Project supported by the Deutsche Forschungsgemeinschaft (VO984/3-1)
In this project, exchange rates between currencies circulating in the Holy Roman Empire, the hanseatic area and the kingdom of Poland of the fourteenth to sixteenth centuries are used to analyze the integration of financial markets. In this context, both the numerous local silver currencies and the fewer interregionally used gold currencies are considered. In a first step, the development of the nominal exchange rates of these currencies is examined. On this basis it is asked whether and how far exchange rates between currencies based on the same kind of precious metal deviated from parity; deviations are then used as indicators of weaknesses in market integration (cf. Heckscher, 1916). Exchange rates between currencies based on different kinds of precious metals can be used to calculate local gold-silver-ratios. Spreads between these ratios likewise indicate weaknesses in integration. Based on this method to analyze financial market integration, the impact of monetary policies pursued by many late medieval governments is to be examined. Specifically, the effects of currency unions and of attempts to exclude foreign money from circulation need to be considered.
Following Mokyr (1990, pp. 4 f.), the analysis of pre-modern economic performance commonly makes a distinction between several types of growth. Solowian Growth is due to an increase in investment; Schumpeterian to an increase in the rate of innovations. Smithian growth, finally, is the result of a growth of the market that gives economic agents more chances to specialize and to increase their productivity. Before industrialisation, such a growth of the market may well have been the most important engine of economic growth as a whole,1 the level of investment being generally low and the rate of innovations moderate (at least compared to modern conditions, cf. Postan, 1967; Hilton, 1975; Thrupp, 1983, p. 170; Mokyr, 1990). Thus, if there was any growth at all between the fourteenth and the sixteenth centuries, it was in all probability due to market integration, which therefore played a key role in the period here under consideration.
As transport costs limit changes to profit from arbitrage between localities, the weight value ratio of the goods traded co-determines the degree of integration. Markets for mass goods whose value per unit of weight is low integrate less easily than markets for goods with a high value per unit of weight. Even under late medieval conditions when commodity money based on precious metals dominated and fiat money was relatively unimportant, there were hardly any markets whose integration was easier to achieve than financial markets. Money - even coins - had a more favourable weight-value ratio than practically any other commodity. As financial market integration was therefore probably more advanced than the integration of markets for goods with a less advantageous weight value ratio, financial markets provide a viable measure to assess the degree of market integration on the whole: they show the optimum that could be reached in this field at any given place and point in time. Hence, analyzing financial market integration is of particular importance.
As indicated above, this project investigates the hanseatic region, the Holy Roman Empire (north of the Alps) and Poland. There are several reasons why this area is particularly suitable for an analysis such as that planned here: On the one hand, the sources are relatively well-preserved and plentiful, with many of them having been edited (see below). On the other hand, there is little prior research. This seems to be due to the large number of different currencies circulating in Germany, Poland and North-Eastern Europe, the monetary situation in this area consequently being comparatively unclear (see below). However, for this project the multiplicity of currencies actually is an advantage. Without it, the relevant sources would contain far fewer notations of exchange rates, which are crucially important for setting up the data base. Moreover, compared to Western Europe and above all Italy, the Holy Roman Empire, the hanseatic area and Poland were underdeveloped as far as financial technologies were concerned. Non-cash payments were known and used, but were - apart from the Netherlands - much less important than south of the Alps. For reasons discussed below this makes it much easier to interpret the relevant data (see below).
In the period of time analyzed here, coins were without exception made of either silver or gold of varying fineness. It is therefore possible to relate exchange rates to the bullion content of the currencies involved. If these currencies were based on the same kind of precious metal, deviations of exchange rates from parity are used as indicators of weaknesses in market integration; if they were based on different metals, spreads between local gold-silver-ratios are used for this purpose. The project proceeds in two steps: First, it is necessary to collect the data, that is, exchange rates and information on the bullion content of the coinage; next, these data are to be analyzed. Additionally, the project considers the links between financial market integration and economic performance, using data on population, urbanization and wage levels as proxies of the latter.
In the context of the project, several publications are planned. On the one hand, a number of articles are to be published in refereed journals, on the other, a handbook of Central and Northern European exchange rates (fourteenth to sixteenth centuries) and a handbook of European monetary standards (fourteenth to sixteenth centuries) are to be published. As for the contributions to journals, the following subjects are envisaged:
- The Influence of Information Costs on the Integration of Financial Market
Question: How did the development of information costs influence the integration of financial markets? (Provisional analyses suggest that there was a strongly significant negative correlation, with higher information costs being linked to less integration.)
- Currency Unions, Optimal Currency Areas and the Integration of Financial Markets
Question: Were well-integrated financial markets a precondition of the formation of currency unions such as the Rappenmuenz-Union or the Wendish Monetary Union, or did the formation of these unions trigger integration?
- Political Autonomy and Market Integration
Question: How did urban monetary autonomy influence the integration of financial markets? (Provisional analyses indicate a strongly significant positive correlation: financial markets between cities that were autonomous with regard to monetary policies were much better integrated than markets between cities whose currencies were supplied by feudal rulers.)
- Financial Market Integration and Economic Growth
Question: Which conclusions can be drawn from financial market integration with regard to the integration of other markets, and how did it affect economic performance as a whole?
The handbook of Central- and Northern European exchange rates will contain data on exchange rate notations as found in the sources and in metric form. Each notation will be supplemented by information on its date and place, on the type of source and on where it is to be found. The handbook of monetary standards will provide information on the tale and fineness of the coins minted between the fourteenth and sixteenth centuries. Here, the date when the standard was established or determined, the face value of the coin, its weight and the weight of fine gold or silver it contained, the type of the source and the place where it is to be found will be given.
Sources and Data
Which sources can be used in order to determine the development of fourteenth to sixteenth century exchange rates? First of all there is a relatively small number of commercial account books: From the Holy Roman Empire of the fourteenth to sixteenth centuries, only about twelve have been published (Mantels, 1866; Koppmann, 1885; Nirrnheim, 1895; Mollwo, 1901; Rörig, 1928; Posthumus, 1953; Woehlkens, 1971; Lesnikov, 1973; Rowan, 1974; Jenks, 1983); the number of un-edited books is not known. However, given the fact that the archive of Bruges alone holds several such books (cf. Murray, 1999), it is plausible to assume that the situation in archives of other important centres of trade is similar, in particular for the sixteenth century. Apart from commercial account books those of gold smiths, who were frequently handling foreign currencies, are of interest. For example, one such book from the years between 1478 and 1515 has been preserved in the archive of the city of Constance.
Commercial correspondence is of similar importance for determining the development of exchange rates, but its preservation is patchier. From Northern Europe, there is only one comprehensive group of sources: the correspondence of the Veckinghusen-family from between 1395 and 1425 (Stieda, 1887; 1894; 1921). Neither quantity nor quality of this material match that of Southern European sources, e.g. from the Datini-archive from Prato. Individual letters have been published in many urban or territorial collections of sources such as the 'Urkundenbuecher' of Luebeck, Mecklenburg or of the Hanseatic League (Verein für Lübeckische Geschichte, 1843, ff.; Verein für meklenburgische Geschichte und Alterthumskunde, 1870, ff.; Höhlbaum, 1882-86, ff. ). These sources, too, contain occasional exchange rate notations.
Account books kept by political authorities are generally better preserved. As for territorial authorities, the Teutonic Order in Prussia is famous for its record keeping, in particular at the end of the fourteenth and in the early fifteenth century (cf. Sarnowsky, 1993). The Order's account books contain much information on exchange rates. As for Central Europe, urban account books are even more comprehensive and plentiful. Many of them have been edited, e.g. those from Hamburg and Basle (Koppmann, 1869, ff.; Harms, 1909, ff.), others need to be examined in the archives. As a rule, the development of exchange rates can be best followed on the basis of these sources. Several problems need to be taken into account when the sources described above are examined. Usually, the author of the source simply mentioned the how large the value of a certain sum given in one currency was in another currency. However, information on how he arrived at the exchange rate on which he based his equation is almost always lacking. Here, there were several possibilities (Spufford, 1986, S. l ff.): the most simple one was the so-called manual exchange, that is, the simultaneous and on the spot exchange of coins of one currency for those of another. A more developed form of exchange made use of bills of exchange; a practice that spread from Italy to Northern Europe since the thirteenth century. Bills are mentioned in commercial account books such as those kept by Hildebrand Veckinghusen (cf. Lesnikov, 1973). Finally, exchange rates were occasionally also imposed by political authorities.
The relative importance of the several ways exchange rates developed is under dispute. As for politically imposed rates, Miskimin (1985/89, pp. 148-151) claimed that late medieval and early modern authorities rarely were able to enforce their directives. As there is evidence that market rates really diverged from political rates (Volckart, 1996, pp. 144 f.), the latter are here excluded from the analysis, which is exclusively restricted to exchange rates that developed under the influence of supply and demand. Regarding the use of bills of exchange, two points need to be considered. For one thing, there is the likelihood that exchange rates mentioned in bills contain a hidden interest rate for the credit involved in the transaction. Therefore, they can systematically deviate from rates based on manual exchange (de Roover, 1968, pp. 32 ff.). However, such deviations can be shown to have existed only if rates mentioned in bills can be compared with rates from the same date mentioned in other types of sources. When that is impossible, there is only the alternative between ignoring these rates and treating them as all others. Furthermore, some historians (e.g. Henning, 1981) claim that already by the fourteenth and fifteenth century, bills and other instruments of credit were an important part of the money in circulation. If that was the case, their bare existence would have influenced the quantity of money and therefore the development of exchange rates. Still, bills could be a fully-fledged substitute for hard money only if they were freely negotiable. There are some isolated instances of their transferability from fifteenth-century Northern Europe, but on the whole their endorsement became common only in the early seventeenth century (Munro, 1991). As in the late Middle Ages and during much of the sixteenth century, coins dominated in circulation (Spufford, 1986, p. xxxi; Day, 1980/87, p. 2), most rates mentioned in the sources - including those contained in merchant account books and commercial correspondence - are ultimately based on manual exchange. In the first part of the project, a data base of the rates needs to be established.
Following this, the bullion content of the coins involved in these transactions needs to be established. The principal class of sources that contain the relevant information are mint ordinances and contracts concluded between the authority that issued the coins and the mint master. Usually, such documents defined the fineness of the alloy from which the coins were to be made, and the number of coins to be drawn from a specified quantity of that alloy. They could be straightforwardly interpreted if it were not for several obstacles. For one thing, in some cases there is no clarity about the exact metric equivalents of the units of weight used between the fourteenth and sixteenth centuries. For another, the ability of medieval and early modern mint technicians to make chemically pure gold and silver has been questioned (Miskimin, 1963, p. 31; Jesse, 1928, p. 160). And finally, once in circulation, money became worn down and defaced. For silver, losses due to wear and tear have variously been estimated at between 2 and 2.75% per decade (Mayhew, 1974, p. 3) and between 0.25 and 0.87% per year (North, 1990, p. 108). Still, losses and wear and tear influenced the amount of specie in circulation, and therefore probably affected the level of prices, but as far as exchange rates are concerned, their effects were less important. Presumably, coins made of both metals suffered alike from defacement, so that its effects on gold and silver cancelled each other out.2 Still, for this reason, too, a margin of error is unavoidable.
Fortunately, there is a group of sources that helps us to minimise such errors. Many late medieval and early modern authorities had foreign money assayed more or less regularly (cf. Ropp, 1878, pp. 223 f.; Cahn, 1895, pp. 169 ff.; Munro, 1972, p. 212 ff.). The interpretation of contemporary assays is, of course, problematic due to our imperfect knowledge of the metric equivalents of ancient units of weight, but if these sources are checked against the results of modern chemical tests (cf. Grierson, 1981; Kubiak, 1986), it is possible to derive a clear enough picture of how much gold and silver really changed hands when money was exchanged. In this way, a group of sources can be used to yield precise data which until now have been largely disregarded by research on medieval and early modern economic history and whose usefulness for estimating the integration of financial markets has passed unnoticed.
Methods of analysis
On the basis of these data, it is possible to estimate the degree of integration of financial markets in the late medieval and early modern Holy Roman Empire, in Poland and in the hanseatic area. To do this, new methods developed in the context of financial markets econometrics, specifically nonlinear threshold autoregression models, are used. The analysis proceeds in two steps. First, we examine the integration of financial markets where currencies based on the same precious metal were traded; next that of markets where both gold- and silver-currencies were exchanged. The most common approach to estimating market integration relies on the so-called 'law of one price'. The better integrated interlocal markets are, the more similar are the effects of supply and demand on prices, which accordingly converge or at least move into the same direction. By contrast, when the law of one price does not apply, that is, if prices differ between localities, this indicates that there were opportunities for arbitrage which remained unused due to high transportation or transaction costs or because of political barriers to entry to the market. The first of the two above-mentioned methods of analyzing financial market integration is based on a special case of the law of one price. In a world of commodity money and perfect monetary integration, exchange rates correspond to the relative bullion content of the currencies. With positive transportation and transaction costs, rates fluctuate around parity. The upper and lower points of these fluctuations are called bullion points; their position depends on the level of transport and transaction costs. If there are data both on nominal exchange rates and on the bullion content of the coinage, it is possible to use a threshold autoregression model in order to estimate the bullion points (between which rates move randomly) and the extent to which nominal rates violate these points. At the same time, it is possible to analyze how much time was needed for such rates to return to the area within the points. Both the length of this span of time and extent to which nominal exchange rates violate the bullion points are used as indicators of weaknesses of market integration. This method is discussed in detail in Volckart and Wolf (2006).
Provisional results suggest that between the fourteenth and sixteenth centuries, nominal exchange rates did not permanently deviate from parity, but that any violation of the bullion points was corrected by processes of arbitrage. Based on this, it is possible to use further sources hitherto disregarded by research: Exchange rates between currencies based on different precious metals and data on their respective bullion content can be used to calculate local gold-silver-ratios. Such ratios can then be interpreted as prices paid on bullion markets, which can be interlocally compared. Once this is done, a broad array of instruments for the analysis of long-term market integration can be used. In particular, we can utilise the method based on the employment of threshold autoregression models in this context, too.
The exchange rate of some type of gold coin sold in a specific place for silver money can be defined as Eh = kCs
Cg , where k represents the sum in the silver currency (Cs) which equalled one gold coin (CG). The par ratio between these currencies is given by Rh= kCsC
CgG. Here, S is the silver equivalent of the unit of account of the silver currency, and G is the fine gold content of the gold coin. The gold-silver ratio valid at this place is the average of the par ratios found per year (i.e. RL1'), subject to the restrictions described in the previous section. Ratios valid at the other places here considered (RLn) are defined analogously and are aggregated on a yearly level. Spreads between both aggregates are therefore given by ∆= |RL1'-RLn'| .
In a further step of the analysis it is necessary to explain regional and temporal differences in the integration of financial markets. To do this, the threshold approach needs to be extended in order to include geographical and institutional variables. First, geographical position and infrastructure are taken into account; this is done by estimating the impact of ports, shipping routes and distance between markets. A core role is played by their distance to centres of silver and gold mining industries such as Kutna Hora in Bohemia or Freiberg in Saxony. Estimates of the bullion output of mining regions exist for the time from c. 1470. Among institutional variables that need to be considered are membership in currency unions such as the 'Wendish Monetary Union' and urban autonomy in questions of monetary policy. Such institutional factors can be captured by introducing dummy variables into the analysis.
1 Reed (1973) found that in early modern Europe, a reduction in transaction costs had a much stronger impact on growth than a fall in production costs of a similar magnitude.
2 Gold may have suffered less from wear and tear than silver. The hardness of both metals is about the same (2.5-3), but as the purchasing power of gold was higher, gold coins circulated slower. On the other hand, silver was more often alloyed to a higher degree with base metals, a practice which increased the hardness of silver money.