellor of the Exchequer, Robert Harley, to introduce a
National Lottery in 1711. Blunt organized this lottery, which was a great
success, and he went on to market a second, much larger-scale lottery
known as ‘the Adventure of Two Millions’; this involved an enormous top
prize of £20,000.79 Similarities exist between the manner in which Blunt
structured this lottery and the way he organized the issue of South Sea
subscription receipts. The draw was divided into five classes, each with a
different, and a successively higher, set of prizes; the lowest class prizes
ranged from £110 to £1,000, and the fifth, and final class featured prizes
from £130 to £20,000. The staged prize draws were deliberately designed
to build excitement, investors being unsure ‘whether to be pleased or dis-
appointed until the last ticket of the fifth and final draw was produced from
the box to win the biggest prize of all’.80 Blunt’s lotteries succeeded in
capturing the attention of much of England, and they were successful in
raising £3.5 m.
The successive issues of subscription receipts mirrored the successive
draws of the lottery, which the population had associated with increased
excitement and higher prizes. To a nation whose thirst for excitement and
gambling had been stimulated by Blunt’s successful lotteries, the successive
South Sea subscription issues may have been viewed in a similar manner to
successive lottery draws. The rising premia of the issue prices of successive
subscriptions over the equivalent stock price might also have been suggestive
of the ever-larger prizes featured in successive lottery draws, with their
attendant excitement. The central role of Blunt in both enterprises is likely
to have furthered this view. It has been demonstrated that individuals often

77
Kahneman, Slovic, and Tversky, Heuristics and biases, pp. 14–18.
78
Dickson, Financial revolution in England, p. 156.
79
Balen, A very English deceit, p. 35.
80
Ibid., p. 35.
© Economic History Society 2005
 14680289, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1468-0289.2005.00304.x by Swinburne University Of, Wiley Online Library on [18/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
IRRATIONAL BEHAVIOUR DURING THE SOUTH SEA BUBBLE 261
place too much emphasis on similarities between new stimuli (i.e. the
subscription issues) and events with which they are familiar (i.e. the lottery),
irrationally overlooking differences. This ‘representativeness heuristic’, 81
might well have caused investors to expect the later subscription issues to
produce greater returns than the earlier issues.
It might be argued that even if investors were tempted to perceive the
subscription issues as akin to successive lottery draws, this would not cause
them to misprice the assets since they had available to them comprehensive
and authoritative valuations of the South Sea stock, most notably produced
and publicized by the highly respected Archibald Hutcheson MP. However,
Hutcheson’s calculations were relatively complex for the average investor
(particularly as many were new entrants to capital markets), and research
has demonstrated that individuals increasingly rely on heuristics, non-
rational strategies, and biases when faced by a complex information envi-
ronment; this is especially true for those who are relatively naïve in a
particular decision domain.82 Increased complexity has also been demon-
strated to lead to increases in risk propensity83 and to a deterioration in the
quality of investors’ decisions.84 In addition, as complexity increases, par-
ticipants in speculative markets increasingly allow the market to drive their
behaviour85 and Shiller argues that consideration of past price changes
when making current decisions is the ‘essence of a speculative bubble
. . . feedback from price increases to increased investor enthusiasm, to
increased demand, and hence further price increases’. 86
The degree to which echoes of the lottery featured in South Sea subscrip-
tion investors’ minds is obviously open to question, but it is clear that sales
of financial instruments at this time exploited a growing addiction to gam-
bling. Excitement attended the earlier lottery draws and featured in many
contemporary accounts of trading during the South Sea Bubble. Dickson
suggests that the period was characterized by ‘an appetite for gain . . . [that]
could and did become uncontrolled’.87 This, he argues, even influenced the
judgement of experienced investors. Chancellor points to the speculative
excess that fuelled the markets in 1720,88 and the central role of excitement
in changing behaviour within speculative markets is well established. 89 It
has also been demonstrated that exposure to excitement in such markets
can impose substantial financial costs.90 Previous research has shown that
excitement in wagering markets is more likely to be raised to levels that

81
Kahneman, Slovic, and Tversky, Heuristics and biases, pp. 23–98.
82
Eiser and van der Pligt, Attitudes and decisions, p. 99; Keren and Wagenaar, ‘Psychology of playing
blackjack’, p. 157.
83
Johnson and Bruce, ‘Risk strategy’, pp. 11–14.
84
Bruce and Johnson, ‘Decision-making under risk’, pp. 72–4.
85
Johnson and Bruce, ‘Effect of complexity’, pp. 769–70.
86
Shiller, ‘Bubbles’, p. 3.
87
Dickson, Financial revolution in England, p. 156.
88
Chancellor, A history of financial speculation, pp. 76–8.
89
Brown, ‘Normal and addictive gambling’, p. 165.
90
Bruce and Johnson, ‘Costing excitement’, pp. 57–61.
© Economic History Society 2005
 14680289, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1468-0289.2005.00304.x by Swinburne University Of, Wiley Online Library on [18/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
262 RICHARD S. DALE, JOHNNIE E. V. JOHNSON, AND LEILEI TANG

damage rational judgement where market participants build on each other’s

excitement in a face-to-face arena.91 This is precisely the frenetic, charged
atmosphere that was evident in Exchange Alley, where South Sea stock and
subscription receipts changed hands; attested to by the London attorney of
a Dutch investor, who, on visiting the alley, proclaimed, ‘it is like nothing
so much as if all the lunatics had escaped out of the madhouse at once’. 92
Certainly, the relative market valuations of stock and subscriptions became
so distorted that it is difficult to find a rational justification for the discrep-
ancy. If market activity had been particularly low then this might have
inhibited prices adjusting to reflect the same information across markets.
However, whilst total trading volume data is not available, there are con-
temporary accounts which indicate that market activity was on a consider-
able scale. For example, Hoare and Company’s trading accounts indicate
that Hoare’s own trading in South Sea stock exceeded £140,000 in 1720. 93
In addition, a single investor, Mr Waller, generated a turnover of £795,000
in his South Sea trading account from March to November 1720, 94 and the
Chancellor of the Exchequer, John Aislabie, stated that such activity was
typical of many investors. In addition, contemporary reports suggest that
one-sixth of the third subscription was resold for cash within a few days of
its issue.95 Consequently, the indications of heavy trading volumes for stock
and subscription receipts during 1720 suggest that the prices of these
substitutable financial instruments should have equalized if investors had
behaved rationally.
The results reported here are clearly dependent on the validity of the
assumptions employed, but if these hold then it appears that investors
simply lost the ability to adequately price these assets, or simply developed
overconfidence in their own intuition96 that later subscription issues would
outperform previous issues. In conclusion, while the precise roles played by
excitement and heuristics are difficult to discern, the results suggest that
South Sea investors may have been the victims of their own irrational
behaviour, or may have, in Shiller’s words, suffered from ‘irrational
exuberance’.97

VIII
The article sets out to resolve some of the controversy that has arisen
concerning the nature of investor behaviour during the South Sea Bubble.
Some studies have attributed the Bubble to irrational behaviour, but these

91
Johnson and Bruce, ‘Successful betting strategies’, pp. 643–8.
92
Wilson, Anglo Dutch commerce, p. 122.
93
Temin and Voth, ‘Riding the South Sea Bubble’, p. 11.
94
Cobbett, Parliamentary history, p. 730.
95
The pattern of trading by the original subscribers cannot be meaningfully analysed because of the
widespread practice of using nominee accounts.
96
Shiller, ‘Bubbles’, pp. 4–5.
97
Shiller, Irrational exuberance, p. 18.
© Economic History Society 2005
 14680289, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1468-0289.2005.00304.x by Swinburne University Of, Wiley Online Library on [18/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
IRRATIONAL BEHAVIOUR DURING THE SOUTH SEA BUBBLE 263
accounts have largely relied on contemporary anecdotes and quotations.
More recent contributions to the debate have attempted to explain the rise
in the South Sea share price within a rational expectations framework,
different accounts attributing the South Sea episode to a rational extrinsic
or intrinsic bubble. Garber, for example, concludes that the South Sea
Bubble was ‘understandable as a case of speculators working on the basis
of the best economic analysis available and pushing prices along by their
changing view of market fundamentals’.98
In an attempt to shed light on the behaviour of South Sea investors,
data is employed on subscription receipts issued by the South Sea Com-
pany in 1720 that has not hitherto been used. The results demonstrate the
lack of a long-run relationship between the prices of South Sea stock and
those of three of the four subscriptions, a widening gap between the valu-
ation of the stock and subscription price series for successive subscription
issues, and a weakening of the short-run relationship between stock and
subscription prices for later issues; with a complete breakdown in this
relationship for the fourth subscription. Despite being directly substitut-
able assets, the equivalent share prices for each of the four subscriptions
also appear to diverge significantly from each other. However, these results
depend crucially on certain assumptions, notably that Freke’s stock and
subscription prices were not subject to systemic distortion, and that the
discount rate employed is appropriate. Steps are taken in the article to
defend these assumptions, and the sensitivity analysis that is conducted
suggests that the results are robust to large deviations from the adopted
discount rate.
Some explanations for the results are offered which preserve the notion
of rational investor behaviour, but none of these proves wholly adequate.
Neal defines an irrational bubble as one where the ‘relationship of an asset
to its market fundamentals simply breaks down because of overzealous
trading or an unrealistic appraisal of the value of the stock’.99 The results
suggest that this description more accurately portrays the South Sea episode
than any alternative based on a rational expectations framework. It is also
suggested that the frenetic activity in Exchange Alley and the naïvety of
many of the investors (who were new to capital markets) may have caused
excitement levels to rise to destructive levels (in terms of rationality) and
led to biased judgements, caused by widespread use of a range of heuristics.
Alternative explanations that are offered include the thirst for gambling in
England at this time, and the manner in which the subscription issues were
designed to mimic previous lotteries, each successive issue building levels
of excitement and greed. Certainly, investor behaviour at this time appears
to mirror that observed in recent studies of gambling markets. Interestingly,
the lottery paradigm has been applied to more recent stock market events.

98
Garber, ‘Famous first bubbles’, p. 52.
99
Neal, Rise of financial capitalism, p. 76.
© Economic History Society 2005
 14680289, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1468-0289.2005.00304.x by Swinburne University Of, Wiley Online Library on [18/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
264 RICHARD S. DALE, JOHNNIE E. V. JOHNSON, AND LEILEI TANG

Alan Greenspan, Chairman of the US Federal Reserve Board, has drawn

this comparison between lotteries and the Internet stock boom:
There is something else going on here, though, which is a fascinating thing to
watch. It is, for want of a better, term, the ‘lottery principle’. What lottery
managers have known for centuries is that you could get somebody to pay for a
one-in-a-million shot more than the value of that chance. In other words, people
pay more for a claim on a very big pay-off, and that’s where the profits from
lotteries have always come from. So there is a lottery premium built into the
prices of Internet stocks.100
What then is the significance of the South Sea Bubble for today’s financial
markets? The central lesson that both investors and policy-makers have to
take account of is the fact that today, as in 1720, investor behaviour can
become manic and irrational. The South Sea episode was short-lived, the
boom–bust cycle occurring within a single year, and there was therefore no
great dislocation of the real economy. However, more extended bubbles,
such as those experienced by Japan in the 1980s and by the USA, in
particular, in the late 1990s, can have more enduring consequences, result-
ing in a serious misallocation of resources and severe economic fluctuations.
Policymakers need to be aware, therefore, of the potential damage caused
by irrational investor behaviour and may need to intervene to prevent
financial market excesses. It is hoped that a better appreciation of the factors
contributing to the South Sea Bubble will inform the current debate on this
important issue.

APPENDIX I: The sources of data used in, and the method of

compilation of, table 1
Table 1 displays the schedule of payments and the issue price (shown as ‘total’ in
the table) for each of the four subscriptions, issued on 14 April, 29 April, 16 June,
and 24 August 1720, respectively. These are based on the schedules of payments
published in Freke’s Prices of Stocks etc and Castaing’s The Course of the
Exchange.101
However, there are two complications in using this data. First, according to both
Freke and Castaing, the schedule for the second subscription was changed around
22 July 1720. The instalment payments remained at £40, but the payment dates
were changed. The revised payment dates are shown in table 1,102 with the original
payment dates shown in brackets.103 The revised schedule had the effect of bringing
forward the final instalment date from 24 April 1723 to 14 December 1722. In
subsequent calculations that employ the data displayed in this table, the original

100
Washington post, 21 Jan. 1999, A1.
101
See, for example, Freke’s Prices of Stocks etc., no. 93, 26 August 1720 and Castaing’s The Course of
the exchange, 26 August 1720.
102
Reported in Freke’s Prices of Stocks etc, no. 84, 26 July, 1720 and Castaing’s The Course of the
exchange, 26 July 1720.
103
Reported in Freke’s Prices of Stocks etc, no. 82, 19 July 1720 and Castaing’s The Course of the
Exchange, 19 July 1720.
© Economic History Society 2005
 14680289, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1468-0289.2005.00304.x by Swinburne University Of, Wiley Online Library on [18/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
IRRATIONAL BEHAVIOUR DURING THE SOUTH SEA BUBBLE 265
schedule of payments is used for prices up to 22 July 1720, and the revised schedule
is used thereafter, since investors would have been aware of the schedule revision.
The second difficulty relates to the fourth subscription. Anderson indicates that
when this was issued on 24 August 1720 it was stated to be payable as to
20 per cent down, with four further equal payments at six-monthly intervals over
two years.104 This two-year repayment period was also reported in the contemporary
press.105 However, it seems that the precise dates of the calls were not initially
specified, so that Freke gives only the payment months (February and August) in
his schedule of payments.106 Castaing publishes no schedule for the fourth subscrip-
tion until October, when the schedule had in any case been changed to reflect a
retrospective adjustment of terms proposed by the South Sea Company. Conse-
quently, Freke’s schedule has been employed, but it has been assumed that payment
dates fall on the 24th of each relevant month (i.e. corresponding to the issue date
of 24 August).

APPENDIX II: The sources of data used in, and the method of
compilation of, table 2
The stock price and subscription price series used to determine the statistics
displayed in table 2 are compiled using Freke’s Prices of Stocks etc, which provides
a continuous series of stock and subscription prices from 14 May 1720. The reasons
for employing this data source are fully discussed in section IV.
The stock price series is taken directly from Freke without adjustment, other
than for the period 23 June to 22 August 1720, when the transfer books of the
company were closed.107 As explained more fully in section IV, the prices quoted
by Freke during this period were ‘for the opening of the books’ and were, therefore,
forward prices. The prices during this period are, therefore, converted to a present
value equivalent by discounting using a 5 per cent discount rate (a justification for
employing a 5 per cent discount rate is given in section IV).
In addition, a number of assumptions and adjustments are made in relation to
the subscription prices quoted by Freke in order to convert them into (fully paid)
stock price equivalents. The assumptions and adjustments are fully discussed in
section IV, and in brief, include the following:
(a) The subscription receipts were not delivered out by the South Sea Company
immediately on the issue date. For the reasons outlined in section IV it is
assumed that all four subscriptions were to be delivered out two months after
the issue date.108
(b) Where Freke quotes both regular and ‘for money’ subscription prices,109 the
former are employed, as discussed in section IV.
(c) Where Freke does not distinguish between ‘for money’ and regular transac-
tions, it is assumed that he is referring to regular transactions.

104
Anderson, Origin of commerce, p. 113.
105
For example, The Weekly Journal, published by John Applebee, 27 August 1720.
106
Freke’s Prices of stocks etc, no. 93, 26 August 1720.
107
Dale, The first crash, p. 172.
108
Ibid., p. 174.
109
See, for example, Freke’s Prices of stocks etc, no. 88, 9 August 1720.
© Economic History Society 2005
 14680289, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1468-0289.2005.00304.x by Swinburne University Of, Wiley Online Library on [18/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
266 RICHARD S. DALE, JOHNNIE E. V. JOHNSON, AND LEILEI TANG

(d) During the two-month period when receipts had not been delivered out by
the Company, the prices quoted by Freke are forward prices. These prices
are therefore discounted from the expected delivery date at 5 per cent
(see section IV).
(e) The prices of all four subscriptions are converted into values that can be
compared directly to the stock price by adding the present value of the unpaid
calls back to the market price (adjusted in the manner indicated above).
(f ) For the second subscription, the schedule of payments was changed on 22 July
1720 (as indicated in table 1 and discussed in appendix I). Consequently, the
original schedule is used in calculations of a stock price equivalent to subscrip-
tion two prices until 22 July, and the revised schedule is used after that date.
(g) The first three subscriptions were entitled to the 10 per cent midsummer
dividend of 1720110 and, consequently, the prices of these subscription receipts
(adjusted in the manner indicated above) are equivalent to stock prices. How-
ever, the fourth subscription was not subject to the 10 per cent stock dividend
and, therefore, the prices for subscription four receipts (adjusted in the manner
indicated above) are raised by 10 per cent to put them on a comparable basis
with the first three subscriptions and the stock price.
Following the assumptions and adjustments indicated above the resulting sub-
scription price series are directly comparable with the stock price series, and it is
these adjusted price series which are used to calculate the statistics displayed in
table 2.

APPENDIX III: The sources of data used in, and the method of
compilation of, table 3
Table 3 compares the stock and subscription prices close to the subscription issue
dates. The source of the data and/or the method of calculation for each row of the
table are as follows:
(a) The issue date, new issue price and first subscription payment are taken from
Freke’s prices of stocks etc.111
(b) The stock price equivalent of the issue price of each subscription (SPE of issue
price) is calculated by adding back to the first payment required for each
subscription the unpaid calls on the subscription concerned112 discounted
at 5 per cent, to arrive at a present value equivalent (at issue date) to these
unpaid calls (the justification for using 5 per cent as a discount rate is given in
section IV).
(c) The equivalent stock price of South Sea stock (ESP) at the subscription issue
date is taken from Freke’s prices of stocks etc, and represents the latest stock price
quoted on the day before the subscription date. During the period when the
transfer books were closed (23 June to 22 August 1720) the prices quoted by
Freke are forward prices, and are therefore discounted at 5 per cent to arrive
at a present value equivalent.

110
Dale, The first crash, p. 162.
111
See, for example, Freke’s prices of stocks etc., no. 93, 26 August 1720.
112
Ibid.
© Economic History Society 2005
 14680289, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1468-0289.2005.00304.x by Swinburne University Of, Wiley Online Library on [18/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
IRRATIONAL BEHAVIOUR DURING THE SOUTH SEA BUBBLE 267
(d) The premium of the subscription price over the stock price at the issue date
represents a comparison of the SPE of the subscription issue price and the ESP
at the subscription issue date. It measures the extent to which the subscription
issue price exceeds or falls short of the prevailing stock price.
(e) The market prices for subscriptions three and four, one week after their issue,
are taken from Freke’s prices of stocks etc. However, because Freke did not
publish subscription prices until 14 May (and Castaing later still) Defoe is used
to determine the equivalent prices for subscriptions one and two; he states that
the first two subscriptions rose initially to a premium of 10–12 per cent over
the issue price.113
(f ) The market premium of subscriptions over the first payment compares the
market price for subscriptions one week after their issue date with the initial
down payment required when the subscriptions were issued. For this purpose,
the market prices are taken from Freke’s prices of stocks etc (without adding back
the unpaid calls) and the down payments are taken from Freke’s Prices of Stocks
etc no 93, 26 August 1720. The comparison shows the market premia over
issue price or profit attaching to successive subscriptions shortly after their
issue.
(g) The equivalent stock price of South Sea stock (ESP) one week after the
subscription issue date is taken from Freke’s Prices of stocks etc. The method of
calculation is indicated in (c) above.
(h) The stock price equivalent of subscription receipts (SPE) one week after their
issue is calculated by adding back to the subscription market prices, as reported
by Freke at the relevant date, the discounted value of unpaid calls on the
subscription concerned, using a 5 per cent discount rate.
(i) The premium of the subscription price over the stock price one week after issue
of the relevant subscription compares the values calculated in (h) with the
values calculated in (g) above. The comparison shows the market premium
attaching to successive subscriptions over the stock price shortly after the
subscription issue date.
(j) The subscription prices used in (e) and (h) above involve some adjustment to
the prices quoted by Freke, since subscription receipts were not delivered out
by the company immediately on the issue date. For the reasons outlined in
section IV it is assumed that receipts for all four subscriptions are delivered
out, or are expected by the market to be delivered out, two months after the
issue date. During this two-month period, Freke quotes prices for regular and
‘for money’ transactions for the third and fourth subscriptions.114 The former
are employed in table 3 to avoid distortions caused by settlement risk associated
with the latter type of contract (see section IV for a full discussion of this issue).
Where Freke does not distinguish between ‘for money’ and regular transac-
tions, it is assumed that he is referring to regular transactions. The prices
quoted by Freke during the two-month period when receipts had not been
delivered out by the company are, in effect, forward prices. These prices are
therefore discounted from the expected delivery date at 5 per cent (the
reasons for employing 5 per cent are fully discussed in section IV).

113
Defoe, True state of the contracts, p. 10.
114
See, for example, Freke’s Prices of stocks etc, no. 88, 9 August 1720.
© Economic History Society 2005
 14680289, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1468-0289.2005.00304.x by Swinburne University Of, Wiley Online Library on [18/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
268 RICHARD S. DALE, JOHNNIE E. V. JOHNSON, AND LEILEI TANG

APPENDIX IV: The methodology employed in the

co-integration analysis
The aim is to examine the relationship between stock and subscription prices. If
these time series were stationary then it would be possible to explore this relation-
ship using a simple OLS regression. However, in many cases, time series of this
sort are not stationary and Granger demonstrates that the relationship can then be
examined using co-integration.115 Engle and Granger showed that the co-integrated
variables can be written in the form of an error-correction model (ECM).116 These
provide a means of determining, simultaneously, the nature of any long-run rela-
tionship and the short-run dynamics that exist between the stock and subscription
prices series, even though the absolute values of the series may be non-stationary.
The starting point for the analysis therefore involves an exploration of whether there
exists a co-integration relationship between the stock price St and subscription price
Ctj series. To test for this it must first be decided whether the stock and subscription
prices contain a unit root (i.e. are non-stationary); achieved by applying the Aug-
mented Dickey-Fuller unit root test. If the series St and Ctj are identified as non-
stationary variables, tests for a co-integration relationship are conducted to decide
whether a stationary linear combination of the two series exists of the following
form:
ẑ t = S t - b j Ctj - m j (3)
where m j and b j are estimated constant and parameter values (discussed more fully
below).
Johansen and Juselius’ multivariate co-integration approach is employed to test
for such co-integration.117 However, since the co-integration test is sensitive to the
order of lags included in the vector auto-regression (VAR) a general to specific
approach is employed to decide on lag length before the co-integration test is
conducted. If the two price series are co-integrated and P is of reduced rank, the
following error-correction model representation (ECM) can be applied to examine
the long- and short-run dynamic system.
DZ t = Q1DZ t -1 + . . . + Q k -1DZ t - k +1 + PZ t -1 + e t (4)
where D is the difference operator; thus DZt is a vector of stationary variables of
stock and subscription prices. The term QiDZt-i captures the short-run dynamics.
The matrix P provides information about the long-run relationship between St and
Ctj . The existence of co-integration implies that matrix P can be represented as
abb¢, where a and b are ( pz ¥ r) matrices (r is the number of co-integration
relationships and pz is the number of variables). In this form, the vector b¢Zt-1
captures the deviation of the system from its long-run relationship and et is the
vector of martingale difference sequences.
The long-run relationship between the stock price (St) and subscription price
( Ctj ) series can be examined using matrix P. If P has zero rank, co-integration does
not exist between the stock and subscription prices. If P has full rank, then both
price series are stationary and OLS can be applied to explore the relationship

115
Granger, ‘Econometric model specificating’, p. 127.
116
Engle and Granger, ‘Co-integration and error correction’, pp. 255–9.
117
Johansen and Juselius, ‘Inference on cointegration’, p. 170.
© Economic History Society 2005
 14680289, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1468-0289.2005.00304.x by Swinburne University Of, Wiley Online Library on [18/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
IRRATIONAL BEHAVIOUR DURING THE SOUTH SEA BUBBLE 269
between St and Ctj . If P has reduced rank, then the price series are co-integrated
and P can be expressed as P = ab b¢. Consequently, if P has reduced rank, the form
of the ECM to be estimated will be given by:
k -1 k -1
c
DS t = a S zˆ t -1 + Â q is,s DS t - i + Â q i ,js DCtj- i + e st (5)
i =1 i =1

k -1 k -1
c
DCtj = a c j zˆ t -1 + Â q is,c j DS t - i + Â q i ,jc j DCtj- i + e ct (6)
i =1 i =1

s
where q is,s and q i ,c j measure the influence of previous changes in the stock
cj
price on
c
current stock and subscription ( j) price changes, respectively; q i ,js and q i ,c j measure
the influence of previous changes in the subscription ( j) price on current stock and
subscription price changes, respectively; ẑ t is the error correction term that iden-
tifies the long-run relationship between stock and subscription prices (i.e. ẑ t = St
- b j Ctj - m j).
The coefficients, as and a c of the long run equilibrium error ẑ t, measure,
j

respectively, the adjustment speed of DSt and D Ctj toward the long-run equilibrium.
These speed adjustment coefficients can also help determine the direction of the
long-term causal relationship between the series, since, for example, if as = 0 and
a c π 0, then a deviation from the long-run equilibrium will be corrected by a change
j

in the subscription prices (not in stock prices); this implies that stock prices are
weakly exogenous for the subscription prices, indicating that more information is
contained within the stock price series. In such a case it would be likely that the
stock prices Granger-cause the subscription prices; although it would be possible
for the subscription prices to also Granger-cause the stock prices.
Short-run information flows between the price series are specifically captured by
s
the coefficients q sc j and q c j, which, respectively, determine the Granger causality
effects from stock to subscription ( j ) prices and from subscription ( j ) to stock
prices. Provided there exists a long-run equilibrium relationship between the price
series then there must exist Granger causality effects from stock to subscription
prices (or vice versa) or in both directions. If the price series are co-integrated and
one of the price series Granger-causes the other series then prediction of the latter
series can be improved by using lagged values of the former series and lagged values
of itself.118 If both series Granger-cause each other, both can provide useful infor-
mation about the other.

University of Southampton

First submitted 5 November 2003

Revised version submitted 15 September 2004
Accepted 31 October 2004

118
Granger, ‘Investigating causal relations’, p. 431.

Footnote references
Anderson, A., An historical and chronological deduction of the origin of commerce, vol. 3 (1764).
Avery, C. and Zemsky, P., ‘Multidimensional uncertainty and herd behaviour in financial markets’,
American Economic Review, 88 (1998), pp. 724–48.
© Economic History Society 2005
 14680289, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1468-0289.2005.00304.x by Swinburne University Of, Wiley Online Library on [18/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
270 RICHARD S. DALE, JOHNNIE E. V. JOHNSON, AND LEILEI TANG

Azariades, C., ‘Self-fulfilling prophecies’, Journal of Economic Theory, 25 (1998), pp. 380–96.
Balen, M., A very English deceit (2003).
Brown, R. I. F., ‘Arousal, reversal theory and subjective experience in the explanation of normal and
addictive gambling’, in W. R. Eadington, ed., Proceedings of the seventh international conference on
gambling and risk taking, 3 (Reno, 1988), pp. 165–187.
Bruce, A. C. and Johnson, J. E. V., ‘Costing excitement in leisure betting’, Leisure Studies, 14 (1995),
pp. 48–63.
Bruce, A. C. and Johnson, J. E. V., ‘Decision-making under risk: effect of complexity on performance’,
Psychological Reports, 79 (1996), pp. 67–76.
Campbell, J. Y. and Perron, P., ‘Pitfalls and opportunities: What macroeconomists should know about
unit roots’, NBER Technical Working Paper, 100 (1991), pp. 1–66.
Carlos, A. M., Moyen, N., and Hill, J., ‘Royal African Company share prices during the South Sea
Bubble’, Explorations in Economic History, 39 (2002), pp. 61–87.
Carswell, J., The South Sea Bubble, revised ed. (Dover, 1993).
Chancellor, E., Devil take the hindmost: A history of financial speculation (1999).
Cheung, F. and Lai, K., ‘Finite-sample sizes of Johansen’s likelihood ratio test for cointegration’, Oxford
Bulletin of Economics and Statistics, 55 (1993), pp. 313–29.
Cobbett, W., Parliamentary history of England (1806–20).
Dale, R. S., The first crash: lessons from the South Sea Bubble (Princeton, N.J., 2004).
Defoe, D., A true state of the contracts relating to the third money subscription taken by the South Sea Company
(1721).
Devenow, A. and Welch, I., ‘Rational herding in financial economics’, European Economic Review, 40
(1996), pp. 603–15.
Dickson, P. G. M., The financial revolution in England: a study in the development of public credit, 1688–
1756 (1967).
Eiser, J. R. and van der Pligt, J., Attitudes and decisions (New York, 1988).
Engle, R. F. and Granger, C. W. J., ‘Co-integration and error correction: Representation, estimation
and testing’, Econometrica, 55 (1987), pp. 251–76.
Flood, R. P. and Hodrick, J. R., ‘On testing for speculative bubbles’, Journal of Economic Perspectives,
4 (1990), pp. 85–101.
Froot, K. and Obstfeld, M., ‘Intrinsic bubbles: The case of stock prices’, The American Economic Review,
81 (1991), pp. 1189–214.
Garber, P., ‘Famous first bubbles’, Journal of Economic Perspectives, 4 (1990), pp. 35–54.
Gonzalo, J., ‘Five alternative methods of estimating long-run equilibrium relationships’, Journal of
Econometrics, 60 (1994), pp. 203–33.
Granger, C. W. J., ‘Investigating causal relations by econometric models and cross-spectral methods’,
Econometrica, 37 (1969), pp. 424–39.
Granger, C. W. J., ‘Some properties of time series data and their use in econometric model specificating’,
Journal of Econometrics, 26 (1981), pp. 1045–66.
Granger, C. W. J., ‘Developments in the study of cointegrated economic variables’, Oxford Bulletin of
Economics and Statistics, 48 (1986), pp. 213–28.
Hutcheson, A., Collection of calculations and remarks relating to the South Sea scheme and stock, which have
already been published with an addition of some others, which have not been made public ‘till now (1720).
Johansen, S. and Juselius, K., ‘Maximum likelihood estimation and inference on cointegration with
applications for the demand for money’, Oxford Bulletin of Economics and Statistics, 52 (1990), pp.
169–210.
Johnson, J. E. V. and Bruce, A. C., ‘Successful betting strategies: evidence from the UK offcourse betting
market’, in: W. R. Eadington and J. A. Cornelius, eds., Gambling and commercial gaming: Essays in
business economics, philosophy and science (Reno, 1992), pp. 635–56.
Johnson, J. E. V. and Bruce, A. C., ‘A probit model for estimating the effect of complexity on risk taking’,
Psychological Reports, 80 (1997), pp. 763–72.
Johnson, J. E. V. and Bruce, A. C., ‘Risk strategy under task complexity: a multivariate analysis of
behaviour in a naturalistic setting’, The Journal of Behavioral Decision Making, 11 (1998), pp. 1–18.
Kahneman, D., Slovic, P., and Tversky, A., Judgement under uncertainty: Heuristics and biases (New York,
1982).
Katsaris, A., Periodically collapsing speculative bubbles: An empirical investigation of the SIP 500 Composite
index, 1888–2001 (Unpub. Ph.D. thesis, Univ. of Reading, 2002).
Keren, G. and Wagenaar, W. A., ‘On the psychology of playing blackjack: normative and descriptive
considerations with implications for decision theory’, Journal of Experimental Psychology: General, 114
(1985), pp. 133–58.
Kindleberger, C. P., Manias, panics, and crashes: A history of financial crises (New York, 3rd edn, 1996).
© Economic History Society 2005
 14680289, 2005, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1468-0289.2005.00304.x by Swinburne University Of, Wiley Online Library on [18/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
IRRATIONAL BEHAVIOUR DURING THE SOUTH SEA BUBBLE 271
Lux, T., ‘Herd behaviour, bubbles and crashes’, Economic Journal, 105 (1995), pp. 881–96.
Murphy, A. E., John Law economic theorist and policy maker (Oxford, 1997).
Neal, L., The rise of financial capitalism: International capital markets in the age of reason (Cambridge,
1990).
Ofek, E. and Richardson, M., ‘DotCom mania: The rise and fall of internet stock prices’, Working Paper
(New York Univ., 2002), pp. 1–42.
Parsons, B. T., The behaviour of prices on the London stock market in the early eighteenth century
(unpub. Ph.D. thesis, Univ. of Chicago, 1974).
Perkins, A. B. and Perkins, M. C., The internet bubble: Inside the overvalued world of high-tech stocks and
what you need to know to avoid the coming shakeout (New York, 1999).
Reimers, H. E., ‘Comparison of tests for multivariate cointegration’, Statistical Papers, 33 (1992),
pp. 335–59.
Scott, W., The constitution and finance of English, Scottish and Irish joint stock companies to 1720,
Vols. I–III (Cambridge, 1902).
Shiller, R. J., Irrational exuberance (Princeton, N.J., 2000).
Shiller, R. J., ‘Bubbles, human judgement and expert opinion’, Cowles Foundation Discussion Paper 1303
(Yale Univ., 2001), pp. 1–16.
Temin, P. and Voth, H., ‘Riding the South Sea Bubble’, Discussion Paper 4221 (Centre for Economic
Policy Research, London, 2004), pp. 1–41.
Thaler, R., Advances in behavioral finance (New York, 1993).
Tirole, J., ‘Asset bubbles and overlapping generations’, Econometrica, 53 (1985), pp. 1499–528.
Weiller, K. and Mirowski, P., ‘Rates of interest in 18th century England’, Explorations in Economic History,
27 (1990), pp. 1–28.
Wilson, C., Anglo Dutch commerce in the 18th century (Cambridge, 1941).
Zeira, J., ‘Information overshooting, booms and crashes’, Journal of Monetary Economics, 43 (1999),
pp. 237–57.

© Economic History Society 2005