SOVEREIGN LENDING SUSTAINED: MARKET POWER AND COHESION IN TIMES OF CRISIS
Philip II of Spain accumulated towering debts, and four times he stopped all payments to his bankers. Nonetheless, he continued to retain access to fresh loans for most of his reign—in the overwhelming majority of cases, from the same lenders he defaulted on. What explains this puzzling symbiosis between the first serial defaulter in history and his financiers?
Using our new data set collected from the General Archive of Simancas, we document a unique way in which Philip II’s Genoese bankers overcame enforcement and collective action problems: lending in overlapping syndicates. Effective coordination between lenders gave this coalition substantial market power vis-à-vis the king. Because of his enormous borrowing needs, the king’s demand for loans could not be satisfied by any other group of lenders; in effect, Philip II had access to only a few lenders, who acted in unison. Networks based on social affiliations can solve the problem of incomplete contracting in a variety of contexts, creating in effect a private-order institution (Greif 2006). Examples include contract enforcement among diamond dealers and securities traders, merchants engaged in long-distance trade, as well as the running of common resources.52 In the case of Philip II, the largest and most important bankers acted as if they were a single financial entity—a lenders’ coalition. Ultimately, cheat-the-cheater incentives (Kletzer and Wright 2000) ensured that a simple lending moratorium of the Genoese was sufficient to force a powerful monarch like Philip II to pay his debts.
The crucial test for our hypothesis comes during the default of 1575. No outside power could come to aid of the Genoese, as the US Marine Corps had done in Santo Domingo in 1904 on behalf of European creditors; Philip II commanded the military of the world’s only superpower at the time. Lenders had few ways to sanction Philip II by means other than a stop to lending. They attempted to impose a transfer stop that would have cut off funding for the troops in Flanders. We document that this penalty was ineffective; ample funds still flowed there. Crucially, Spain’s major setback in the Netherlands in 1576 was not driven by a funding crisis but instead by the volatile politics of the time. The Fugger and other bankers continued to transfer funds for the Spanish sovereign, and enough money was made available in the Netherlands to pay off the mutineers. The case of Philip II, then, cannot be claimed as an example of sanctions in the Bulow and Rogoff style. Banker irrationality or sentiment also played no role in lending to the Spanish monarch. Contrary to the argument made by Braudel (1966), banker turnover was minimal. There was no mass exodus of lenders following the defaults. This suggests that expectations were not massively disappointed by the temporary payment stops and general settlements with bankers.
When the payment stop of 1575 came, neither new nor existing lenders undermined the Genoese lenders’ moratorium. To offer the Crown fresh funds would have been a bad idea for any lender. The king’s borrowing needs were so high that he would eventually have to settle with the Genoese coalition. Because the Genoese acted in unison, any lender who offered funds to Philip II during the moratorium would most likely have been cheated, in line with the predictions of Kletzer and Wright (2000). As a detailed analysis of the correspondence between the Fugger makes apparent, this consideration was certainly important for bankers at the time—and a key reason why the famous Augsburg banking dynasty chose not to offer new loans.
Lending occurred under conditions of anarchy, with neither side being able to make firm, binding commitments. Why established lenders in the Genoese coalition repeatedly agreed to debt reductions and a resumption of lending is also probably best explained by the market power derived from the group’s cohesion (Kovrijnykh and Szentes 2007). This ensured that even after earlier debts had been reduced, future profits would be ample.53 Far from indicating banker irrationality and the significance of lender sentiment, the boom-and-bust cycles of the sixteenth-century Spanish monarchy reflect the efficiency and flexibility of private-order institutional arrangements.
Portions of this chapter are based on Drelichman and Voth 2011a.
1 “Cada día boy descubriendo contra estos jinobeses casos, que si a S. Mg. y a sus ministros no nos combiniese cumplir nuestras palabras, no me bería harto de su sangre.”
2 For example, Braudel (1966) cites the case of Michael Cantacuzenus, who was hanged on the orders of the sultan in 1578 in order to seize his wealth. Cases such as these—similar to the proscriptions that dealt such a blow to Roman aristocracy in the late republic (Mommsen 1881)—became exceedingly rare in western Europe after 1500.
3 In addition, cities in northern France and the Low Countries sold annuities from the thirteenth century onward (Tracy 1985).
4 One (important) caveat is that states have successfully restructured bonds issued under their own national law, but held by foreigners. Greece in 2012 is a case in point.
5 The argument is strengthened if we take into account the diversification benefits of foreign lending (Chabot and Kurz 2010).
6 They also conclude that British investors fared markedly better than US ones, who mainly lent to the countries that experienced trouble in the 1930s. In part, the UK policy of steering investment toward the British Empire helped to avoid the worst losses.
7 Sir John Simon, British field officer, 1934, cited in Eichengreen and Portes 1989a.
8 His data also suggest that much of the trade lost between creditor and debtor countries is compensated for by higher exports to other countries. Trade diversion may have welfare costs, but these are typically too small to explain the scale of output declines found in the literature.
9 Laura Alfaro and Fabio Kanczuk (2005) present a similar finding in their calibration of a contingent debt service model.
10 Scaled by output, the average is lower—around 5 percent—although the peak during the 1930s still approaches 40 percent.
11 Using higher-frequency data, Eduardo Levy-Yeyati and Ugo Panizza (2011) demonstrate that defaults typically occur at the lowest point of output during a recession and that many recoveries start in the next quarter.
12 Kenneth Kletzer (1984) supplies another early example of a reputation-based model. Jonathan Eaton and Raquel Fernandez (1995) provide a survey of this literature.
13 See, among others, Braudel 1966; Thompson 1994a, 1994b; Lovett 1982.
14 Roland Benabou (forthcoming) offers a model of (individually) rational reality denial.
15 Rodríguez-Salgado (1988) provides an account of the 1557 default as well as the run-up to the 1560 one.
16 The standard series in use is by Ulloa (1977). It suffers from the double counting of asientos contracted by field commanders in Flanders, which left most details to be negotiated later in consolidated contracts between the king and bankers’ representatives in Madrid. Our database includes only the final agreements, which superseded those made elsewhere, and fully specified all terms and conditions. For more details, see chapter 3.
17 We discuss these conditional clauses in detail in chapter 7.
18 We exclude the 5 million ducat loan that accompanied the general settlement of 1577 as well as the 7.2 million ducat one that followed the resolution of the 1597 bankruptcy.
19 For bankers residing in Spain, we use the Spanish spelling of the banking families’ names throughout. For those residing abroad, such as the Fugger family, we keep the original-language spelling.
20 AGS, Contadurías Generales, Legajo 85. “Gerónimo Grillo y Esteban Grillo. Traslado del asiento con ellos tomado a 13 de marzo de 1572.”
21 AGS, Contadurías Generales, Legajos 84 and 85.
22 The Fugger family never stopped lending for more than nine consecutive years.
23 Braudel (1966, 362) argued that each bankruptcy ruined another group of lenders. These were then in turn promptly replaced by a new group (and nationality), which, lemming-like, readily extended credit to the king. He also argued that “the Fugger and their acolytes … were to withdraw (apart from brief reappearances in 1575 and 1595) fr
om the dangerous business of the asientos,” and that “the decree of 1st September, 1575, then, was a blow struck at the entire fortunes of the Genoese…. To the Genoese this brought massive losses” (ibid., 351–52, 355). We argue that Braudel is mistaken in both cases.
24 For a complete yearly chart of repeat lending, see the appendix in Drelichman and Voth 2011a.
25 Our coding of the asientos in the archive of Simancas allows us to separate the transfers to Flanders from those to other destinations, which were not part of the penalty suggested by Conklin. For other discussions of the transfers during the bankruptcy years, see Lapeyre 1953, 22; Vázquez de Prada 1962, 330–33; Ulloa 1977, 795–96.
26 In 1569, funds earmarked for the Duke of Alba’s troops were seized by Elizabeth II after the ships had to contend with adverse weather in the English Channel (Lovett 1982).
27 Note that the only evidence for this transfer is Conklin (1998). The source that he cites does not contain information on this particular transaction.
28 Both the transfers to Flanders and cost of the Flanders campaign exclude the cost of settling mutinies. For the data on mutinies and their cost, see Parker 2004; Tenace 1997. The cost of the Flanders campaign is part of the military expenditure series presented in chapter 4. The data on transfers are from our database of asientos.
29 The Sack of Antwerp further enhanced Spain’s Black Legend—the view that Spain acted in both the colonies and Europe with greater moral callousness than other powers.
30 “It was the outbreak of a large-scale mutiny … in conjunction with the increasingly strong desire for peace in the southern Netherlands and Don John’s procrastination in taking up this post as governor-general, which, in July and August 1576, led to a spectacular improvement of the rebel position” (Swart 1978, 25).
31 The Flemish ecu was worth approximately 0.98 ducats at the time. The eventual cost of the settlement was larger, since the mutineers were joined by others and because they received back pay for the months of the mutiny.
32 At various junctures, the king and his advisers considered forced conversions of juros, but ultimately decided against them. Juro interest reduction would eventually take place in the seventeenth century.
33 See AGS, Contadurías Generales, Legajo 84. “Tomás de Marín. Asiento tomado con Pirro Boqui en su nombre.”
34 In this sense, the alternative considered by Bulow and Rogoff (1989) was not available to the king.
35 Whether the Genoese with their high degree of collaboration constituted a cartel has been debated in the historical literature (Alvarez Nogal 2003). We do not take a view on their pricing behavior. We refer to them as a network simply because of their co-lending and behavior during the defaults.
36 See, for example, the Fugger correspondence summarized in Karnehm 2003. See also several pieces of official correspondence in De Carlos Morales 2008.
37 The largest loan by the Fugger was for 1.3 million ducats in 1594—a year in which the silver fleets did not sail.
38 See, for example, AGS, Contadurías Generales, Legajo 85, where several loans made by Lorenzo Spinola are collateralized with bonds held by Nicolao de Grimaldo.
39 AGS, Contadurías Generales, Legajo 85. “Lucián Centurión y Agustín Spinola. Traslado del asiento con ellos tomado a 2 de mayo de 1569.”
40 These bankers were Nicolás and Visconte Cattaneo, Alberto Pinelo, Miguel de Mena, Constantin Gentil, Benito Sabago, and Juan Antonio De Negro. Many of them also lent through syndicated loans with the Spinola and Centurión families.
41 AGS, Contadurías Generales, Legajo 84. “Tomás de Marín. Asiento tomado con Pirro Boqui en su nombre.” We never observe Grimaldo and Marín lending together to the king. Still, they both belonged to the network because they extended loans jointly with other bankers.
42 AGS, Contadurías Generales, Legajo 88. “Agustín Spinola, hijo de Francisco difunto. Asiento tomado con él sobre un millón de ducados que provee en Italia.”
43 AGS, Contadurías Generales, Legajo 88. “Lo que por mi mandado se asienta y concierta con Francisco Spinola genovés sobre 30,000 escudos.”
44 The text of the medio general, for example, specifies that Esteban Lomelín is Nicolao de Grimaldo’s son-in-law. AGS, Consejo y Juntas de Hacienda, Libro 42. Similar family relationships are occasionally mentioned in the text of the asientos.
45 Three families both intermarried with network members with who they co-lent and with other network members who did not feature on the same loan documents.
46 The earlier defaults affected loans by the Fugger and Welser to Charles V. The settlements involved large transfers of physical assets—including mines, land, and tax farms—that are difficult to value. Furthermore, our series of asientos extend back only to 1566.
47 There is one possible exception: the transfer of a hundred thousand ducats to Flanders in 1576. The initial request by Francisco Garnica, one of the king’s officials, was for fifty thousand ducats to be advanced by the Crown with the rest to be paid from the next year’s tax increase. Although such an arrangement would constitute a loan, there is no evidence that the Fugger actually advanced any money.
48 The Fugger had a history of agents actually extending credit to the king of Spain in the 1550s. Some of the loans extended by those agents were caught in the 1556–60 default. After this episode, rules were tightened and agents could no longer lend without authorization from “headquarters” (Ehrenberg 1896).
49 “Du siehst, daß sich von Tag zu Tag die Servitios, so wir dem Künig [von Spanien] thun müeßen hauffen.” … [W]irds ain grosse Notturfft erfordern, dem T[omás] Miller ain Bys einzulegen, wir khummen sonst burlando ins Decret.” Karnehm 2003, 408–9. According to Charles Karnehm (ibid.), “burlare” here means “fraudulently”; an alternative translation is “mockingly.”
50 “Die Sp[ani]er [werden sich] unser zu ewigen Zeitten … bedienen wellen, uns aussaugen, und nött[igen], wan wir dann nit jederzeit thun werden, was Sie wellen, so wirdt man uns das Decret fürwerffen, und sagen, man wöll uns darein schließen und tractieren wie die Genueser, wie dan schon vor Augen.” Letter from Hans Fugger to Marx Fugger, September 5, 1576, cited in Karnehm 2003, 408–9; emphasis added. Nor was the idea that the king would strong-arm the Fugger into ever-more concessions idle speculation. Several transfers from Spain during the crisis were organized by the Fugger agent in Madrid, in direct contravention of the orders from Augsburg—because Miller could see no way to say “no” to the king and his advisers in the hour of need (Ehrenberg 1896).
51 The logic here is related to the argument in Wright 2002.
52 Examples include Wade 1987; Baker 1984; Bernstein 1992; Greif 1993.
53 We examine the profitability of lenders in detail in chapter 6.
CHAPTER 6
SERIAL DEFAULTS, SERIAL PROFITS
The king continued to borrow massively throughout his reign, using the help of bankers to raise short-term financing. The Genoese banking network kept incentives aligned; the king’s best strategy was to service his debts, so that he had access to capital markets in the future. Here we look at the bankers’ side: How much did they profit as a result of “lender power”? We calculate returns on lending to the Castilian crown, taking into account defaults and the bankers’ cost of funds. Our calculations demonstrate that loans to Philip II were highly profitable. Defaults and reschedulings reduced the rate of return, but profitability net of these losses was still high—and markedly higher than the return on alternative investments. This was true on average and also applied to the vast majority of banking dynasties individually. As a consequence, few financiers ever stopped lending to Philip II. Profitable lending also explains why the number of exits from the stage after the 1575 default was no greater than before.
MEASURING RETURNS
To estimate rates of return, we need to reconstruct the cash flows of each contract. The database of asientos described in chapter 3 contains detailed information about the agreed-on cash flows, collateral and
foreign exchange clauses, payment dates, and fringe benefits. These can be used to calculate the rate of return for each lending contract.
We now illustrate this process with an example. The brothers Pedro and Francisco de Maluenda entered into a contract with the king on July 13, 1595.1 They agreed to deliver 349,464 ducats in Lisbon in thirteen payments.2 The first payment, for 26,856 ducats, was due eight days after the contract date. The remaining twelve payments, of 26,884 ducats each, were due at the end of each month, starting in July 1595. The king promised to repay as follows:
• A payment of 75,000 ducats from the general treasury in November 1595
• A payment of 97,000 ducats one month after the arrival of the first treasure fleet
• The amounts in the first two payments would accrue 1 percent monthly (simple, not compounding) interest starting from the month of August
Lending to the Borrower from Hell: Debt, Taxes, and Default in the Age of Philip II (The Princeton Economic History of the Western World) Page 22