Inflation in Spain
Elements of the standard formula
The standard formula recommends that the government make overvalued tokens that it redeems for full-bodied coins. To create tokens, the government should terminate free minting of small denominations and thereby end the associated automatic mechanism governed by the minting and melting points, with the purpose of divorcing the value of small coins from that of the metals they contain. To determine the exchange value of the small coins, the standard formula instructs the government to offer to convert token coins into fullbodied coins. The possibility of pushing overvalued small coins into circulation was discovered before the importance of convertibility was appreciated.
The standard formula ties up smaller stocks of valuable metals in a country’s coins than did the comprehensive medieval commodity money system. A government hungry for revenues might be expected to recognize that. By moving from a full-bodied to a token system, a government could, at no cost in inflation, collect one-time revenues equal to the value of the precious metal in its subsidiary coinage. This chapter describes how the government of seventeenth-century Castile tried to do just that, thereby implementing part, but only part, of the standard formula.1 The government did not promise to convert the subsidiary coins into silver. However, it eventually realized that market fundamentals alone could not determine the exchange rate for these token coins2 and used announcements to manipulate the rate of exchange of token coins for full-bodied coins.
It is not surprising that the role of convertibility was not immediately appreciated, because theoretically a token system can work without convertibility. Without convertibility, a token coin system is incomplete but it is not incoherent. With wide bands for the subsidiary coinage and without convertibility, our model asserts that there are many exchange rates that are consistent with equilibrium. Technically, the exchange rate is indeterminate. 3 The theoretical indeterminacy of the exchange rate means that government attempts to pin it down by directive or persuasion make sense. In the episodes to be described in this chapter, watch how the government manipulated the exchange rate.
The Castilian experiment
In chapter 4, we described the arrival of the new cylinder press technology in Spain, and the construction of a new machine called the Ingenio. King Philip II recognized that this new technology offered possibilities beyond those of simply making better versions of the old coins. In a royal decree of 1596, he announced that pure copper coins produced in the cylinder presses of the Segovia mint would replace the billon coinage previously produced for the smaller denominations. The text explained his decision to substitute copper for silver (Del Rivero 1918–19, document 14):
We have been advised by men of great experience that the silver put in those billon coins is lost forever and that no profit can be drawn from it, except in their use as money, and that the quantity of silver that is put to that use for the necessities of ordinary trade and commerce in this kingdom is large. We have also been advised that, since we have established a new Ingenio in the city of Segovia to mint coins, if we could mint the billon coinage in it, we would have the assurance that it could not be counterfeited, because only a small quantity could be imitated and not without great cost if not by the use of a similar engine, of which there are none other in this kingdom or the neighboring ones. And it would thus be possible to avoid adding the silver.
Until then, copper, silver, and minting costs each formed a third of the face value of billon4 coinage. With Philip II’s decree, the silver was withheld and the copper content reduced.5
Philip II sought to raise government revenues without generating inflation. The same decree of 1596 ordered that a limited quantity of the new coins be minted each year, and that an equal quantity of existing small denomination coins be retired, until the outstanding billon coins were eliminated. Gold and silver coins were not changed. In terms of our model, the aim was to replace m1 (small denomination coins) with token coinage, but to maintain a narrow interval between the melting and minting points for the large denomination silver coins composing m2. The price level would stay within the interval between the melting and minting points for silver coins so long as they continued to circulate.
Philip II died in 1598. His successors Philip III (1598–1621) and Philip IV (1621–65) used the Ingenio to produce copper coins far in excess of the quantity of m1 of 1596, so much that the silver coins composing m2 disappeared. That released the price level from the constraints imposed by the melting and minting points for silver coins, and set loose the quantity theory cast in terms of copper coins as the determinant of the price level.
The Castilians struggled with exchange rate indeterminacy. They discovered that the face value of token coins, the number inscribed on them, could be made irrelevant. They learned to manipulate that number either to collect more seigniorage or to fight inflation. At first sight, some of these manipulations may seem absurd, but on closer study they were not. For example, on occasion the government “cried down” coins by saying that the numbers on copper coins represented not what the Roman numerals usually meant, but something else decreed by the government. We see these measures as serious attempts to respond to the government’s growing awareness of the exchange rate indeterminacy implicit in the system of token coinage it had set up.
Monetary manipulations
The complexity of the Castilian episode requires additional notation to keep track of units of account, exchange rates, and government operations designed to alter them.
The unit of account was the maravedi (mr), originally representing 
of a silver real, or 0.094g of fine silver. After copper coins had been issued, people distinguished between silver mr, the original unit, and copper mr. Copper mr were the units in which the government assigned numbers to copper coins. The actual copper coins were not called maravedis, but for example, cuartillos, cuartos, ochavos, and blancas, to which the government initially assigned the numbers 8, 4, 2, and ½, respectively. We shall take the cuarto as our representative token coin.
Starting in 1602, the copper coins were stamped with numbers (see fig. 14.1), originally denoting the number of copper mr assigned to those coins. We say “originally” because the government recurrently tampered with the numbering system in two ways, restamping or “crying down” particular coins.
Figure 14.1 Two 8-maravedis coins (“cuartillos”) minted at the Segovia mint (1622, left, and 1604, right). The original face value VIII (8 mr) is marked to the right of the central coat of arms. Both coins were reduced to 4 maravedis in 1628. The coin on the right was stamped “XII” (12 mr) in 1642 in Burgos. (Author’s collection.) Photograph: Robert Lifson
Restamping
In this operation, the government took a coin that was initially stamped x and restamped it jx, where j might equal 2, 3, or 4. Figures 14.1, 14.2, and 14.3 show various restamped coins. Figure 14.1 shows two coins that testify to both restamping and crying down. Both coins, called cuartillos, were originally stamped VIII mr. Figure 14.2 shows one coin that was restamped three times and another coin that had originally been issued over 100 years earlier and that was restamped in 1603. Figure 14.3 shows a coin originally issued in the time of the Roman emperor Domitian that was restamped fifteen centuries later to circulate at 12 mr; evidently, only the most recent marking mattered.
Restamping operations were administered as follows. The government offered a citizen the opportunity to bring to the mint j coins originally worth x maravedis, have all be stamped with jx, allow the government to keep j – 1, and to retain the last coin.6 Such an operation left the coin owner indifferent. To break indifference, the mints of Castile offered a small premium, in the form of an allowance for the costs of travel to the mint.
Figure 14.2 Two 2-maravedis coins (“ochavos”), restamped in the seventeenth century. The coin on the right was restamped three times: “IIII S” in Seville in 1603, “VI G” in Granada in 1636, and “IIII” in 1654 (the date is visible at the bottom, the numeral is on the other side of the coin). The coin on the left still bears the recognizable legend of Ferdinand and Isabel (1480–1504): it was thus over a hundred years old when it was restamped in 1603. (Author’s collection.) Photograph: Robert Lifson.
Figure 14.3 Bronze coin of Emperor Domitian, 1st c. A.D. This coin was stamped 12 mr in Granada in 1636. (Photograph: Bibliothéque nationale de France, Cabinet des Médailles, Paris.)
Crying down coins
When the government cried down a coin, it told people to disregard the number stamped on the coin when using it, and instead to refer to another number from a government edict. Let x denote the prevailing number of copper mr per cuarto, either written on the coin or in the pertinent government edict, depending on the momentarily prevailing convention. The government cried down the cuarto by saying that a coin originally stamped x1 mr was now to be regarded as x1/j mr, where j might be 2, 3, or 4, regardless of the stamp. This cry-down operation reset x from x = x1 to x = x1/j. Thus, sometimes the number on the coin mattered and sometimes it did not. The actual number of mr represented by a given coin could be set by the edict that had cried down the coin.
The two coins in figure 14.1 were marked 8 at issue. In August 1628, both were cried down to 4. Later, in 1642, the one on the right was restamped XII as part of the factor-of-3 restamping. So for this cuartillo, 8 became 4 (by being cried down in 1628) and 4 became 12 (by being restamped in 1642).
Theory of the Castilian tokens
We modify our model to accomodate restampings and cry-downs.
Additional notation
Table 14.1 summarizes notation. We let m2 denote the stock of silver coins, measured in silver mr, and m1 the number of copper coins called cuartos. We use the parameter x to denote the number of copper mr per cuarto, as determined by the cumulative effects of successive stampings and government edicts crying down the cuarto. Changes over time in x induced by restamping and crying down will be a big part of our story. We let e be the exchange rate of copper maravedis per silver maravedi, and e* = e/x the cuarto-silver maravedi exchange rate; p is the price of consumption goods in cuartos and p* = p/e* is the price of consumption goods in silver maravedis. We let φ–1 be the world price of copper and 
the world price of silver, each denominated in consumption goods per ounce of the metal. As we shall see, Castile was ultimately to use enough copper in making coins to move the price of copper in European markets.7
Table 14.1 Notation for the Castilian episode.
| Variable | Meaning | Units |
| m1 | cuartos, vell´on coins | number of disks |
| m2 | silver coins | silver mr |
| x | number assigned to a cuarto | copper mr/cuarto |
| e | exchange rate | copper mr/silver mr |
| e* = e/x | exchange rate | cuarto / silver mr |
| p | price of cons goods | cuarto/cons good |
| p* = p/e* | silver price of cons goods | silver mr / cons good |
| φ−1 | world price of copper | cons goods / oz copper |
| b1 | content of a cuarto | ounces of copper / cuarto |
 |
silver price of a cuarto | silver mr /cuarto |
| e*−1m1 | stock of cuartos | silver mr |
| b2 | content of silver coins | ounces of silver / silver mr |
| φ2−1 | world price of silver | cons goods / oz silver |
Multiple regimes
The list in table 14.1 contains exogenous variables that the government took as given, policy variables directly under the government’s control, and endogenous variables. We take φ and φ2 as exogenous, and we note that b2 remained fixed throughout the period.8 The government, manipulated the policy variables m1, b1, x, in order to achieve particular outcomes in terms of the endogenous variables m2, P, and e. We see the Spanish king’s advisors as seeking to diminish m2, and thereby reap the resulting silver as government revenues. We also see them as concerned with keeping the price level subject to the discipline of the silver standard. They discovered that several regimes were possible, depending on whether m2 was positive or not.
To see this, consider these bounds imposed by the supply mechanism. The upper bound was the melting point for silver, γ2 = φ2/b2, expressed in silver maravedi per consumption good. In other words, γ2 is the price level in terms of silver mr above which it is profitable for people to melt silver mr. The lower bound is (1 – σ2)γ2, the price level associated with the minting point for silver mr. For simplicity, we will assume that σ2 = 0, so that the two bounds coincide and the price level in silver mr, px/e, must be γ2 so long as silver mr have not all been melted:
where p* is the price level measured in silver mr per unit of consumption good.
The quantity theory of money (2.7a) in terms of prices denominated in cuartos is
while in terms of prices denominated in silver maravedis the quantity theory is
Two regimes are possible, depending on whether or not m2 > 0. In each regime, the quantity theory equation plays a different role.
When equation (14.1) prevails, the price level is determined by (14.1). Then the role of the quantity theory equation (14.4) is to determine the mix of the currency stock between m1 and m2. That is, it determines m2 as a function of 
The government set x, by what it stamped on the coins or proclaimed in government documents, and it affected m1 through its minting policies. The market set e. The government’s purpose was to stay in this first m2 > 0 regime but to lower m2.
In another regime, m2 = 0. The restriction on the price level p* imposed by the melting point of silver (14.1) vanishes. This makes the quantity theory equation (14.4) collapse to a quantity theory cast purely in terms of cuartos:
We can measure both m1 and p in silver mr by making a pure units change, and multiplying both the numerator and the denominator by 
Depending on the regime in which it found itself, the government experimented with different policies. Within the m2 > 0 regime, it used several methods to increase 
and to reduce m2 to its benefit. One was simply to mint new coins and increase m1 to purchase silver coins. Another was to reduce the b1 parameter, and then use a restamping to exchange new, lighter coins for the outstanding heavier ones. Finally, a direct restamping, or j-fold increase in x, allowed the government to collect a fraction (j – 1)/j of the existing money stock.
Within the m2 = 0 regime, the government tried to decrease p* enough to bring back some silver mr, that is, to make m2 > 0 again by equation (14.1), and thereby discipline the price level with the silver standard. Since 
it could do so by acting on p through equation (14.6), that is, by decreasing the actual quantity of cuartos in circulation (by “consuming vellón,” to use the phrase of the time). Or it could try to change x/e by a cry-down, that is, a reduction in x.
We see the government of Castile learning about properties of our model, such as the existence of the two regimes we have just described. The price data from the beginning of the seventeenth century are symptomatic of the m2 > 0 regime. However, chapter 6 told how Juan de Mariana had warned earlier of a possible m2 = 0 regime. Either the king’s advisors learned the hard way, or they deliberately ran the risks that Mariana foretold because they saw no better way to meet their goals.
More learning: aspects of indeterminacy
Our model has another property with which the government appeared to wrestle, namely indeterminacy. What matters in the quantity theory formulas (14.3) and (14.4) is x/e. While the government acted upon x through restampings and cry-downs, the market set e. It is theoretically possible, though not necessary, that the market could always have reset e proportionally with x, thereby removing the effects of restampings or cry-downs on m2 or the price level denominated in silver mr. However, the time series evidence about e and x in figure 14.6 (later in this chapter) shows that movements in e did not always cancel movements in x, so that the restampings and cry-downs mattered. We want now to mention some of the theoretical possibilities for the co-movement of x and e.
The form of equation (14.4) indicates that multiple equilibria are possible, because in response to a change in x, either m1, e, or p can adjust. We mention two types of equilibria:
1. Equilibria with x/e a constant. In such equilibria, fluctuations in x are irrelevant changes in units. Notice that if equation (14.4) holds for one x > 0 with m1 > 0, it also holds for any other positive value of x with x/e held constant. Here the market immediately cancels the effects of any government adjustments of x.
2. Equilibria in which e stays fixed while x is altered. In these equilibria, movements in x affect either the stock of silver coins m2 (if m2 > 0) or the price level denominated in silver mr (if m2 = 0). We consider each possibility in turn.
a. Suppose that m2 > 0, so that some silver coins circulate. Then an increase in x is consistent with fixed e and fixed 
provided that m2 shrinks in response to the increase in x. Notice that the price level p denominated in cuartos varies inversely with x in this case. Philip II’s advisors wanted that. The melting point for silver prevents the price level from rising in response to an increase in xm1/e while m2 > 0.
b. If (x/e)m1 ≥ kγ2, where γ2 is the upper bound on the price level implied by the melting point for silver, then further rises in x cause either the price level measured in silver mr or the exchange rate e to rise. Which form inflation takes is irrelevant because both lead to the same value of the price level p denominated in cuartos.
The evidence presented in the next section shows that x/e fluctuated over time, ruling out a pure type 1 equilibrium. But aspects of a type 1 equilibrium were reflected in particular episodes.
The evidence
Figures 14.4, 14.5, and 14.6 summarize the Spanish monarchs’ experiments with token coinage (see the appendix of this chapter for sources).9 Figure 14.4 plots the nominal (m1x, in copper maravedís) and real (
, in silver maravedis) values of the vellón stock.
For a single vellón coin called the cuarto, figure 14.5 shows the intrinsic value and the value in exchange or the market value, each measured in silver maravedis per cuarto.10 The intrinsic value measures b1p*/ø in our model; b1 is the ounces of copper per cuarto, φ is the ounces of copper per unit consumption good, and p* is silver maravedis per unit consumption good. The market value measures e*–1, in units of silver maravedis per cuarto. Note that b1p*/φ < e*–1 implies 
Thus, a larger gap between 
and e*–1 implies a larger “fiat component” to the value of cuartos. The figure implies that, most of the time, the exchange value of cuartos exceeded their melting point, often substantially.
Figure 14.4 Nominal value m1x (in copper mr) and real value 
(in silver mr) of the vellón stock, 1595–1680. 1 ducat (D) = 375 maravedis (mr). The dates of restampings and cry-downs are marked by Δ and ∇, respectively.
Figure 14.6 plots the movements in x and e. We observe fluctuations in both series. Movements in x are dictated by the government’s sequences of restampings and cry-downs. The series e represents observations on the market-determined exchange rate between copper mr and silver mr.
Figure 14.5 Market value (e*–1 = x/e) and intrinsic value (b1p*/φ) (in silver mr) of a vellón cuarto coin, 1600–1660. The dates of restampings and cry-downs are marked by Δ and ∇, respectively.
The chronology of operations was roughly as follows (see table 14.4 for details). Implementation of Philip II’s edict of 1596 led to a gradual replacement of older, partly silver small coinage with the new copper coins minted in Segovia. In 1602, the government decided to produce new copper coins with half as much copper as the previous ones. In figure 14.5, the cuarto’s intrinsic value became only ⅛ of its market value. The government observed that e remained constant. It then recalled the coins minted before 1602 and restamped them (increased x) by a factor of 2. In figure 14.6, x doubled. The operation was successful, in that large quantities of coins were voluntarily brought to the mint for restamping, and the government collected 50% of the amounts brought. No inflation ensued, encouraging the government of Castile to mint large quantities of the new coins, not just in Segovia but in the other mints as well, even though these other mints continued to rely on the medieval technology, producing poor quality coins known as “thick billon” (vellón grueso). Figure 14.4 shows two periods of substantial minting: from 1602 to 1608, when a concerned parliament extracted from the king a promise to halt minting, and again from 1618.
Figure 14.6 Index numbers of x (copper mr/cuarto) and e (copper mr/silver mr) for a vellón cuarto coin, 1600–1660.
These expansions of m1x continued until 1626, when the vellón coinage had completely replaced silver as medium of exchange. Comparing the nominal and real balances of the vellón coinage in figure 14.4 indicates that by 1626 an upper bound on real balances had been reached, corresponding to k, the demand for all real balances.11 Inflation set in, as e began to increase rapidly (fig. 14.6). Alarmed policy makers abruptly halted the minting of copper coins, a measure that stabilized e for a while. They spent two years deliberating over the best way to stem the inflation and the incipient fall in real balances shown in figure 14.4. They started an open-market operation to exchange copper coins for bonds but soon cancelled it. That they contemplated this operation shows that they understood how the price level was under sway of a quantity theory cast in terms of the vellón coins of the form (14.6). That is, they believed that p* varied directly with m1e*–1. Eventually, in 1628, they devalued (lowered e*–1≡ x/e), by crying down x, halving the face value of the copper coinage overnight. This brought the market exchange rate between copper and silver coins part of the way back to its pre-1602 value: the premium on silver fell overnight from 84% to 10%.
Castile issued no more copper coinage until 1660, and therefore ceased to act on m1 directly. But it carried out four restamping operations (increases of x) to multiply the face value of the copper coins by 2, 3, or 4. These restampings are detailed in table 14.4 of the appendix; they are also marked by upward-pointing arrows in figure 14.4, and recognizable as sharp increases in x in figure 14.6.12 Each of these restamping operations was soon followed by a cancelling cry-down or reduction in x. The devaluations are marked by downward-pointing arrows, and their effects appear in the downward movements in nominal values in figure 14.4 and figure 14.5, and in x in figure 14.6. Each restamping operation generated revenues, as individuals brought in coins to exchange them unit per unit, affording seigniorage rates of ½, ⅔, or ¾. But the government found that the revenues it could raise were diminishing. As shown in figure 14.4, the real balances of vellón fell over time.
Figure 14.6 shows abrupt adjustments in x that were not accompanied by abrupt movements in e (for example, in 1602 and in 1651) so that the data are not consistent with the straight type 1 equilibrium mentioned earlier. The restamping of 1651 is particularly noteworthy, for it had no effect on the silver premium, either in Hamilton’s monthly series or in Micón’s daily notations. Yet government accounts show that the restamping was very successful, as appears in table 14.2. On the other hand, movements in e during cry-downs at times came close to matching the movements in x, as shown in table 14.3 which compares the overnight changes in x with the corresponding overnight changes in e.
Table 14.2 Extent of restampings. The first column gives the number of existing coins called for restamping, the second gives the number of coins that were presented.
Table 14.3 Responses of e to changes in x.
The end of the token coin experiment
From 1660 to 1680, Castile issued small coins made of a mixture of copper and 7% silver, this time using the cylinder press technology in all mints. Initially, the government tried to give the coins a face value above intrinsic content, but depreciation forced it to abandon that project, and the coins were issued at close to intrinsic value, like medieval billon. Extensive counterfeiting (in spite of the use of the new technology) ultimately forced the government to abandon billon altogether: the government’s technological advantage had been lost. After 1680, the small coinage in Castile was, like elsewhere in Europe, full-bodied copper, with a return to the hammering method. Not until after the advent of the Bórbon dynasty were the hammers and the cylinder press replaced with imported screw presses in 1728. The Castilian experiment with token coins was over.13
Figure 14.7 Real balances of assignats during phases of the French hyperinflation, 1789–96. Source: Sargent and Velde (1995).
Comparison with inflation during the French Revolution
The shape of the time series for real balances in figure 14.4 resembles a graph for real balances of the French paper assignats issued during the French Revolution, figure 14.7 (taken from Sargent and Velde 1995). The assignat was first issued in 1789 as a claim redeemable at auctions of church land nationalized in 1789. The government nationalized the land so that it could get revenues by then selling it. The government intended to use proceeds from the land sales to service and retire the national debt. The state issued assignats to pay for goods and services and to service its debt. To absorb or “back” assignats, the government auctioned church lands in exchange for assignats. It promised to burn the assignats redeemed at these auctions, and for a while actually did so. Vertical lines in figure 14.7 divide the period into three broad regimes of monetary policy. The line in 1793 indicates the beginning of the Terror, when price and wage controls and prohibitions on holding stores of value other than the assignat were enforced with instruments of Terror. The line in mid-1794 indicates the end of the Terror.
Before mid-1792, nominal issues of assignats saw mild inflation and large and steady increases in the real value of the total stock. During this period, assignats displaced metal as a means of exchange. The metal left France for other countries.14 During this first period, the French government was mostly faithful to the auction scheme. But war raised the government’s need for revenue and induced it to suspend the auctions and to print assignats. Inflation accelerated and real balances started to fall. The policy response to this situation was the set of economic measures enforced by the Terror. Thus, our second regime was the Terror, whose economic components included a set of harshly enforced legal restrictions designed to stimulate the demand for assignats and thereby increase the base of the inflation tax. The Terror arrested the inflation and the fall in real balances. Vast quantities of assignats were printed without causing inflation. In July 1794, with France winning the war and the coalition that supported the Terror eroding in France, the Terror ended. This began the third regime, with deficits again being financed by printing assignats. This policy produced the classic symptoms of hyperinflation: accelerating inflation and falling real balances.
Apart from their time scales, figures 14.4 and 14.7 have similar patterns and witness similar monetary experiments. The different time scales show a much more rapid pace of events during the French Revolution. In both cases, a token currency was issued. In Spain it was made of copper, in France of paper. In both cases there initially occurred a buildup of real balances of the token currency, accompanied by only modest depreciation of the token money. In both cases, real balances of the token currency eventually constituted virtually the entire stock of real balances, driving out other monies and threatening to activate the quantity theory. In both cases, there eventually emerge periods of persistent inflation accompanied by declining real balances of the tokens. Thus, in the Castilian example, in the period from 1602 to the early 1620s, the token currency replaced other currencies, with little observed inflation, as occurred at first when the French revolutionaries introduced a paper currency.15 In the French case, the subsequent period of declining real balances of the token money exhibited all of the features of the twentieth-century hyperinflations described by Cagan (1956). It is likely that related forces account for the long decline in real balances in figure 14.4, because the repeated restampings and subsequent devaluations must eventually have been expected, prompting people to economize on m1.
Unintended consequence for Sweden
An unintended consequence of Spain’s monetary reform was to bring a full-bodied copper money and also inflation to Sweden.16 An important element in the story is reflected in figure 14.5, where the intrinsic value
shows interesting movements in the 1620s. Its rise in the early 1620s was partly caused by the rise in the price of copper in terms of silver that accompanied Spain’s massive coining of vellón coins. Its fall after 1626 reflects Spain’s suspension of new coining.
Mechanization came to Sweden in 1625. The following year, Spain stopped minting copper. When Spain withdrew from copper, it caused the demand for copper from Sweden (the biggest producer in Europe) to fall, because Spain had been consuming an amount equal to half of Sweden’s copper output (Wolontis 1936, 221).17 The Swedish king, Gustavus Adolfus, repeatedly pressed his French ally to buy Swedish copper and to mint it. In 1636, he nearly succeeded, but the project was blocked at the last minute by the French Cour des Monnaies (Spooner 1972, 189–90). To promote demand for Swedish copper, Sweden then moved to a copper standard, which it retained until the mid-eighteenth century, when it went to a paper currency. The Swedish experience differed from the Spanish because the coins were full-bodied rather than fiduciary. The Swedish government’s aim was to enhance the demand for copper; a full-bodied coinage achieved this aim more aptly than a token one. The inflation that occurred in Sweden came from a declining copper content of the unit of account (Heckscher 1954, 88–91).
Concluding remarks
The governments in seventeenth-century Castile experimented with token coins. They understood and implemented parts but not all of the standard formula. Informed by advisors who recognized that the state could reap revenues by instituting a token subsidiary coinage, kings of Spain issued inconvertible token coins, at first in moderate amounts that let them gather revenues without causing inflation. But the kings eventually issued enough token coins to cause the first big fiat money inflation in Europe. For decades after inflation had emerged, the kings struggled sometimes to repair the system but other times to raise more revenues from it. In doing so they used a variety of measures that led contemporary observers to formulate early versions of the quantity theory of money. Following their footsteps, we have interpreted these Castilian experiments in terms of key parameters of our model. For us as well as for the early quantity theorists, purposefully or not, the kings of Castile performed wonderful and informative experiments. They were to be copied.
Appendix: money stocks and prices
Table 14.4 presents a chronology of monetary operations on copper coinage in Castile between 1597 and 1680, pieced together from various sources.18 We distinguish between calderilla, that is, coins made any time before 1602 and containing some silver, like the coin on the right in figure 14.2, and the pure copper coinage that followed. Monetary ordinances and reforms treated those two types of coins differently.19
Our estimates of the copper stock of money are constructed by keeping track of the stock of calderilla and pure copper separately. The starting point for both series is in 1596: the initial stocks are 3 million ducats (1 ducat = 375mr) for calderilla (Domínguez Ortiz 1960, 252) and 0 for pure copper.
Table 14.4 Chronology of billon and pure copper coinage (vellón grueso, segovianos) in Castile, sixteenth and seventeenth centuries. The mint equivalent is expressed in mr per marc (230g).
*: non-Segovia coinage was ordered restamped by 2 on Feb 11, 1641.
Quarterly time series for the nominal stocks are then constructed by accounting for increases due to minting of new coins, crying up and restamping, and decreases due to by crying down and depreciation. Like Motomura (1997), we set depreciation at 1% per annum, which is consistent with what is known about wear and tear of coinage and allows us to match contemporary estimates of the money stock. Output, whether due to minting or restamping, is computed from Motomura’s figures until 1642 and Domínguez Ortiz (1960, 273) for later restampings. See García de Paso (2000a, 2000b) for more precise estimates. For the restampings of February and November 1641, we keep track of Segovia coinage and non-Segovia coinage separately (the former representing 15% of the total copper coinage).
The silver premium series, which is quarterly before 1650 and monthly after 1650, is in Hamilton (1934, 93, 96; 1947, 28). For the period 1618 to 1667, we have a daily premium series collected by a Genoese banker based in Madrid named Cosme Micón (see García de Paso, whom we thank for alerting us to these data).
1 The story of vellón inflation in Castile is told in Schrötter (1906), Hamilton(1934), Domínguez Ortiz (1960), Motomura (1994), Lozanne (1997), García de Paso (2000b). The exact chronology of monetary operations between 1597 and 1664, as well as the data and calculations used in this section, are described in Velde and Weber (2000), and summarized in the appendix of this chapter. This chapter benefited from the comments of Josíe García de Paso.
2 Without a promise to redeem, the “fundamentals” of a token coin are necessarily weak.
3 Kareken and Wallace (1981) write about the indeterminacy of international exchange rates in environments where there is no special source of demand for domestic over foreign currencies.
4 The Spanish word for billon is vellón, which later came to designate the pure copper coins.
5 In response to parliamentary opposition, 0.3% of silver was actually put in the new coins until 1602.
6 We say “originally worth” rather than “originally stamped” because the coin that was brought in might have been cried down earlier, for example, as the coins in figure 14.1 both were.
7 See chapter 18 for an account of a worldwide model of bimetallism in which the monetary authority’s choice of commodity affects its equilibrium price.
8 With one shortlived exception from December 1642 to March 1643, when the silver real was cried up by 25%.
9 Compare figure 14.4 with figure 10.4 on page 175.
10 The intrinsic value is constructed as the market value, in silver mr, of the copper content of the coin, copper prices taken from Motomura (1994) until 1626, and from Amsterdam prices for Swedish copper (Posthumus 1946, 271– 72) with a 30% markup for transport costs. The market value is constructed as the face value of the coin at each point in time, deflated by the premium on silver as for figure 14.4.
11 Velde and Weber (2000) estimate the pre-existing silver stock at 15 to 25 million ducats, based on the population of Castile and estimates of per capita money holdings in other countries at the time. Hamilton (1934, 91) states that 95% of transactions used vellón. Castilian vellón even began circulating in bordering areas of France in 1626 (Spooner 1972, 178).
12 Since figure 14.6 charts the story of a pure copper coin, only three restampings appear; compare with table 14.4, right columns.
13 See García de Paso (2000a) for the reform of 1680–86.
14 See Hawtrey (1919) and Sargent and Velde (1995).
15 A notable difference is that, in the Castilian case, there was no explicit backing for the token coinage.
16 See Friedman and Schwartz (1963, 699–700) and Brandt and Sargent (1989) for differing analyses of the Roosevelt administration’s silver purchase program on the world price of silver and thereby on monetary affairs in China.
17 In figure 14.5, note the fall in the world price of copper after the mid 1620s, roughly coinciding with the termination of Spain’s producing new copper coins.
18 The account in Hamilton (1936) is useful but incomplete and in places erroneous. It needs to be complemented with Domínguez Ortiz (1960). A good study of the Valladolid mint (Pérez García 1990) is very helpful as well. See also Carrera i Pujal (1943–47). Numismatists provide precious information: in particular, Fontechay Sánchez (1968) catalogs the stamps used in the various restamping operations.
19 In 1641, a difference was even made between the segovianos, coins minted in the Ingenio of Segovia like those of figure 14.1, and the other coins, variously called vellón grueso, pechelingues, or moneda de Cuenca.
