Top Banner

Click here to load reader

37

Familiarity, Convenience, and Commodity Money: Spanish ......Familiarity, Convenience, and Commodity Money: Spanish and Mexican Silver Dollars in Qing and Republican China Warren Bailey

Sep 14, 2020

ReportDownload

Documents

others

  • Familiarity, Convenience, and Commodity Money:

    Spanish and Mexican Silver Dollars in Qing and Republican China

    Warren Bailey and Bin Zhao *

    22nd June 2009

    Abstract

    Using the available monthly and annual data from 1866 to 1928, we study the premiums at

    which foreign silver coins and silver-backed currency circulating in China sold relative to melt

    value, at both Shanghai and San Francisco. Silver-backed Chinese currency typically sold at a

    slight premium to its melt value, while Mexican and particularly Spanish silver dollars often sold

    at substantial premiums above melt value, averaging 2.25% and 16.09% respectively. Silver

    money premiums appear related to several measures of Chinese and global trade,

    competitiveness, and prosperity, including the extent of opium imports. We also offer potential

    psychological and cultural explanations for the observed premiums.

    JEL Classifications:

    Keywords: China, silver, silver dollar, Spanish piece of eight, international trade

    * Johnson Graduate School of Management, Cornell University, Sage Hall, Ithaca, NY 14853-6201, wbb1@cornell.edu; School of Business, University of Wisconsin, 5284 Grainger Hall, 975 University Avenue, Madison, WI 53706, bzhao2@wisc.edu. We thank Richard Burdekin, Hazem Daouk, Koralai Kirabaeva, David Ng, Allison Truitt, and Xiaoyang Zhang for discussions, helpful comments, and other assistance. This paper is based on Zhao (2006), the second author’s master’s thesis completed at the Department of Applied Economics and Management, Cornell University. © 2009 Warren Bailey and Bin Zhao

  • 1

    1. Introduction

    International trade between China and Western countries started hundreds of years ago.1

    The ban on maritime trade during the late Ming Dynasty (1659-1683) was eventually reversed,

    and the volume of such trade grew enormously by the nineteenth century. There was great

    demand for Chinese porcelain, brocades, sugar, ginger, silk, tea, and furs, but there was little

    demand for western goods in China. At the same time, there was no standard Chinese currency

    and commerce had to make due with a broad variety of bullion, metal coins, banknotes, and

    other forms of money. Thus, large numbers of foreign silver coins were brought to China to

    facilitate trade.2 This form of money first appeared in China during the Ming dynasty (Peng,

    1994). It was estimated that about 2.2 billion Chinese and foreign silver coins were circulating in

    China at the end of 1930 (Leavens, 1939). Aside from the increase in foreign trade, silver had

    acquired importance in China since taxes were typically denominated in the metal (Lin, 2006).

    Perhaps because of the huge supply of silver from Spanish colonies in the New World

    that flowed into global circulation in coin form, the Spanish silver dollar became the de facto

    standard money for trade with China, and was in common use in both domestic and international

    commerce throughout Asia. The Spanish silver dollar, also known as the “piece of eight”, and its

    Mexican descendant were so significant that, at times, the value of such coins greatly exceeded

    the melt value of the bullion they contained. There were many other dollar-sized silver coins in

    circulation including those of Hong Kong, Japan, the U.K., the U.S., and France struck

    specifically for Chinese and colonial trade.3 China’s central and provincial governments also

    minted silver coins at points, while small silver bars in a variety of shapes were produced by

    silver shops and silversmiths. Figure 1 depicts some of the coins that circulated in the nineteenth

    and early twentieth centuries that are the focus of this study.

    Because minting standards varied from place to place and given the constant threat of

    counterfeit coins, much effort went into assessing silver coins and bullion. Chinese merchants

    were typically suspicious of all but the standard silver coins. Foreigners seeking to buy Chinese

    products often purchased Spanish silver coins at a substantial premium above their bullion value

    1 Much silver came to China by the galleon trade in which silver from the New World was traded for Chinese products at Manila. See Von Glahn (1996) for a description. Another source was arbitrage based on differing relative prices of gold and silver in Europe versus Asia (Von Glahn, 1996a, 1996b; Lin, 2006). 2 Opium was also brought to China to balance the gap between China’s exports and imports with the West. 3 The use of foreign silver coins from more developed economies was not unique to Asia. See, for example, Gervais (1982).

  • 2

    because of their credibility. However, premiums for these coins varied from port to port

    (Greenfield and Rockoff, 1995) and at first they were accepted only by weight. With the passage

    of time and growth in trade, Chinese merchants found the Spanish dollar increasingly convenient

    given their large supply, uniform standard, and clearly marked year and mint. The opening of

    new ports and the end of the monopoly of the East India Company in 1834 both contributed to

    the rising demand for Spanish silver coins. They were collected worldwide to be used for trade

    with China (Willem, 1959). After Mexico gained independence in 1821, it started minting its

    own silver coins, and stopped minting Spanish silver coins. Consequently, there was a shortage

    in the supply of Spanish silver coins at that time (Peng, 1994). The first recorded appearance of

    the Spanish dollar’s Mexican descendent in China was in 1854 (Sandrock, 1995).

    By Gresham’s law, large denominations are expected to carry a premium while small

    denominations should trade at a discount (Rolnick and Weber, 1986). The Spanish silver dollar

    in China may be a contradiction, in that it represented a relatively modest denomination in terms

    of wholesale trade which nonetheless circulated at a premium.4 Even in the nineteenth century

    United States, the Spanish silver dollar circulated at a premium of 0.25 percent to 1 percent

    (Rolnick and Weber, 1986). Not all the foreign silver coins circulating in China were traded at a

    premium. Many Chinese were unfamiliar with the U.S. trade dollar, first issued in 1873

    specifically to facilitate American trade with Asia. Although it contained slightly more pure

    silver than Spanish and Mexican dollars, it initially traded at a discount (Greenfield and Rockoff,

    1995).

    The aim of this study is to understand the drivers of the premiums for Spanish and

    Mexican silver coins, an interesting feature of China’s economy in the late Qing dynasty and its

    Republican aftermath. Foreign silver coins evidently served China well. Understanding their

    premium can help illustrate the relationship between the form of currency and an economy.

    While silver dollars were not “consumable” in the sense of commodity money based on food or

    cigarettes (Burdett, Trejos, and Wright, 2001), they offer an interesting venue to test theories of

    what drove the demand for such money, examine the valuation impact of perceived differences

    in quality and desirability across different types of commodity money, and also allow us to

    4 For ordinary Chinese workers, however, one silver dollar would have been a substantial sum. Peng (1994) describes typical wages in late Qing Shanghai as 25 cents per day for a textile worker, 40 cents a day for plasterers and carpenters, and 85 cents a day for highly-skilled shipbuilders.

  • 3

    speculate about other factors that affected the decision-making processes of the individuals cand

    businesses that used these forms of money in daily life.

    2. A first look at the premiums for silver coins and silver-backed currency

    During the period that we study, the basic local currency unit in Shanghai was the tael, a

    hypothetical amount of pure silver (1.0813 English ounces) that circulated in the form of ingots

    and demand deposits, though there were few tael-denominated bank notes and no coins.5 In an

    attempt to simplify banking and commerce, the Shanghai foreign banking community defined

    the Shanghai tael in January 1857 and also switched silver dollar accounts from Spanish dollar to

    Mexican dollar terms. Monetary confusion continued to reign in China, as different cites and

    government departments defined their own unique tael. Even within the modern banking sector

    in Shanghai, the exchange rates among hypothetical taels, taels in silver ingot form, and silver

    dollar coins varied over time, and the varying quality of physical ingots and coins yielded further

    uncertainty and expense to businesses and consumers.

    In the discussions that follow throughout the paper, we distinguish between what we call

    the melt premium and the coin premium. By melt premium, we mean the market price of a coin

    or currency in excess of its value if converted to silver bullion and sold at the London silver

    price. By coin premium, we mean the difference between the price of a coin expressed in

    Shanghai taels and the value of that coin’s silver content computed as the London melt value

    times the Shanghai taels per U.K. pound exchange rate. As we shall see, these two measures can

    differ significantly. For example, both silver coins and silver-backed paper taels may display a

    melt premium at times when demand for China’s products is high. At the same time, the coin

    premium may be positive if there is a specific demand for silver in recognized coin form. The

    coin premium may be negative if silver-backed taels are perceived as more convenient and less

    costly to handle.

    Figure 2 includes scans of typical newspaper columns from the North China Daily News,

    a British publication produced in Shanghai which is the source of much of our data. The

    relatively short column for 4th May 1877 shows a variety of exchange rates and other data, much

    of which we explain in our discussion of the 5th July 1888 column below. Of immediate interest

    5 In contrast, silver dollars, fractional dollar silver coins, and dollar-denominated bank notes of various issuers were common.

  • 4

    are the quotes for coins, “Mex. Dollars – 76.25” and “Carolus – 83.00”. “Carolus” refers to the

    old Spanish silver dollar with portrait of King Charles. The quotes indicate the cost in Shanghai

    taels of 100 coins. Given that the weight of pure silver in a Spanish or Mexican dollar was just

    about 73% of that of a paper tael,6 there is clearly a substantial coin premium, particularly for the

    older Carolus dollar.

    The column for 5th July 1888 is included because it is different and longer than the earlier

    column. There are a number of spot, discount, and forward exchange rates, prices in local

    currency for Mexican dollars and precious metal bars, and exchange rates and silver prices

    received from London by telegraph.7 Following are descriptions of some of the more important

    items in the column.

    “Tel. Transfers” is the price, in British shillings and pence, for buying or selling

    Shanghai taels via telegraph to and from London. The quotation reflects the British monetary

    system of the time, in which one pound equals 20 shillings and one shilling equals 12 pence.

    “Sight” is the price, in British shillings and pence, for a bank draft for Shanghai taels that can be

    cashed immediately. Other rows in the first part of the newspaper column refer to spot and

    forward rates for drafts or transfers to and from other cities. “Mexican dollars” is the price, in

    Shanghai taels, of 100 Mexican silver dollars. For example, the quotation, “73.2.675”, means

    that one Mexican dollar traded in a range of 0.732 to 0.73625 taels.8 No local price for the

    “Carolus” Spanish silver dollar appears at this time. “Copper Cash” is the number of copper cash

    (a small copper coin with a square hole in the middle) per Shanghai tael. Therefore, the price of

    one copper cash is the inverse of this number. The small copper coins were another form of

    commodity money, though their use was confined to very small transactions and was uncommon

    among foreigners.9

    6 To compute the amount of pure silver in a Shanghai tael, multiply its weight, 565.65 grains, times its fineness, 0.935374, times the “Shanghai convention” for converting paper taels into bullion, 0.98, to yield 518.512 grains of pure silver. To compute the amount of pure silver in a Spanish or Mexican dollar, multiply its weight, 417.8 grains, by its fineness, 65/72, to yield 377.14 grains. This implies that the dollars contain 72.7351% of the pure silver of a paper Shanghai tael. Circulated dollars would likely weigh 2 or 3 grains less, and would trade accordingly. See Kann (1927) and Leavens (1939), page 93. 7 A direct telegraph connection between London and Shanghai was established in 1871. Such connections are known to have increased the geographic integration and efficiency of financial markets (Garbade and Silber, 1978; Michie, 1987). 8 “Bar Silver” is the price, in taels, for a standard bar of silver. Though the tael was also a fixed amount of silver and the bar price does not often fluctuate appreciably, it represents yet another form of commodity money that sometimes varied from its theoretical value. 9 See Chen (1975) for a description of China’s traditional silver-copper bimetallic monetary system.

  • 5

    The newspaper column also includes information from London, and a few of these series

    are critical to our work. “Bar silver” at “42 1/16 d” refers to the global benchmark price in

    British pence for one ounce of 92.5% pure sterling silver in ingot form. There is also a price, 41

    1/2 British pence, for Mexican dollars, though this series did not appear consistently in the

    newspaper. Given the relative weights of silver, the Mexican dollar also traded at London at a

    significant melt premium, which is consistent with historical accounts (Willem, 1959).10

    Figure 2 also includes a scan of a typical exchange rate column from the San Francisco

    Chronicle. At the time this column was published in 1877, two currencies, paper dollars and

    gold dollars, were in use in the U.S., a legacy of both the Civil War “greenback” period and the

    lingering uncertainty about the course of the nominally bimetallic currency standard. “Legal

    Tenders” at bid 94 3/4 and ask 95 1/2 indicate that paper dollars sold at a slight discount to gold

    dollars, which were more widely accepted in the western states. Prices for both U.S. trade

    dollars11 and Mexican dollars are indicated, as is the previous day’s price of silver at London.

    We can translate the London silver price into U.S. gold dollars to compute the premium over

    melt for coins at San Francisco using the mint parity exchange rate of 4.86656 gold U.S. dollars

    per British pound (Officer, 1986). The newspaper clipping indicates the previous day’s London

    silver quote is 53 3/4. When converted to dollars and adjusted for the weight and fineness of

    London silver versus dollar coins, the newspaper quotes at San Francisco on 4th May 1877 imply

    premiums of two or three percent for the coins.12

    Figure 3 plots Shanghai melt premiums for Spain’s Carolus silver dollar coin, Mexico’s

    silver dollar coin, and the silver-backed Shanghai tael. As described previously, the melt

    premium is computed by comparing the market price of a particular coin to its value if melted

    and sold at the contemporaneous price of silver in London. The figure shows that the Mexican

    silver dollar often sells for a premium of a few percent above its London melt value, though there

    10 As established in an earlier footnote, an uncirculated Mexican dollar contains 377.14 grains of pure silver. To compute the amount of pure silver in an ounce of sterling silver, multiply its weight, 480 grains, times it fineness, 0.925, to yield 444 grains of pure silver. Thus, the melt value of the Mexican dollar at Shanghai is 84.9414% of the London price of sterling silver, but the prices in the newspaper column indicate that the Mexican dollar trades for 98.6627% of the value of sterling silver translated into taels. Thus, the melt premium is about 13.72%. 11 To compute the amount of pure silver in a U.S. trade dollar, multiply its weight, 420 grains, times its fineness, 0.9, to yield 378 grains, which is slightly higher than the amount contained in the Spanish and Mexican dollars. See Leavens (1939), page 93. 12 Start with 377.14 (378) grains of pure silver in the Mexican (U.S. trade) dollar, divide by 444 grains in a sterling ounce, multiply by 4.86656, then by 53.75, divide by 240 (pence per pound), and each Mexican (U.S.) coin contains about 92.5782 U.S. cents of silver, but sells for more: bid 95, ask 95 1/2 (bid 94 ¾, ask 95 1/2).

  • 6

    are periods when it sells at a discount of up to ten percent. The convenience of drafts, demand

    deposits, and other forms of the “paper” Shanghai tael is evident: it typically sells for a premium

    over its melt value. The plot also shows that the Carolus dollar is not always traded, but when its

    price does appear in the newspaper, its melt premium is often enormous. There are points in the

    mid 1870s and mid 1890s when the Carolus melt premium approaches or exceeds 30% of its

    melt value. While we have no data on the cost of moving silver coins and bullion between

    London and Shanghai, Leavens (1939) estimates a cost of 2.5% to move silver between New

    York and Shanghai, a broadly comparable distance.13 Thus, the premiums indicated in Figure 3

    are far beyond transactions costs.14 Although the figure does not plot coin premiums, it is evident

    that the Spanish and Mexican coins almost always trade for no less than the paper tael,

    suggesting that coin premiums are typically substantial.

    In looking at the plot, a question that arises immediately is whether “bad money drives

    out good money”, as in a basic version of Gresham’s Law. Carolus dollars are evidently “good

    money”: the Spanish coin typically sold at a premium to other forms of money. Furthermore, bad

    does may indeed sometimes drive out good: at times, the Spanish coin disappeared from the

    market as there are no quotations in the newspaper. At the same time, it is not obvious why there

    should be significant price premiums for the Spanish coin over the Mexican coin which is

    identical in weight, purity, form, and familiarity.

    A practical problem for statistical tests that is evident in the plot is the extent of missing

    observations. The Carolus dollar prices do not appear in the newspaper for most of 1879 through

    1882, from August 1885 to August 1890, and past December 1899. The Mexican dollar series

    ends in 1924. The Shanghai tael series is complete except for about ten observations.

    Figure 4 plots melt premiums for Mexican and U.S. trade dollars at San Francisco, with

    the corresponding melt premiums from Shanghai for comparison. The time-span of the plot is

    much smaller than that for Shanghai because coin prices were published in the San Francisco

    Chronicle for a period of less than 35 years. The melt premium for Mexican silver dollars at San

    Francisco is broadly similar to that observed in Shanghai. The U.S. trade dollar was quoted in the

    newspaper for a period of less than a decade. Within a few years of its introduction in 1874, it

    13 For a broadly-related study, see Officer (1986) on gold bullion and coin movements between the U.S. and U.K. 14 In no sense should transactions costs limit the deviations between market prices and mint parity. The was no “free coinage” of these forms of money: the supply of genuine Carolus dollars was limited by the number of coins minted decades previously while the supply of new Mexican dollars was set by the Mexican government.

  • 7

    was typically traded at melt premiums in excess of ten percent, and at a higher price than the

    Mexican silver dollar. While intended to facilitate Asian trade by American merchants, the coin

    posed a unique problem for U.S. monetary authorities as many found their way back to the U.S.

    and competed with ordinary U.S. silver dollars in circulation. The trade dollar was demonetized

    for use in the U.S. in July 1876 and the last significant quantities were minted in 1878. It is ironic

    that the coin achieved its greatest success and highest melt premiums after it lost its legal tender

    status and ceased production.15

    3. Testable hypotheses

    The state of China’s monetary system in the late Qing dynasty and early Republican

    period leads to several testable hypotheses about the melt premiums observed for the Shanghai

    tael, the Spanish and Mexican “pieces of eight”, and the U.S. trade dollar relative to their melt

    value based on mint parity and the London silver price. In addition to these primary hypotheses,

    we will briefly test a few other ideas towards the end of the paper.

    Our first hypothesis motivates tests that link the melt premium on silver money to

    economic fundamentals:

    H1: Melt premiums for China’s silver money were driven by the demand for money to

    purchase Chinese products and, therefore, are positively correlated with measures of the

    demand for Chinese exports.

    Given data limitations, our primary monthly measure of demand for China’s exports is a proxy

    for the value of global trade detailed below. We also have annual numbers on China’s export

    trade. Furthermore, the London price of silver bullion may serve as another proxy that can

    explain silver money melt premiums:

    H2a: Melt premiums for China’s silver money are positively correlated with changes in

    the global price of silver, a proxy for the demand for Chinese exports.

    15 See Willem (1959) for a detailed history of this coin and Greenfield and Rockoff (1995) for a summary.

  • 8

    Alternatively, it may be the case that the direction of causality is reversed. Costly silver is not a

    consequence of heightened demand for China’s exports but instead dampens demand for those

    products:

    H2b: Melt premiums for China’s silver money were driven by the competitiveness of

    China’s exports and, therefore, are negatively correlated with changes in the global price

    of silver.

    A variation on H2b runs as follows. As China’s demand for imports rises, the foreign exchange

    market tends to experience selling of taels and silver coins as U.K. pounds and other foreign

    currencies are needed to purchase foreign products. Therefore, premiums for the drafts and

    demand deposits in Shanghai taels that facilitate exports drops off, and melt premiums decline.16

    Another set of hypotheses exploits the existence of another form of money in China, the

    small copper coins known as cash. If demand for China’s exports is booming, the demand for

    both copper coins, larger silver coins, and other forms of money should increase together:

    H3a: The price of small Chinese copper coins was driven by the same fundamental

    economic forces as the prices of China’s silver money. Therefore, melt premiums for

    China’s silver money are positively correlated with changes in the Shanghai price of

    copper cash.

    Alternatively, copper coins may have served as a partial substitute for silver money:

    H3b: Small Chinese copper coins served as a partial substitute for silver money.

    Therefore, melt premiums for China’s silver money are negatively correlated with

    changes in the Shanghai price of copper cash.

    Indeed, prior to 1800, copper coin was expensive, and silver was cheap in China. Later, silver

    became expensive and copper coin became cheap (Peng, 1994).

    16 See King (1988b), page 213.

  • 9

    We explore the dynamics of the relationship between melt premium on silver money and

    the health of China’s export economy with two competing hypotheses as follows:

    H4a: The market for silver money forecasts changes in China’s export trade. Therefore,

    the melt premium for silver money leads measures of the demand for Chinese exports.

    H4b: Increases in China’s export trade gradually drive demand for silver money to

    facilitate trade with China. Therefore, the demand for Chinese exports leads the melt

    premium for silver money.

    At points in history, large shocks to the global value of silver are believed to have had a

    substantial impact on China’s competitiveness and prosperity (Friedman, 1990; Burdekin, 2008).

    For another facet of the associations between money and trade, we consider the

    possibility that some of the forces at work in determining melt premiums can be detected in the

    relationship between those premiums and the size of China’s opium trade as registered by

    customs statistics:

    H5a: Opium served as an alternative currency that foreigners brought to Chinese ports to

    purchase Chinese exports. Therefore, the correlation between the size of the opium trade

    and the melt premiums is negative.

    H5a suggests that imported opium was an additional form of currency beyond silver and copper.

    It may have lessened the dependence of foreign traders on silver coins and, thus, reduced melt

    premiums for silver money.17 Alternatively, imported opium may have been just another foreign

    product in search of locally-accepted silver currency with which to buy Chinese products:

    H5b: Opium was sent to China to purchase Chinese products and increased the demand

    for silver money with which to purchase those products. Therefore, the correlation

    between the size of the opium trade and the melt premiums is positive.

    17 Lin (2006, p. 109) describes some evidence, in U.K. government statistical abstracts, that opium gradually replaced silver as a medium of payment for British businesses in the China trade.

  • 10

    During the period (1867 to 1927) for which we have annual data, legal imports of opium

    comprised, on average, slightly more than ten percent of China’s total trade as measured by the

    sum of imports and exports recorded by the customs agency. Lin (2006) describes how the

    opium trade produced a large infrastructure of distribution and payments across China,

    stimulating employment and investment. Given the importance of this import to China’s

    economy, it is plausible that the opium trade may be associated with the melt premiums for silver

    coins and silver-backed money.18

    In addition to the factors we have been able to measure to test the hypotheses just

    outlined, there are no doubt other stories that we would test if we had access to data. For

    example, lacking a unified central bank and monetary policy, reputable foreign banks supplied

    much of the private paper money in use in the modern sector of China’s economy. The supply of

    such money, and demand deposits, fluctuated through time more to suit the needs of the issuing

    banks than the economy generally. Thus, the melt premium for taels should be negatively

    correlated with the supply of bank notes and demand deposits in the economy.19 As another

    example, we might augment H2a and H2b with a related hypothesis if we could measure the net

    flow of silver coins and bullion in and out of China. If higher exports lead to a flood of Carolus

    dollars, Mexican dollars, and silver ingots into China, there should be a negative association

    between the net flow of silver and the melt premiums for silver money, yielding a prediction of

    the same sign as H2b. We may proxy for the flow of silver using the balance of trade as any

    trade surplus (deficit) for China would be met, in part, with an inflow (outflow) of silver.20

    To this point, all the hypotheses outlined apply to the melt premium for all forms of silver

    money circulating in China, not just coins. This includes the melt premium of the

    hypothetically-defined Shanghai tael, in addition to the melt premiums of silver coins. These

    premiums reflect the cost of shipping bullion and coins between London and Shanghai (and San

    Francisco), the cost of minting bullion into coins, and the cost of melting coins to yield ingots, in

    addition to the pure demand for silver money for commercial purposes. We also conduct some

    18 Bailey and Truong (2001) report a few correlations between measures of the opium trade and broader economic conditions in colonial Malaya and Netherlands East Indies. 19 See King (1988a) p. 210. 20 See Jamieson (1893). Furthermore, he notes that, because bullion and coin movements were not taxed or otherwise regulated, China’s Imperial Maritime Customs only began recording such flows in 1888.

  • 11

    tests on coin premiums for Spanish and Mexican silver dollars relative to the value of the

    Shanghai tael.

    Finally, the historical literature alludes to a legacy demand for the Spanish coin in silk-

    producing and trading areas of China.21 While no explanation or motivation for this demand is

    offered, we nonetheless formulate one final testable hypothesis:

    H6: Increases in demand for China’s silk drive demand for the Carolus dollar. Therefore,

    coin premiums for the Carolus dollar are positively related to the rate of growth of

    China’s silk exports.

    Other series (detailed below) that we use as explanatory variables in some tests may be

    thought of as addressing the hypotheses outlined above or as control variables. For example,

    monthly series on global stock prices and interest rates may proxy for the general level of global

    prosperity and, thus, the demand for China’s exports. Alternatively, the interest rate may proxy

    for the cost of holding inventories of silver coins and, therefore, should be negatively correlated

    with the premiums paid for such coins.

    4. Data and methodology

    4.1 Data overview

    The testable hypotheses imply a variety of factors that might be useful in explaining the

    observed melt premiums and coin premiums in China. As we detail throughout the paper, we

    encountered many limitations (such as the lack of monthly data on China’s trade) in the

    construction of this old data set. The maximum sample period for the series we collect is June

    1866 to January 1928. The exchange rate between taels and pounds and the prices of Mexican

    and Spanish coins described previously are the principal targets of our investigation. Wherever

    possible, we use the “sight” exchange rate as our measure of the pound value of the tael because

    it commences appearing in the newspaper earlier than telegraphic transfers or other forms of

    currency exchange.22 The coin and tael price series are collected from the North China Daily

    News, as is the price of copper cash. Also among the previously-described series, the London

    21 See King (1965) p. 177. 22 Given missing values for “sight”, about 20 of the observations of the pound – tael exchange rate are the telegraphic transfer rate.

  • 12

    silver price is used both to compute premiums on China’s silver money and to proxy for demand

    for China’s exports as in H4a and H4b. It is gathered from the North China Daily News

    whenever possible, else from other colonial or British newspapers. Our observations on coin

    prices at San Francisco are limited by their availability in the San Francisco Chronicle.

    Additional series are constructed to proxy for the volume of China’s trade and to control

    for general global economic conditions. The global stock return proxy is the British pound return

    on an equally-weighted index of U.S. and U.K. stock indexes. See Bailey and Bhaopichitr

    (2004).23 The global trade volume proxy is the total British pound value of imports and exports

    of the U.K. and the U.S., also following Bailey and Bhaopichitr (2004). The consol bond yield

    for a perpetual British government bond represents global long-term interest rates. It is available

    from the historical data files of the NBER (www.nber.org). All the series described to this point

    are available at monthly frequency. However, we have no monthly series on the state of China’s

    trade or real economy. Therefore, we conduct some tests on annual data tabled in Yang (1931)

    and Hsiao (1974), and originally constructed from publications of China’s customs agency for

    the period from 1864 to 1927. Specifically, we collect from these sources the value of total

    Chinese exports, total imports, exports of key commodities identified by Yang (1931) including

    silk, and the value of opium imports.24

    4.2 Methodology

    Before describing how the data will be used to test the hypotheses previously outlined,

    we must consider the properties of the raw series. Non-stationarity is a common characteristic for

    many economics time series. Indeed, a casual look at the plots of silver money melt premiums

    suggests that some series may be non-stationary. This implies problems with some of the

    estimators we employ if we use these variables in undifferenced form. Thus, for a more formal

    test of this issue, we use Dickey-Fuller tests (unreported but available upon request) with six lags

    to test whether our series contain a unit root and need to be differenced.

    Our main concern regarding non-stationarity applies to the melt premium series.

    Augmented Dickey-Fuller tests with up to six lags readily reject a unit root in the melt premiums

    23 We also tried Shanghai and Hong Kong stock returns from this source in some of the tests detailed below, but they proved to be uniformly statistically insignificant. 24 Interestingly, customs duties and statistics were denominated in yet another hypothetical currency unit, the Haikwon tael defined as 583.3 grains of pure silver and, thus, equal to 1.1140 Shanghai taels.

  • 13

    for the Mexican dollar at Shanghai, the Mexican dollar at San Francisco, and the Shanghai tael.

    Therefore, we use these series in undifferenced form. In contrast, we cannot reject the presence

    of a unit root in the melt premiums for the U.S. trade dollar at San Francisco. Therefore, we use

    this series in first-differences. The melt premium series for the Spanish silver dollar is

    problematic. Though augmented Dickey-Fuller tests cannot reject the presence of a unit root,

    inclusion of intercept and trend terms reject the presence of a unit root. Therefore, we use the

    Carolus melt premium series in undifferenced form.

    Following common empirical practice in economics and finance, we express the trade,

    silver price, copper price, and global stock index in log-differences, and also formally confirm

    that the raw series contain unit roots. Finally, we follow previous researchers and the unit root

    tests to first-difference our global interest rate series.

    Given our understanding of the structure of the data, we briefly describe the statistical

    techniques that will be used. The monthly silver money melt premium series will be related to

    the explanatory variables with linear regressions (and Newey and West (1987) adjusted standard

    errors) to test hypotheses H1 through H3b. The lead-lag associations between silver money

    premiums and the growth in trade will be estimated with VARs to test hypotheses H4a and H4b.

    Similar methodologies will be used to estimate a few other tests over annual data to

    accommodate the available Chinese trade data. This includes a test of H5 for a relationship

    between Carolus coin premiums and the silk trade.

    5. Empirical results

    5.1 Summary statistics

    Table 1 presents univariate summary statistics on the monthly variables in the data set.

    The often substantial melt premiums for silver coins suggested in the plots are confirmed by the

    estimated means and medians. The mean (median) melt premiums for the Mexican dollar at

    Shanghai, the Carolus dollar at Shanghai, and the Mexican dollar at San Francisco are 2.25%

    (2.08%), 16.09% (13.84%), and 1.66% (1.45%) respectively. Thus, melt premiums are typically

    larger than the mean (median) of 0.75% (0.82%) for the silver-backed paper Shanghai tael,

    implying positive coin premiums. While the melt premium for the Shanghai tael is small, it is

    nonetheless typically positive, which suggests a small premium for the convenience of bank

    drafts and bank deposits over bar silver.

  • 14

    Standard deviations for these series are on the order of two or three percent per month,

    except for the Carolus melt premium which is around nine percent per month. Although we treat

    the Mexican and Carolus melt premiums as stationary based on the Dickey-Fuller statistics

    previously discussed, they display substantial serial correlation, as high as 93.06% for the

    Carolus melt premium. This motivates the inclusion of autoregressive terms in some of the

    regression estimates we present subsequently. The extent of missing observations is also evident

    from the table. The U.S. trade dollar series is of particularly limited use given the small number

    of observations. However, it is interesting that the popularity this coin eventually garnered in the

    China trade as described in historical accounts is evident in the melt premium it fetched in San

    Francisco’s foreign exchange market.

    As was evident from the plots and Dickey-Fuller tests discussed previously, the U.S.

    trade dollar series presents stationarity problems and is used in first-differenced form. While

    Table 1 presents summary statistics only for this series in first-differences, we also compute the

    means and medians for the undifferenced melt premiums to compare to the statistics for the other

    silver money melt premiums. The mean (median) melt premium for the U.S. trade dollar (not

    reported in the table) is 9.28% (11.09%).

    Among the other series, the rate of growth of global trade displays strong negative first-

    order serial correlation. The series for silver, copper coins, global stock returns, and the global

    interest rate sometimes display less prominent autocorrelation though relatively high dispersion.

    The upper right-hand triangle of Table 2 presents parametric cross-correlations among

    the variables in the data set. Strong associations among most of the melt premiums measured in

    Shanghai and San Francisco are evident. For example, the correlation (p-value) between the tael

    melt premium at Shanghai and the Mexican dollar melt premium at San Francisco is 77.52%

    (

  • 15

    consistent with silver coins and copper coins being competing forms of commodity money

    (H3b). Correlations between the melt premiums and the other variables are less obvious, while

    there appears to be no worry about multicolinearity among the explanatory variables. The

    estimates in the lower left-hand triangle of the table for non parametric correlations are

    qualitatively similar.

    5.2 Tests to explain silver money premiums over melt value

    Table 3 presents estimates of regressions that relate monthly silver money melt premiums

    to the variables that test H1, H2a, H2b, H3a, and H3b. H1 predicts that melt premiums are

    driven by the demand for Chinese exports. This is weakly confirmed by the marginally

    significant slopes on the global trade growth proxy found for the Mexican dollar, Spanish dollar,

    and Shanghai tael melt premiums at Shanghai. For the Mexican dollar melt premium at

    Shanghai, strongly significantly positive slopes on the rate of growth of the London price of

    silver are consistent with H2a, the notion that silver proxies for demand for China’s products.

    They reject H2b, the idea that silver governs competitiveness of China’s exports. There is also

    some evidence that this may also be true for the U.S. trade dollar melt premium in spite of the

    relatively short time series of data we have for this series.

    Marginally significant positive slopes on the rate of growth of the price of copper cash in

    the case of the autoregressive specification for the Mexican dollar reject H3a (copper and silver

    monies are competitors) in favour of H3b (demand for different forms of money moves together)

    Significant or marginally significant positive slopes on the global stock index return for some

    specifications for the Mexican, Carolus, and tael melt premiums at Shanghai weakly confirm the

    notion that melt premiums are positively correlated with global prosperity. If the global stock

    index return proxies for demand for China’s exports, this is also consistent with H1.

    Throughout Table 3, autoregressive terms inserted to control for the autocorrelation

    structure of dependent variables are typically very significant but tend to clarify, rather than

    subsume, the explanatory power of the other variables.

    Panel A of Table 4 presents estimates of similar regressions that relate annual melt

    premiums to annual measures of Chinese trade that are available to us, plus annual versions of

    the silver price and copper cash price variables previously employed. Given the small number of

    years for which data on the Carolus dollar and trade dollar is available, we report only results for

  • 16

    the Mexican dollar premium and Shanghai tael premium, based on just over 50 observations.

    Given the power of the autoregressive terms included in the monthly regressions in the previous

    table, we report only specifications with the autoregressive term.

    The results for the Mexican silver dollar melt premium are consistent with H1, given the

    marginally significant positive slope on the rate of growth of total Chinese exports in the first

    specification and the significantly positive slope on the rate of growth of “key” Chinese exports

    in the second specification. In contrast, slope coefficients for the Shanghai tael melt premium are

    insignificant for the export measures. Thus, the demand for coins, but not money in more

    abstract form, seems positively related to the demand for China’s exports. However, these results

    are weak given that the number of observations is small. Furthermore, results (unreported) which

    exclude the autoregressive terms do not display the marginally significantly positive slope on

    total exports, though the strongly significant positive slope on “key” exports persists.

    The only other notable result in Panel A of Table 4 is the finding of marginally

    significant or significant negative slopes for the rate of change of the silver price, for both the

    Mexican dollar at Shanghai and the Shanghai tael. While the positive slopes in Table 3 suggest

    that silver proxies for the demand for China’s exports (H2a), the negative slopes on silver in the

    annual data results are consistent with silver governing the competitiveness of China’s exports

    (H2b). While this contradicts the monthly results, note that the significance of the slope

    coefficients in the annual results vanishes for (unreported) results that exclude the autoregressive

    term.

    Panel B of Table 4 offers brief evidence on associations between melt premiums and the

    opium trade. Regressions similar to those of Panel A replace the broader Chinese trade indicators

    with the extent of the opium trade, which we measure as the total value of opium imported

    divided by the sum of all imports and all exports. The regression results indicate statistically

    significantly positive slope coefficients on the size of the opium trade for melt premiums of both

    the Mexican dollar at Shanghai and the Shanghai tael. This is consistent with H5b, the notion

    that opium arriving in Chinese ports in search of Chinese exports increased the demand for silver

    forms of money, rather than the idea (H5a) of opium as another form of commodity money. As

    was the case for Panel A, however, other (unreported) specifications indicate that the

    significance of the coefficients on the opium measure vanish if the autoregressive term is

    included. While this reprehensible trade caused great suffering to millions of Chinese people and

  • 17

    is perhaps best forgotten, opium may have had a significant influence over broader monetary and

    economic conditions in China.

    5.3 Tests of dynamic associations between silver money premiums and trade

    Table 5 presents estimates of VAR systems that relate monthly silver money melt

    premiums to a proxy for global trading activity with lags of up to six months. Panel A presents

    the VAR estimates that relate the melt premium for Mexican silver dollars at Shanghai and the

    rate of change of the global trade proxy with six monthly lags. The number of usable

    observations, 578, hints at the effect of missing observations on the procedure. The substantial

    serial correlation in both series is evident, as is confirmed by the test statistics reported at the

    bottom of the panel. The tests also indicate statistically significant feedback from lagged global

    trade growth to Mexican silver dollar melt premium, but not from lagged Mexican silver dollar

    melt premium to global trade growth. This is consistent with the notion that growth in trade

    gradually drives up the melt premium (H4b), rather than the melt premium anticipating

    subsequent growth in trade (H4a).

    Panel B presents the VAR estimates that relate the melt premium for Carolus silver

    dollars at Shanghai and the rate of change of the global trade proxy. The number of usable

    observations, 181, is substantially lower than for the Mexican silver dollar. The VAR results for

    the Carolus melt premium reflects the significant autocorrelation of the two series, while there is

    no evidence of significant feedback between them to confirm either H4a or H4b.

    In Panel C, the two-equation system relates the Shanghai tael melt premium and the rate

    of change of the global trade proxy with six monthly lags. The results are qualitatively similar to

    those previously reported in Panel A for the Mexican silver dollar. Autocorrelation dominates

    both series, feedback from lagged global trade growth to the Shanghai tael melt premium is

    significant (H4b), and feedback in the other direction is insignificant.

    Similar to the Carolus silver dollar VAR reported in Panel B, Panel D presents weak

    VAR estimates using 205 usable observations on the melt premium for Mexican dollars at San

    Francisco. It may be the case that the relatively small number of observations for this series

    contributes to the weakness of the results relative to what was found for the Mexican dollar at

    Shanghai and the Shanghai tael.

  • 18

    We do not estimate VAR results for the melt premium on U.S. trade dollars at San

    Francisco because the length of the available time series is very short. For the same reason, we

    do not estimate VAR systems using the annual Chinese trade series rather than the monthly

    global proxies. Unfortunately, an exhaustive search led to the conclusion that the statistical

    returns of China’s customs service were never reported at higher than yearly frequency.

    5.4 Other tests

    The results of the previous two tables explain melt premiums for coins and for the silver-

    backed Shanghai tael. We also estimated similar regressions (unreported) with coin premiums,

    rather than melt premiums, as dependent variables. Relative to the findings for melt premiums,

    there is no evidence of H1 as slopes on the global trade growth proxy are insignificant. As was

    the case for the melt premiums, strongly significantly positive slopes on the rate of growth of the

    London price of silver are consistent with H2a (silver as proxy for demand for China’s products)

    and reject H2b (silver governs competitiveness of China’s exports) for the Mexican dollar coin

    premium. Marginally significant positive slopes on the rate of growth of the price of copper cash

    in the case of the autoregressive specification for the Mexican dollar reject H3a (copper and

    silver monies are competitors) in favour of H3b (demands for different forms of money move

    together). Significant or marginally significant positive slopes on the global stock index return

    for the Mexican and Carolus coin premiums at Shanghai weakly confirm the notion that such

    premiums are positively correlated with global prosperity and proxies for demand for China’s

    exports (H1). As was the case for the melt premium regressions, autoregressive terms are

    typically very significant.

    Sandrock (1995) notes that, during political or economic crises, the amount of trade

    would dry up the supply of foreign coins and, thus, increase coin premiums. While Sandrock

    (1995) cites the period of the Taiping Rebellion in the middle 1850s as an example and this is

    prior to the time period (1866 to 1928) for which we have data, the history of the late Qing

    dynasty and early Republican eras is littered with such events. Fears about the soundness of the

    local or global financial system could also increase coin premiums. For a quick look at this

    notion, we conduct a simple “event study” over our monthly data as follows. First, we create two

    dummy variable series. The “Chinese strife dummy” is set to zero except for months that

    represent unusually chaotic times: the Sino- Japanese War (August 1894 to April 1895), the

  • 19

    Hundred Days’ Reform (June to September 1898), the Boxer Rebellion (June 1900 to September

    1901), the Republican Revolution (October 1911 to March 1912), major protests (May 1919),

    and the general strike (July 1925). The “Global financial crisis dummy” is set to zero except for

    months of global financial chaos that may have affected economic conditions and sentiment in

    China: the Panic of 1873 (September 1873 to March 1879), the Panic of 1907 (May 1907 to June

    1908), and the First World War (June 1914 to November 1918). The selection of events and

    determination of their duration is, of course, arbitrary.

    We employ these dummy variables as explanatory variables in monthly regressions with

    coin premiums for the Mexican dollar at Shanghai, the Carolus dollar at Shanghai, and the

    Mexican dollar at San Francisco as dependent variables. Specifications include both the dummy

    variables by themselves and combined with the basic set of explanatory variables used in Table

    3. Surprisingly, the only regressions that show significance for the “Chinese strife dummy” are

    those for the Mexican dollar coin premium at San Francisco. The slope coefficients are positive

    and strongly statistically significant. Slopes on the “Global financial crisis dummy” and all

    slopes for the other coin premium series are uniformly insignificant.

    Finally, given mention in King (1965) of a unique demand for Carolus dollars related to

    the silk trade, we relate the Carolus coin premium to annual data on the rate of growth of the

    value of China’s silk exports. Given the very small number of annual observations available for

    the Spanish dollar, we estimated a simple regression of the Carolus coin premium on the rate of

    growth of the value of silk exports. The result (unreported) is statistically insignificant and, thus,

    offers no support for H6.

    6. Potential explanations beyond conventional economic forces

    Our empirical results have only limited ability to explain the melt premiums and coin

    premiums. It remains puzzling how such premiums could arise and persist when close substitutes

    (silver bullion, the silver-backed Shanghai tael, other foreign currencies) existed, there were no

    exchange controls or other frictions that would motivate a black market premium for coins, and

    there was no issue of differences in denomination.25 We can imagine other factors affecting the

    premiums that are so evident in Figures 3 and 4 and which our empirical tests are, at best, only

    25 Sargent and Velde (1999) discuss complexities given coins of the same composition but differing denominations.

  • 20

    able to explain partially. While it may be impossible to quantify these forces with periodic data

    that we can use in empirical analysis, it is worthwhile to discuss them.

    Given China’s political and economic turbulence during the decades we study, it is likely

    that the demand for silver coins reflected highly time-varying confidence in the Qing and

    Republican governments of the time. Indeed, premiums for foreign silver dollars might be

    thought of an index of Chinese consumer and business confidence, rather than just a

    consequence of economic factors such as the demand for China’s exports. The premium paid for

    coins over the silver-backed tael might simply be thought of an insurance premium: unlike tael

    bank deposits or silver-backed paper money, silver coins did not bear the risk of a collapse of the

    financial system or expropriation by governments, warlords, or other economic and political

    powers.

    “Confidence” leads very quickly to consumer, business, and investor psychology as a

    force affecting markets, prices, and decision-making, an idea with a long history. Keynes (1936)

    argued that short-term investor “sentiment” has a dominant impact on investment. The

    psychology of individual economic decision-makers is a subject that has received increasing

    attention recently. Kahneman and Tversky (1973), for example, find that decision-makers often

    misuse information. Useful applications of psychological concepts to economics and finance

    abound (Kahneman, 2003), and the emerging field of behavioral finance presents many

    theoretical and empirical illustrations of how investors can make poor decisions and incorrectly

    value assets due to behavioural biases or irrationality. Premiums and discounts for closed-end

    funds studied by Lee, Shleifer, and Thaler (1991) are a good example. Errors in judgement by

    individual investors are implied by the finding of Elton, Gruber and Busse (2004) that substantial

    investments have gone into index funds which charge high fees (over 2% per year) for passive

    holdings of broad stock indexes while identical funds with much lower fees are readily available.

    Investor sentiment, irrationality, and poor decision-making can distort investment

    decisions and financial market prices, and a Spanish silver coin in China a hundred years ago

    could be viewed, and priced, as a store-of-value, in addition to its utility in facilitating

    transactions. It is easy to imagine a sentiment-driven demand to hoard silver coins that varied

    with the ebb and flow of confidence in late Qing and early Republican economic and political

    conditions. Furthermore, if sentiment and fundamentals are correlated (Baker, Ruback and

    Wurgler, 2005), information about fundamentals, as we have used in this study, may have some

  • 21

    power to explain the often substantial premiums on Spanish and Mexican silver coins. Thus, the

    phenomena of substantial coin premiums for Spanish and Mexican silver dollars, and our

    correlation and regression results, may be manifestations of mass biases in decision-making.

    Such episodes have appeared throughout economic history, with the dot com bubble and today’s

    ongoing financial crisis as recent examples.

    With psychology in mind, we may even invoke “culture” as another potential explanation

    for the puzzling, sometimes-enormous coin premiums. The anthropology literature on money

    offers some interesting anecdotes concerning the perception of different types of currency. Guyer

    (2004) describes a time in the 1880s when a French government-supported explorer found that

    local people in Congo demanded outdated flintlock-style rifles for payments, rejecting the more

    modern rifles used by the French military. Dickinson (2007) describes associations between

    changes in the quality of Ukraine’s paper money after the 1991 dissolution of the Soviet Union

    and the sense of confidence Ukrainians expressed about their new country. In a study of modern

    Vietnam, Truitt (2006) describes “a lack of confidence in the state-issued currency even in the

    face of an expanding money economy”, noting that “money is an important register by which to

    analyze the contentious politics of symbolic exchange.”

    In the case of late Qing and early Republican China, the ongoing crumbling of the

    Chinese state and the rise of foreign political and economic power may have affected the

    substantial premiums for foreign silver coins that we observe. For example, the evolving

    language used by Chinese to describe foreign silver coins represents the changing view of the

    West in China. Lin (2006) describes how early silver European coins, which were often rough in

    quality and appearance, were referred to as “barbarian coins”, while starting in the late

    seventeenth century, the very precise machine-made European silver coins were referred to more

    respectfully as “Buddha heads” due to the royal portraits on the coins. These coins may have

    been perceived as a reliable store of value, to the point that the exchange market often offered

    premiums for them.

    A useful parallel is the manner in which foreigners approached China’s indigenous

    commodity money, the copper cash. Notar (2004) describes associations between social status

    and the form of currency a particular individual would use in Republican-era China. King (1965)

    quotes a foreigner writing in the North China Herald in 1887: “We have cash for our dealings

    with the poorer natives, dollars in our social and retail dealings, and taels for our higher

  • 22

    commercial affairs”. Foreigners typically refused to handle copper cash, and the economy

    evolved a system of “chits” with which foreigners could make small payments without using

    copper cash. While some of the objection to copper money related to its inconvenience, it was

    also shunned because it represented the backwardness and unpleasantness of the “native”

    economy. This contrasts with the favourable perception of foreign silver coins: they and, to a

    lesser degree, the Shanghai tael designed by the foreign banks inspired confidence since they

    represented the efficient, prosperous, and technologically superior Western world.

    Again, none of the ideas sketched in this section of the paper help us uncover additional

    economic factors and other proxies that can pick up time-variation in melt premiums or coin

    premiums due to time-variation in psychological or cultural factors. Furthermore, with the

    exception of the example of the demand for outdated rifles described by Guyer (2004), nothing

    in the previous discussion explicitly motivates or documents a market premium for a particular

    form of commodity money. These ideas are nonetheless interesting to speculate about, and

    illustrate the usefulness of concepts from beyond economics for interpreting our evidence, even

    if they cannot literally boost the r-squared coefficients of statistical tests.

    7. Summary and Conclusions

    This paper examines premiums paid for Spanish and Mexican silver coins which served

    as important currencies in China in the nineteenth and early twentieth centuries. During this

    period, the prices for these coins at Shanghai and San Francisco often exceeded melt value by a

    wide margin. Data for the period from 1866 to 1928 including basic economic indicators, prices

    of other forms of money, trade, and global economic measures are used to explain the melt

    premiums.

    While silver-backed Chinese currency typically sold at a slight premium to its melt value,

    Mexican and particularly Spanish silver dollars often sold at substantial premiums above melt

    value. Regressions offer several glimmers of evidence consistent with theoretical predictions

    concerning what caused those premiums. They are correlated with several measures of Chinese

    and global trade, competitiveness, and prosperity. The extent of the legal opium trade also

    appears to explain some of the premiums. The mere existence of these premiums strongly

    suggests that “bad money drives out good” is not always the case. In a chaotic economy where

  • 23

    individual decision-makers are experienced at handling different forms and qualities of money,

    “good money” has its price, and often sells for a substantial premium.

    A principal constraint to conducting research on this interesting historical period is the

    unevenness or absence of needed data. If, for example, there was data back to the early 1800s

    when the Spanish silver dollar gained much popularity in China, our evidence might be more

    conclusive, and certainly less noisy. Silver coins may have experienced reduced importance in

    the period for which we have data to study, while the end date of our data excludes interesting,

    more recent monetary events in the 1930s.26 Furthermore, the monthly price series for the

    Spanish dollar at Shanghai and both the Mexican and U.S. trade dollars at San Francisco are

    substantially shorter than the entire time period that we study. The availability of Chinese trade

    data on only an annual, rather than monthly, basis further constrains our tests. Throughout the

    paper, small time series and missing observations confine us to relatively simple reduced-form

    specifications.

    While the alternative psychological and cultural explanations for melt premiums that we

    offer cannot be quantified and used in our empirical procedures, they remind us that “the human

    factor” is potentially important in explaining economic phenomena like these silver money

    premiums. The combination of data, empirical tests, and speculative discussions based on

    concepts beyond conventional economics and finance give up a richer picture of this interesting

    currency market and its economic, political, and historical setting.

    26 See, for example, Burdekin (2008).

  • 24

    References

    Bailey, Warren, and Bhaopichitr, Kirida, 2004, How Important Was Silver? Some Evidence on

    Exchange Rate Fluctuations and Stock Returns in Colonial-Era Asia, Journal of Business

    77, 137-173.

    Bailey, Warren, and Truong, Lan, 2001, Opium and Empire: Some Evidence from Colonial-Era

    Asian Stock and Commodity Markets, Journal of Southeast Asian Studies 32, 173-193.

    Baker, Malcom, Ruback, Richard S., and Jeffrey Wurgler, 2005, Behavioral Corporate Finance:

    A Survey, unpublished Harvard Business School working paper (September).

    Burdekin, Richard C.K., 2008, US pressure on China: Silver flows, deflation, and the 1934

    Shanghai credit crunch, China Economic Review 19, 170 – 182.

    Burdett, Kenneth, Trejos, Alberto, and Randall Wright, 2001, Cigarette Money, Journal of

    Economic Theory 117-142.

    Chen, Chau-Nan, 1975, Flexible Bimetallic Exchange Rates In China, 1650-1850: A Historical

    Example of Optimum Currency Areas, Journal of Money, Credit and Banking 7, 359 -

    376.

    Dickinson, J.A., 2007, Changing Money in Post Soviet Ukraine, in Senders, Stefan, and Allison

    Truitt, eds., Money: Ethnographic Encounters, Oxford: Berg Publishers, 27 - 42.

    Elton, Edwin J., Gruber, Martin J., and Jeffery A. Busse, 2004, Are Investors Rational? Choices

    Among Index Funds, Journal of Finance 59, 261-288.

    Friedman, Milton, 1990, The Crime of 1873, Journal of Political Economy 98, 1159-1194.

    Garbade, Kenneth D., and Silber, William L., 1978, Technology, Communications, and the

    Performance of Financial Markets: 1840-1975, Journal of Finance 33, 819-32.

    Gervais, Raymond, 1982, Pre-Colonial Currencies: A Note on the Maria Theresa Thaler, African

    Economic History 11, 147-152.

    Greenfield, R.L., and Rockoff, H., 1995, Gresham’s Law in Nineteenth-Century America,

    Journal of Money, Credit and Banking 27, 1086-1098.

    Guyer, Jane I., 2004, Marginal Gains: Monetary Transactions in Atlantic Africa, University of

    Chicago Press.

    Hsiao, Liang-lin, 1974, China’s Foreign Trade Statistics, 1864 – 1949, Cambridge MA: Harvard

    University East Asian Research Center.

  • 25

    Jamieson, G., 1893, Effect of the Fall in Value of Silver in China, Journal of the Royal Statistical

    Society 56, 646-69.

    Kahneman, Daniel, 2003, Maps of Bounded Rationality: Psychology for Behavioral Economics,

    American Economic Review 93, 1449-1475.

    Kahneman, Daniel, and Amos Tversky, 1973, On the Psychology of Prediction, Psychological

    Review 80, 237-251.

    Kann, E., 1927, What is a Shanghai Tael?, Chinese Economic Journal 1, 769 – 772.

    King, Frank H.H., 1965, Money and Monetary Policy in China 1845 – 1895, Cambridge,

    Massachusetts: Harvard University Press.

    King, Frank H.H., 1988a, The History of the Hongkong and Shanghai Banking Corporation.

    Vol. 2: The Hongkong Bank in the Period of Imperialism and War 1895-1918:

    Wayfoong, the Focus of Wealth, Cambridge: University Press.

    King, Frank H.H., 1988b, The History of the Hongkong and Shanghai Banking Corporation.

    Vol. 3: The Hongkong Bank between the wars and the Bank interned, 1919-1945: Return

    from Grandeur, Cambridge: University Press.

    Leavens, D.H., 1939, Silver Money, Bloomington, Indiana: Principia Press.

    Lee, Charles, Andrei Shleifer, and Richard Thaler, 1991, Investor sentiment and the closed end

    fund puzzle, Journal of Finance 40, 75-110.

    Lin, Man-houng, 2006, China Upside Down: Currency, Society, and Ideologies, 1808-1856,

    Cambridge, MA: Harvard University Asia Center.

    MacKenzie, Compton, 1954, Realms of Silver: One Hundred Years of Banking in the East,

    London: Routledge and Kegan Paul.

    Michie, R. C., 1987, The London and New York Stock Exchanges 1850-1914, London: Allen &

    Unwin.

    Newey, Whitney K., and West, Kenneth D., 1987, A Simple, Positive Semi-Definite,

    Heteroskedasticity and Autocorrelation Consistent Covariance Matrix, Econometrica 55,

    703-08.

    Notar, B. E., 2004, Ties that Dissolve and Bind: Competing Currencies, Prestige, and Politics in

    Early Twentieth-Century China, in Werner, Cynthia, and Duran Bell, eds, Values and

    valuables: From the sacred to the symbolic, Lanham, MA: Rowman & Littlefield, 128 -

    157.

  • 26

    Officer, L. H., 1986, The Efficiency of the Dollar-Sterling Gold Standard, 1890-1908, Journal of

    Political Economy 94, 1038-1073.

    Peng, Xinwei, 1994, A Monetary History of China, Bellingham, WA: Center for East Asian

    Studies,Western Washington University.

    Rolnick, A.J., and Weber, W.E., 1986, Gresham’s Law or Gresham’s Fallacy?, Journal of

    Political Economy 94, 185-199.

    Sandrock, J. E. Copper Cash and Silver Taels: The Money of Manchu China, Baltimore, MD,

    Gateway Press, Inc., 1995.

    Sargent, J. and Francois R. Velde, 1999, The Big Problem of Small Change, Journal of Money,

    Credit and Banking 31, 137-161.

    Truitt, Allison, 2006, Big Money, New Money, and ATMs: Valuing Vietnamese Currency in Ho

    Chi Minh City, Research in Economic Anthropology 24, 283 – 308.

    von Glahn, Richard, 1996a, Fountain of Fortune: Money and Monetary Policy in China 1000 –

    1700, Berkeley: University of California Press.

    von Glahn, Richard, 1996b, Myth and Reality of China's Seventeenth-Century Monetary Crisis,

    Journal of Economic History 56, 429-454.

    Willem, J. M., 1965, The United States Trade Dollar, Racine, Wisconsin: Whitman Publishing

    Company.

    Yang, Tuan-liu, 1931, Statistics of China’s Foreign Trade During the Last Sixty-Five Years,

    Monograph No. IV, National Research Institute of Social Sciences, Academia Sinica,

    Shanghai.

    Zhao, Bin, 2006, Familiarity, Convenience, and Money: Spanish and Mexican Silver Coins in

    China, master’s thesis, Department of Applied Economics and Management, Cornell

    University.

  • 27

    Figure 1. Spanish, Mexican, and U.S. silver coins commonly used in China First row depicts the Spanish silver coin known as the “piece of eight” (given its denomination of eight reals) and the “Carolus” dollar given its portrait of King Charles III. The second and third rows show the standard Mexican silver dollar and the U.S. trade dollar respectively. All coins show evidence of “chopping”: Chinese and Southeast Asian merchants punched an indentifying mark on a coin deemed genuine.

  • 28

    Figure 2. Typical exchange rate columns from the North China Daily News and the San Francisco Chronicle The table includes columns from North China Daily News of 4 May 1877 and 5 July 1888, and 4 May 1877 San Francisco Chronicle.

  • 29

    Figure 3. Premiums over melt value for silver money at Shanghai The figure depicts end-of-month melt premiums for Spanish silver dollars, Mexican silver dollars, and silver-backed Shanghai taels. They are computed by comparing the price of each type of money at Shanghai (converted to U.K. pounds) to their melt value at the London silver price.

    -0.1

    -0.05

    0

    0.05

    0.1

    0.15

    0.2

    0.25

    0.3

    0.35

    0.4

    186606 187006 187406 187806 188206 188606 189006 189406 189806 190206 190606 191006 191406 191806 192206 192606

    frac

    tion

    of L

    ondo

    n si

    lver

    mel

    t

    Carolus dollar Mexican dollar Tael draft

  • 30

    Figure 4. Premiums for silver coins at San Francisco, December 1876 to November 1897 The figure depicts end-of-month melt premiums for Mexican silver dollars and U.S. trade dollars at San Francisco. They are computed by comparing the gold U.S. price of each coin (translated to U.K. pounds at mint parity) to the melt value of each coin at the London silver price. Shanghai premiums for Spanish and Mexican silver dollars are included for comparison.

    -0.1

    -0.05

    0

    0.05

    0.1

    0.15

    0.2

    0.25

    0.3

    0.35

    0.4

    187612 187806 187912 188106 188212 188406 188512 188706 188812 189006 189112 189306 189412 189606

    frac

    tion

    of L

    ondo

    n si

    lver

    mel

    t

    Carolus dollar Mexican dollar Mexican dollar (SF) US Trade dollar (SF)

  • 31

    Table 1. Univariate summary statistics for monthly currency premiums and other monthly variables The sample period for the end-of-month data is June 1866 to January 1928. Coin and currency premiums are computed comparing the market value of the coin to its value if melted and sold as bullion. Nobs Mean median std dev skew kurt rho1 rho2 rho3 rho4 rho5 rho6 Mexican silver dollar premium 687 0.0225 0.0208 0.0309 0.4098 2.7405 0.7111 0.5374 0.4238 0.3518 0.3121 0.2829 Carolus silver dollar premium 287 0.1609 0.1384 0.0944 0.1336 -1.1862 0.9306 0.8931 0.8774 0.8511 0.8463 0.8407 Tael premium 730 0.0075 0.0082 0.0219 0.3721 7.3021 0.5544 0.3184 0.1963 0.1359 0.0800 0.0842 Mexican silver dollar premium (SF) 244 0.0166 0.0145 0.0239 4.1902 41.0614 0.2450 -0.0090 0.1361 0.1361 -0.0314 0.0076 Change in trade dollar premium 68 0.0012 0.0020 0.0127 -0.5176 2.1614 -0.2072 -0.0759 0.0402 -0.0913 -0.0363 0.1118 Rate of change of British pence per ounce of sterling silver 739 -0.0012 0.0000 0.0333 -0.3832 8.0207 0.1340 0.0142 -0.0207 0.0107 -0.0446 -0.0552 Rate of change of tael price of copper cash coins 709 -0.0010 0.0000 0.0161 0.0349 5.9998 -0.0065 -0.1034 0.0111 0.0018 0.0149 0.0101 Rate of change of proxy for global stock prices 717 0.0042 0.0043 0.0239 0.0870 3.6153 0.0483 0.0504 -0.0512 0.0286 0.0491 -0.0541 Rate of change of proxy for global trade 673 0.0024 -0.0022 0.0772 -0.0073 0.6133 -0.3187 0.0461 0.0785 -0.1876 0.1137 -0.1457 Change in consol bond yield 734 0.0014 0.0000 0.0522 -0.3708 11.1301 0.0732 -0.0297 0.1823 -0.0432 -0.0771 0.1085

  • 32

    Table 2. Cross-correlations of melt premiums and other monthly variables The sample period for the end-of-month data is June 1866 to January 1928. Each cell contains the estimated correlation, its p-value, and the number of observations available to compute the correlation. Upper right triangle is parametric correlation while lower left triangle is Spearman rank correlation. Mexican Change Rate of Rate of Rate of Rate of Mexican Carolus dollar in trade change change of change of change of Change in dollar Dollar Tael premium dollar of silver cash global global consol

    premium premium Premium (SF) premium price price equities trade yield Mexican dollar 1.000 0.042 0.775 0.406 0.136 -0.156 -0.116 -0.015 0.046 0.020 Premium 0.475

  • 33

    Table 3. Regressions to explain monthly melt premiums The sample period is June 1866 to January 1928. Residuals are adjusted for serial correlation and heteroskedasticity following Newey and West (1987). T-statistics appear in parentheses beneath each coefficient estimate.

    Rate of Rate of Rate of Rate of Auto change change of change of change of Change in regressive of silver cash global global consol Adjusted constant term price price equities trade yield r-squared Nobs

    Mexican 1. 0.0225 -0.1414 0.2234 -0.0232 0.0207 0.0020 0.0329 580 dollar (11.40) (-2.70) (2.34) (-0.47) (1.74) (0.06) premium 2. 0.0052 0.7461 -0.2268 0.1734 0.0456 0.0330 -0.0117 0.5772 579 (5.34) (21.13) (-4.41) (3.16) (1.73) (2.59) (-0.56) Carolus 3. 0.1914 -0.1928 0.3835 0.2031 0.0759 0.0664 -0.0082 219 dollar (20.63) (-0.73) (0.84) (0.75) (1.54) (0.27) premium 4. 0.0116 0.9328 -0.3708 0.0325 0.0036 0.0530 -0.1019 0.8891 210 (3.02) (46.48) (-3.12) (0.25) (0.04) (1.95) (-1.31) Tael 5. 0.0081 -0.1671 0.0237 -0.0493 0.0058 -0.0070 0.0644 620 premium (6.27) (-4.06) (0.45) (-1.15) (0.57) (-0.29) 6. 0.0031 0.5821 -0.2054 0.0123 -0.0043 0.0191 -0.0159 0.3993 620 (4.30) (12.05) (-4.45) (0.32) (-0.16) (1.75) (-0.86) Mexican 7. 0.0168 -0.1034 0.0727 -0.0647 0.0005 0.0692 0.0018 242 dollar (7.96) (-0.95) (0.70) (-0.84) (0.03) (1.34) premium (SF) 8. 0.0108 0.2997 -0.2248 -0.0020 -0.0214 -0.0046 0.0519 0.0904 235 (2.69) (1.84) (-2.22) (-0.02) (-0.38) (-0.33) (1.00) Change in 9. 0.0002 -0.7695 0.0832 -0.0456 -0.0068 0.0319 0.6283 66 trade dollar (0.24) (-5.58) (1.22) (-1.18) (-0.59) (1.42) premium 10. 0.0002 -0.1101 -0.7061 0.1419 -0.0501 -0.0098 0.0239 0.5973 63 (0.21) (-3.25) (-5.43) (2.14) (-1.20) (-0.89) (1.08)

  • 34

    Table 4. Regressions to explain annual melt premiums with Chinese trade measures The sample period is 1866 to 1927. Residuals are adjusted for serial correlation and heteroskedasticity following Newey and West (1987). T-statistics appear in parentheses beneath each coefficient estimate. Raw series on exports and imports are measured in terms of their value in customs taels, rather than in terms of physical quantities. Panel A: Melt premiums and broad measures of China’s international trade Rate of Rate of Rate of Rate of Rate of Rate of Auto change change of change of change of change of change of China’s Regressive of silver cash China’s China’s China’s key China’s key trade Adjusted constant Term Price price exports imports exports Imports balance r-squared Nobs Mexican 1. 0.0099 0.4866 -0.0405 0.1192 0.0328 0.0052 0.2015 53 dollar (2.32) (4.15) (-1.74) (1.38) (1.79) (0.35) premium 2. 0.0114 0.5002 -0.0503 0.1317 0.0346 -0.0114 0.2348 51 (2.56) (4.06) (-2.22) (1.41) (2.76) (-1.00) 3. 0.0252 0.2586 -0.0267 0.0970 0.0270 0.0262 53 (5.63) (3.15) (-1.04) (1.02) (0.88) Tael 4. 0.0057 0.3289 -0.0255 -0.0110 -0.0044 0.0063 0.1221 56 premium (2.23) (2.75) (-1.87) (-0.35) (-0.25) (0.33) 5. 0.0066 0.3200 -0.0261 -0.0165 0.0013 0.0009 0.1288 54 (3.20) (2.55) (-1.71) (-0.53) (0.16) (0.21) 6. 0.0067 0.3297 -0.0290 -0.0129 0.0110 0.1395 56 (1.69) (2.65) (-2.06) (-0.36) (0.34) Panel B: Melt premiums and the scale of legal imports of opium into China Rate of Rate of Value of opium Auto change change of imports / Regressive of silver cash (value of all imports + Adjusted Constant Term Price price value of all exports) r-squared Nobs Mexican dollar premium 1. 0.0114 0.5232 -0.0301 0.1113 0.0127 0.2567 52 (3.31) (5.16) (-1.26) (1.37) (2.55) Tael premium 2. 0.0053 0.3961 -0.0238 -0.0066 0.0107 0.2118 55 (2.77) (3.18) (-2.26) (-0.18) (2.49)

  • 35

    Table 5. Vector Autoregressions (VARs) for monthly melt premiums and global trade proxy The sample period for the end-of-month data is June 1866 to January 1928. Results for the melt premium for U.S. trade dollars at San Francisco are not estimated given the small number (50) of usable observations. Panel A: Mexican silver dollar melt premium versus rate of change of global trade Mexican dollar Global premium trade Coefficient T-statistic Coefficent T-statistic Mexican dollar premium{1} 0.6084 14.4284 -0.2254 -1.5457 Mexican dollar premium{2} 0.0720 1.4534 0.2273 1.3266 Mexican dollar premium{3} 0.0158 0.3184 -0.1540 -0.8972 Mexican dollar premium{4} 0.0360 0.7270 0.3139 1.8348 Mexican dollar premium{5} 0.0040 0.0805 -0.1316 -0.7731 Mexican dollar premium{6} 0.0413 1.0007 -0.0672 -0.4705 Global trade{1} 0.0242 1.9847 -0.3101 -7.3536 Global trade{2} 0.0309 2.4180 -0.0517 -1.1685 Global trade{3} -0.0045 -0.3531 0.0449 1.0138 Global trade{4} -0.0338 -2.6452 -0.1457 -3.2931 Global trade{5} -0.0138 -1.0652 -0.0183 -0.4082 Global trade{6} -0.0046 -0.3749 -0.1170 -2.7795 Constant 0.0042 3.5624 0.0035 0.8610 Chi-sq. test Mexican dollar premium 97.9804 0.0000 1.0681 0.3804 Chi-sq. test Global trade 2.4904 0.0219 15.1809 0.0000 Usable Observations 578 Total Observations 625

    Panel B: Carolus silver dollar melt premium versus rate of change of global trade Carolus dollar Global Premium Trade Coefficient T-statistic Coefficent T-statistic Carolus dollar premium{1} 0.7353 9.5707 -0.0243 -0.1279 Carolus dollar premium{2} 0.0878 0.9305 0.1913 0.8197 Carolus dollar premium{3} 0.1206 1.2909 0.0433 0.1874 Carolus dollar premium{4} -0.1585 -1.7463 -0.0369 -0.1646 Carolus dollar premium{5} 0.1407 1.5969 -0.1537 -0.7058 Carolus dollar premium{6} 0.0595 0.8592 0.0749 0.4370 Global trade{1} 0.0115 0.3687 -0.5737 -7.4151 Global trade{2} -0.0154 -0.4332 -0.3048 -3.4578 Global trade{3} -0.0898 -2.5020 -0.1932 -2.1782 Global trade{4} -0.0487 -1.2996 -0.3104 -3.3515 Global trade{5} -0.0437 -1.1525 -0.2222 -2.3696 Global trade{6} 0.0046 0.1405 -0.1687 -2.0801 Constant 0.0037 0.7184 -0.0187 -1.4540 Chi-sq. test Carolus dollar premium 269.2903 0.0000 0.6531 0.6875 Chi-sq. test Global trade 1.4502 0.1984 10.5675 0.0000 Usable Observations 181 Total Observations 330

  • 36

    Table 5 continued Panel C: Shanghai tael melt premium versus rate of change of global trade Tael Global Premium trade Coefficient T-statistic Coefficient T-statistic Tael premium{1} 0.5152 12.7750 -0.3400 -2.1472 Tael premium{2} 0.0249 0.5474 0.2094 1.1742 Tael premium{3} 0.0220 0.4835 -0.1924 -1.0787 Tael premium{4} 0.0631 1.3902 0.2872 1.6129 Tael premium{5} -0.0558 -1.2324 -0.1417 -0.7968 Tael premium{6} 0.0582 1.4566 -0.0790 -0.5041 Global trade{1} 0.0120 1.1785 -0.3190 -7.9823 Global trade{2} 0.0322 3.0164 -0.0643 -1.5337 Global trade{3} 0.0217 2.0307 0.0397 0.9466 Global trade{4} -0.0079 -0.7394 -0.1397 -3.3236 Global trade{5} -0.0128 -1.1828 -0.0143 -0.3377 Global trade{6} -0.0070 -0.6917 -0.1184 -2.9617 Constant 0.0028 3.4056 0.0048 1.4543 Chi-sq. test Tael premium 45.6498 0.0000 1.3762 0.2218 Chi-sq. test Global trade 2.2251 0.0392 17.2208 0.0000 Usable Observations 628 Total Observations 667

    Panel D: Mexican silver dollar melt premium at San Francisco versus rate of change of global trade Mexican dollar Global premium (SF) trade Coefficient T-statistic Coefficent T-statistic Mexican dollar premium (SF){1} 0.2256 3.1495 0.0595 0.2805 Mexican dollar premium (SF){2} -0.1768 -2.2752 -0.2233 -0.9705 Mexican dollar premium (SF){3} 0.3220 2.5539 -0.1284 -0.3439 Mexican dollar premium (SF){4} 0.0395 0.2994 0.2994 0.7669 Mexican dollar premium (SF){5} -0.1435 -1.0526 0.2504 0.6202 Mexican dollar premium (SF){6} 0.1308 1.1285 0.3223 0.9391 Global trade{1} -0.0073 -0.2971 -0.4899 -6.7324 Global trade{2} -0.0060 -0.2172 -0.3843 -4.7215 Global trade{3} -0.0311 -1.0981 -0.1464 -1.7488 Global trade{4} -0.0778 -2.7758 -0.2509 -3.0225 Global trade{5} -0.0764 -2.6754 -0.1187 -1.4048 Global trade{6} -0.0376 -1.4690 -0.1924 -2.5359 Constant 0.0091 3.3338 -0.0086 -1.0654 Chi-sq. test Mexican dollar premium (SF) 3.7014 0.0017 1.0214 0.4126 Chi-sq. test Global trade 1.7376 0.1142 10.5064 0.0000 Usable Observations 205 Total Observations 245

Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.