یکپارچگی بازار مالی در دوره مدرن اولیه در اسپانیا: نتایج از مدل تصحیح خطای آستانه

The application of a threshold error correction model to the exchange rates of the Spanish Ducado and the Dutch Groat in Seville and Medina del Campo in the 16th century indicates that the band of arbitrage inactivity was 6%.

The silver and gold currencies offered arbitrage opportunities between the markets for silver and gold as well as foreign exchange. However, transaction costs, which may have been rather substantial in the past, hindered arbitrage and led to a band of “arbitrage inactivity” for the exchange rate around its par value. This issue was analyzed empirically in some recent papers using threshold autoregressive or error correction models [Prakash and Taylor, 1997, Canjels et al., 2004, Volckert and Wolf, 2006 and Kugler, 2009]. Prakash and Taylor (1997, Table 5) report that in subperiods of the classic gold standard deviations from par value of the Dollar/Pound rate were possible only in the range of 0.15–0.46% until arbitrage operations were triggered. This finding is not surprising since the introduction of the telegraph, the steamship and the railroad as well as improved safety of transport routes decreased significantly the transaction costs for exploiting arbitrage by shipping precious metals during the 19th century. Volckert and Wolf (2006) find larger bands of arbitrage inactivity in the range of 0.34 to 0.99% for the silver currencies of Flanders, Lübeck and Prussia in the years 1385 to 1450. This difference is surprisingly small since one would expect that the information and transport revolution in the second half of the 19th century led to a huge decrease in transaction costs. However, the application of a threshold error correction model to data relating the Basle Pound and the Rhinegulden and the market for gold and silver for the period 1365–1429 indicates that transaction costs prevented arbitrage when the difference of the gold–silver ratio and the real exchange rate was within a 7% band (Kugler, 2009). Such a large band of “arbitrage inactivity” is supported by information on other indicators for transaction costs, namely the up to 5% fees for money transfer by bills of exchange as well as the up to 10% difference between deposit and loan rates of banks in medieval and early modern times (Homer and Sylla, 2005, 76, 136). Nevertheless, it seems of some interest to explore financial market integration with other historical data sets. In this paper we consider the integration of the markets between the Spanish Ducado and the Dutch Groat in two Spanish towns, namely Seville and Medina del Campo. We use annual data from 1564 to 1603 and apply a threshold vector error correction model.

This paper carries out an empirical analyis of the dynamic adjustment of the exchange rate between the Spanish Ducado and the Dutch Groat in Seville and Medina del Campo. The application of a threshold vector error correction model to annual data from 1564 to 1603 provides the result that deviations of up to 6% between the two market places were possible without leading to arbitrage operations. Larger deviations were offset by an adjustment of the exchange rate at Medina del Campo within a year. By contrast to Volckert and Wolf (2006) concerning the currencies of Flanders, Lübeck and Prussia in the years 1385 to 1450 this result supports the view of relatively large transaction costs in late medieval and early modern times: the analysis of the exchange rate of the Basle pound and the Rhinegulden in the 14th and 15th century led to a similar threshold estimate, namely approximately 7% (Kugler, 2009). Our results necessarily suggest the question why the adaptation of the two exchange rates took only place in Medina del Campo. This is probably brought about by two facts: First, that in Seville the silver market was very liquid given the large inflow of this metal from Latin America; and second that the costs of transportation by sea compared to those on land (as at least partly necessary to Medina del Campo) were much lower and less time-consuming. Therefore, we have to expect a faster adjustment of the relative prices of silver and gold in Seville. As a consequence, this market experienced relative changes of gold and silver prices first which were only later transmitted to Medina del Campo provided that the deviation was large enough to cover the transaction costs.