THE GOLD STANDARD AND INTERNATIONAL SPECIE FLOWS

Between 1870 and 1914 many countries adhered to a gold standard, strictly tying national money supplies to gold stocks and standing ready to redeem currency for gold at a fixed exchange rate. The operation of the gold standard had a number of implications for economic activity in open economies, as summarized in the classical adjustment mechanism. Consider the effects of a sudden outflow of gold from a country caused, for example, by the payment of insurance claims. Through the price-specie-flow mechanism (most famously described by David Hume in 1752), the gold outflow amounts to a reduction in the nation's money supply, and so reduces the domestic price level. As the price of domestic goods falls relative to the price of foreign goods, exports will increase, imports will decline, and the resulting trade surplus will return gold to the nation. This gold flow will reduce the foreign price level and raise the domestic price level until trade imbalances and international price differentials are eliminated.⁵

With integrated financial markets, this trade adjustment mechanism can be augmented by capital flows, whereby the reduction in the money supply from a gold outflow raises domestic interest rates. This will attract loanable funds from abroad, reversing the initial gold outflow. Central banks may also use policy tools to affect the speed with which equilibrium is reached. Under what came to be known as “the rules of the game,” central banks were supposed to respond to a sudden gold outflow by contracting the domestic money supply and raising interest rates.⁶

These policies would enhance the trade surplus and capital inflow effects of the classical adjustment mechanism, accelerating a return to equilibrium.

Prior to 1913 the United States lacked a central bank to provide liquidity during periods of monetary stringency and so the economic effects of abnormal gold flows were often magnified. Fabio Canova, and Ellis Tallman and Jon Moen, building on work by O. M. Sprague in 1910, argue that unusual gold flows were the cause of most financial crises and business cycles in the United States before the founding of the Federal Reserve.⁷

Canova shows that external gold shocks were an important factor in explaining fluctuations in high-powered money and economic downturns in the pre-Federal Reserve era. Tallman and Moen find evidence of a monetary transmission mechanism where shocks to gold high-powered money had non-neutral effects on interest rates and output in the face of sluggish price adjustment.

There is, however, far less research on the sources of exogenous gold shocks. Take, for example, the many studies that have examined the causes of the Panic of 1907. Most of this literature explains the Panic as the ultimate outcome of a liquidity crisis in London in the wake of an unusually large amount ($70 million) of gold exports from England to the United States during 1906. Under the gold standard, such a sudden gold outflow represented a massive shock to the British money stock. In response, the Bank of England not only nearly doubled its discount rate between September and October 1906, but it also instituted a discriminatory policy against capital flows (in the form of finance bills) to the United States. These actions practically cut off gold exports to the United States, and England reversed its position from a net gold exporter to a net gold importer. The actions pushed the United States into a recession and made New York markets susceptible to a financial panic.

Scholars disagree about what caused the gold outflows that ultimately led the Bank of England to take such drastic measures. Charles Kindleberger cites various international events ranging from an Italian banking failure to the Boer War and the Russo-Japanese War as possible sources of the British gold outflow; R. S. Sayers adds the fact of a bumper crop of cotton produced in Egypt in autumn 1906.⁸

Although each of these events may have resulted in some gold outflows from Britain, it remains the case that during this period, British gold exports to the United States were nearly 230 percent greater than to any other country in 1906, as can be seen in Table 1. If we add nearly three million pounds sterling of gold exports returned to Paris (as repayment of specie lent to London in the spring to cover earthquake payments), then British gold exports to the United States in 1906 were more than 250 percent greater than to any other country.

What events particular to the United States could have precipitated such unusual imports of specie from England? Milton Friedman and Anna Schwartz, and Charles Goodhart note that Secretary of the Treasury L. M. Shaw had, throughout 1906, subsidized gold imports, but they discount the impact these actions may have had in changing the usual direction of specie movements.⁹

In the end, most authors blame the gold flows on excessive speculation in the New York equity market in 1906.¹⁰ There is, however, little hard evidence on the magnitude or timing of such flows; there is even some question about the very likelihood of the flows. Excess speculation implies a return differential in favor of the United States. To investigate this question, Table 2 reports simple summary statistics of U.S. and U.K. stock returns as measured by the S&P 500 and the U.K. All Shares Index in the year prior to the earthquake, April 1905—March 1906. Although the average U.S. monthly stock return was 0.90 percent versus 0.12 percent for the United Kingdom during this period, there were many months during this period when returns on British equities outpaced American equities, as shown by the range and standard deviation of returns.

As an additional check, we test whether U.S. equity returns were significantly greater than U.K. returns in the year and six months before the earthquake. We first regress American stock returns on a constant and on the return on the U.K. All Shares Index for the period 1900–1909. The return on the U.K. index is significant at the 10-percent level. We then estimate two more regressions that incorporate dummy variables to test if there were abnormal returns in U.S. equities prior to the earthquake. The dummy variable, PREQUAKE, is set equal to one in the year leading up to the earthquake in the first specification, and set equal to one in the six months before the natural disaster in the second model. As Table 2 shows, U.K. stock returns continue to be significant at the 10-percent level while the PREQUAKE dummy is not significant at even the 40-percent level in both regressions. The results suggest that it is difficult to support the claim that British speculation in an American stock boom created the sizeable gold flows that forced the Bank of England to nearly double its discount rate in September and October 1906.

In contrast, our evidence shows that the magnitude and timing of earthquake-loss-related funds do fit the pattern of gold flows we see in autumn 1906. British insurers (and others, including a significant number of German firms) stalled on paying claims immediately after the April disaster. It was not until late summer 1906 that an agreement was reached among the firms to accept liability for claims. In their comprehensive monetary history of the United States, Friedman and Schwartz offer no definitive statement on the basis for the Bank's rate increases, saying only that “During September and October 1906, when the New York money market was relieved by gold imports from Great Britain and Germany, the London money market was in turmoil.”¹¹

A close reading of their statement is instructive: the authors note that New York was relieved by gold flows from Great Britain and Germany; our evidence shows that the two largest sources of foreign insurance underwriters in San Francisco at the time of the earthquake and fire were England and Germany.¹²

Contemporary observers certainly noted this connection, as seen in the epigraph. But while Sayers and Kindleberger make brief reference to the observation that the San Francisco earthquake resulted in gold exports from England to the United States (echoing a suggestion first made by a banker in 1908 and repeated by Sprague in his 1910 volume on bank crises), they appear to be referring mainly to the gold flows that occurred immediately after the earthquake in Spring 1906.¹³

In contrast, our article utilizes data from contemporary newspapers as well as internal documents from British insurance companies to show that insurance payments in the aftermath of the San Francisco earthquake were at the heart of the unusual gold flows throughout 1906. The payment of claims by British insurance companies to policyholders in San Francisco (delayed until late summer by the companies' initial dispute of their liability, then further postponed by an extension of the period in which to make claims—and the typical insurance clause allowing 60 days to pay claims once they were adjusted) caused the massive gold exports that ultimately resulted in defensive actions by the Bank of England, setting off the chain of events that culminated in the Panic of 1907.

THE SETTING AND THE DISASTER

By the time the transcontinental railroad was completed in 1869, San Francisco had already established itself as the center for export trade from the Pacific Coast region. Endowed with an excellent natural harbor and easy coastal and river access to the agricultural and natural resource riches of the west, San Francisco had developed strong economic ties to other countries, particularly to Britain. Most of the wheat exported from the west coast and bound for England was financed through San Francisco, and a sizeable number of London banks had offices in that city.

At the same time, other British financial institutions sought to expand their business in the area.¹⁴

Prominent among these were the British insurance companies. Prior to 1907 California law allowed insurers to underwrite both marine insurance and fire insurance. Firms insuring the extensive ocean trade between San Francisco and England, then, had the opportunity (and perhaps the incentive, from the standpoint of diversification) to expand into fire insurance. In 1852 the Liverpool & London & Globe fire insurance company placed an agent in San Francisco: this was the first such fire insurance firm (either foreign or domestic) in the city. Two years later, three more British firms were writing business in San Francisco and the first American insurance firm set up an office in the city. It was not until 1858, however, that a San Francisco-based company was established.¹⁵ By 1890 in California there were 127 Class 1 term American fire insurance firms underwriting a total of $1.72 billion in risks. On the other hand, there were 52 foreign firms that underwrote a total of $1.22 billion in risks; more than 6 percent of these policies were insured by German firms, and nearly 27 percent of the policies were carried by British companies. In fact, the fire insurance company writing the most policies in California was Liverpool & London & Globe, with total risks of $173 million. In comparison, 28 percent of all fire insurance policy risks in New York were underwritten by foreign firms, whereas in Illinois foreign companies insured less than 20 percent of the value of all risks¹⁶

Twenty-five years later, these patterns persisted. At the end of 1905 slightly more than half of insured risks in California were underwritten by American firms, with almost 40 percent of business still carried by foreign firms, most of whom were based in Britain. On the other hand, California-based firms were writing only 7 percent of fire insurance business in the state.¹⁷

The city of San Francisco was even more dependent on foreign fire insurers than the state as a whole. By the turn of the century, it was estimated that at least half of all fire insurance policies in San Francisco were issued by British companies.¹⁸ One explanation for the dominance of British firms is the long history of trade relations between the city and Britain; another is simple economics: as agents from the London and Lancashire insurance firm noted, the profit on San Francisco business equaled 30 percent—“three times greater than that yielded by its business generally.”¹⁹ Evidently, adjusters failed to consider earthquake risk.²⁰

On Wednesday, 18 April 1906 an earthquake of Richter magnitude 8.3 hit San Francisco. Most of the damage was not done by the tremor itself (which was especially severe in areas of landfill where liquefaction occurred) but by the fires that followed. Many of the city's buildings had been made of wood; this material was far more plentiful and inexpensive than brick as a result of the city's central place in the coastal lumber trade. The combination of close quarters, highly flammable building materials, and earthquake-damaged water mains hampered the efforts of firefighters. Ultimately more than four square miles—about half of the city—were destroyed. Although fewer than 1,000 of the city's 375,000 residents were killed, damage to the city was estimated at between $350 million and $500 million, on which approximately $235 million in insurance policies were held.²¹

Word of the disaster in San Francisco spread throughout the United States within hours and the impact was felt almost immediately in financial markets. In particular, news of the earthquake led to a sell-off and a significant drop in the price of shares in the New York and London stock exchanges. On 26 April The New York Times reported that the San Francisco disaster led directly or indirectly to about a $1 billion (or nearly 12.5 percent) decline in the market value of NYSE stocks; railway stocks alone fell more than 15 percent.²²

In London, share prices for leading insurance companies plummeted following news of the San Francisco disaster, with stock prices for most insurance companies involved in the quake suffering losses of 15 to 30 percent in the two weeks following the disaster. Equity prices for London and Lancashire, one of the largest British insurers in San Francisco, posted a 30-percent decline. At the same time, shares of London Assurance fell more than one-third, from 75 to 51.5.²³

Funds for relief and rebuilding flowed into the city quite quickly. Given the preference of Californians for specie over banknotes, these funds came largely in the form of gold; in fact, San Francisco's bankers formed a clearinghouse to coordinate the distribution of gold money to their depositors shortly after the earthquake.²⁴

Figure 1 shows that, in late April and May of 1906, nearly $50 million of gold poured into the United States from Germany, France, the Netherlands, and England (whose contribution alone amounted to $30 million). The New York Times and the records of the National Monetary Commission indicated that 80 percent of these funds were transferred to San Francisco; much of the rest was used to replenish the gold reserves of New York banks depleted by specie shipments to the West Coast.²⁵ These flows were apparently also aided by the U.S. Treasury's policy of subsidizing gold imports by offering to place government deposits in banks while gold was in transit.

THE INSURERS

Because San Francisco was a major market for British fire insurance companies, much of the brunt of financing San Francisco's recovery was borne abroad.²⁶

As recreated in Table 3, The Economist reported at length on the consequences of the disaster for British insurance firms, pointing out that they had more than $87 million in policies in San Francisco, with $46 million in losses.²⁷

Of course, the amount of policies underwritten could not precisely predict the size of insurance payments to be made; some properties survived the earthquake undamaged. At the same time, the insurers (in a particularly bad public relations move) indicated early on that no payments would be made on damage that resulted from the earthquake itself. This was a sticky point: If the fires that followed the earthquake produced the damage but the fires themselves were caused by the earthquake, insurers claimed that they were not liable.²⁸

There were no clear means of allocating damage to the earthquake or to the fire, although some insurers proposed a 60–40 split: 40 percent of each policy claim would be denied “on the ground that the destroyed buildings were first damaged in that proportion by the temblor.”²⁹

In contrast to relief payments, most insurance claims took months to settle.³⁰

Not only did some companies equivocate regarding their liability, but the adjustment of claims was also made extremely difficult by the loss of business records in the fires which had been centered in the city's financial district. Most firms waited for guidance from a report issued under the auspices of the Insurance Department of New York State that outlined how firms should settle claims. This four-point plan was not finalized until the end of July. At that time, almost all insurers affected by the disaster—including the British fire insurance houses—signed on and agreed to settle their claims in accordance with the New York agency.³¹ By October it was estimated that more than $100 million in insurance checks had been received in San Francisco.³² Ultimately, British insurers paid out £10 million (or $48 million) for earthquake damage.³³

What made these British liabilities all the more significant for international financial markets was the fact that most foreign insurers decided to pay the claims out of their “home funds” rather than out of the reserves they held in the United States.³⁴

This, in itself, guaranteed that the earthquake would result in massive funds flows from Britain to San Francisco. At the same time, however, the British companies were hesitant to liquidate the home securities in which they held reserves. Given the size of losses involved, such sales would undoubtedly depress stock prices, as The Economist noted on 5 May 1906. As a consequence, a number of firms negotiated term loans with their banks and so postponed securities sales for a few months.³⁵

The pattern and timing of insurance payments can be seen in the experience of one British company, the Norwich Union Fire Insurance Society, which had over $15 million in San Francisco policies in effect.³⁶

Two days after the earthquake, British firehouses gathered in London to discuss their liability in the disaster. The great majority of British insurance companies ultimately assumed liability for fire damage caused indirectly by the earthquake. Members of the Norwich Union, however, had a special clause protecting them against losses caused by or through an earthquake. These firms initially denied any liability in the San Francisco disaster.

Because the refusal of some British firehouses to assume liability produced a large public outcry in San Francisco, members of the Norwich Union reconsidered their liability in the earthquake. Legal counsel in the United States advised these British firehouses to pay claims in order to avoid possible litigation. Lawyers for the firehouses noted that British companies were likely to lose in court given that a San Francisco jury would be influenced by the “extraordinary conditions” surrounding the quake.³⁷

Persuaded by these arguments, directors for the Norwich Union accepted liability for damage caused by the earthquake.

Some British firehouses met early claims by selling assets held by New York trust companies. Norwich Union, for example, liquidated its portfolio in July 1906.³⁸

In the following months, the company utilized financial bills of exchange—and even borrowed from London banks—to settle insurance liabilities. Cablegrams from Mr. Hare of Norwich Union, collected in the CGNU Archives, indicate that “earthquake” gold was placed on British ships and transferred to the United States:

While the summer of 1906 saw British insurance firms shipping gold to San Francisco, New York financial institutions also faced reduced gold reserves resulting from their own specie transfers to San Francisco in the immediate aftermath of the earthquake.³⁹

New York firms were putting pressure on the Secretary of the Treasury Shaw to resume his policy of subsidizing gold imports by placing deposits in banks while gold was in transit. In this manner, The Financial Times predicted that New York firms would import $20 to $35 million in gold from London in the late summer and early fall of 1906 to shore up reserves lost as a result of the earthquake and fires.

In sum, then, the disaster gave rise to a massive flow of gold to San Francisco and New York, immediately after the earthquake and again in the autumn of 1906. As shown in Figure 2, England exported $30 million in gold to the United States in April and May; this was followed in September and October by a further $35 million in specie. In total, quake-related payments to the United States represented 40 percent of seasonally adjusted British gold exports for all of 1906⁴⁰

and over 80 percent of seasonally adjusted gold imports into the United States that year.⁴¹ As Table 1 shows, these flows were nearly two-and-half times more than British gold flows to any other country in 1906.

Confirmation of the importance of insurance payments comes from figures on gold imports into the port of San Francisco. Typically, the city accounted for a negligible amount of American gold imports in any given month. This was not the case, however, in autumn 1906. As shown in Figure 3, the city of San Francisco alone received through its port approximately $11 million in the late summer and early fall of 1906. This accounts for approximately 9 percent of all seasonally adjusted U.S. gold imports in 1906. In addition, the records of the National Monetary Commission indicate that New York transferred $10 million of gold to San Francisco in the fall of 1906. One periodical noted that “with money pouring in at the rate of $1,000,000 a day from insurance alone, the banks of San Francisco are more crowded with deposits than ever in their history.”⁴²

The New York Times noted on several occasions that these flows imposed a substantial drain on New York banks.⁴³ In turn, New York financial institutions imported gold from Europe to shore up bank reserves.

THE CENTRAL BANKS

The magnitude of the British funds flows can be put into stark perspective by noting that they represented a 14-percent loss in the gold money stock of England.⁴⁴

Faced with its largest two-month net gold outflow in the period 1900–1913—as shown in Figure 4—and the lowest ratio of reserves to deposits since the 1893 crisis, the Bank of England began raising its discount rate.⁴⁵ On 12 September the rate was raised from 3 1/2 to 4 percent; at the time The Economist noted that “It is seldom, however, that the raising of the rate by only 1/2 per cent. suffices to check an outflow of gold, and it remains to be seen whether this is going to prove an exception to the rule.”⁴⁶ In fact, it was not an exception: by 19 October the Bank had raised the discount rate to 6 percent, a level not seen since 1899.⁴⁷ Market watchers anticipated even greater measures to stem the outflow of gold and urged British bankers “to refrain, as far as possible, from renewing the American finance bills….”⁴⁸

As shown in Fig. 5, the bank held the discount rate constant for the remainder of 1906 and then subsequently lowered it in early 1907. This would seem to indicate an easing of credit conditions in England. On the contrary, the bank threatened joint stock companies with a 7-percent rate on money if they did not stop discounting U.S. finance bills, credit instruments used to import gold and to profit from exchange rate fluctuations.⁴⁹

By December the Bank of England had stopped discounting U.S. bills altogether, but not the bills of other countries. The Bank of England adhered to this policy until the panic of 1907.⁵⁰ As shown in Figure 6 these policies reversed England's position from a net gold exporter to a net gold importer by the end of 1906.

The gold drain was not confined to the London money market. Insurance companies from other parts of Europe, particularly France and Germany, had also issued policies in San Francisco, accounting for about 10 percent of the payments owed by foreign firms. Like their British counterparts, these firms paid their claims in accordance with the New York report (see The Times reports for foreign insurance companies, September-December 1906). As seen in Figure 5, the German central bank raised its discount rate to 7 percent in the fall of 1906 to counter gold losses of nearly 10 million pounds sterling, or more than 10 percent of its specie reserve, between 7 August and 31 October 1906. With a bank rate higher than that of the Bank of England—6 percent—gold poured into the Reichsbank in early 1907. The specie reserve of the Reichsbank increased from 36 million pounds sterling on 7 January 1907 to nearly 45 million pounds by the end of February.⁵¹

In addition, the German and French central banks acted in concert with the Bank of England and discriminated against American finance bills for the next year disrupting the usually well-functioning price-specie-flow mechanism of the gold standard.⁵² Sprague considered this action “the most important financial factor in the panic of 1907.”⁵³ Ira Cross estimated that by “violating the rules of the game,” European central banks had reduced the amount of finance bills in the London market from about $400 million to $30 million by the summer of 1907.⁵⁴ Reductions of similar magnitudes were also apparent on the Paris and Berlin markets. These actions stopped the flow of gold to the United States and put the squeeze on American financial markets.

As finance bills came due (without the possibility for the usual rolling-over of extant bills) there was a scramble for liquidity and a sell-off of railroad securities by American firms in early 1907. The decline in stock values led to a short but sharp “Rich Man's Panic” in March, helping to push the United States into a recession. The repatriation of finance bills led to the net export of more than $30 million in gold during the summer of 1907.⁵⁵

As shown in Figures 7 and 8, the U.S. gold stock contracted nearly 10 percent between May and August of 1907 and American industrial production fell 30 percent in the second half of the year. Nathan Balke and Robert Gordon estimate that real GNP, a less volatile measure of macroeconomic activity, declined 6.7 percent in the last half of 1907.⁵⁶

The New York money market entered the fall of 1907 low on gold reserves and vulnerable to shocks that might otherwise have been temporary in nature. Some scholars attribute the proximate cause of the Panic to the failed attempt by Auguste Heinz and his associates to corner the copper market. Others blame the collapse of New York's second largest trust company, Knickerbocker. Both events probably contributed to the panic that erupted in late October 1907, leading to a series of bank runs in New York. Within a few weeks, the crisis spread to other regions of the United States.

Several measures were taken to contain the crisis. The New York Clearinghouse Association issued loan certificates, a money substitute used to clear accounts between banks. Clearinghouse loan certificates artificially increased the money supply and freed up currency for depositors who demanded cash. Federal aid came in the form of public funds deposited by the U.S. Treasury at key New York City banks, and J. P. Morgan formed a money pool with bankers to provide liquidity assistance to trust companies and the stock market.⁵⁷

These measures eased conditions in the money market, but failed to prevent the suspension of specie payments.⁵⁸

As short-term interest rates rose to over 10 percent as seen in Figure 9, gold poured into the United States from England and the rest of Europe. Figure 1 shows that the United States imported over $100 million in gold during November and December. Although the specie arrived too late to prevent a panic, gold shipments probably shortened the period of suspension and reduced the duration of the recession. On the other hand, specie exports drained European money centers of gold, helping to transmit the “localized” New York panic to international financial markets.⁵⁹

By one estimate, the resulting shock to the British money supply depressed output 0.9 percent below its trend level between September 1906 and March 1907.⁶⁰

To summarize, the historical narrative provides evidence that the San Francisco earthquake was the catalyst for a liquidity crisis in the New York and European money markets through its effect on gold flows. The Bank of England, in conjunction with other European central banks, responded by raising their discount rates and discriminating against U.S. bills for the next year. These actions set off a chain of events—the interest rate increases that led to a recession and stock market decline—that ultimately ended in the panic of 1907.⁶¹