Lending to the Borrower from Hell: Debt, Taxes, and Default in the Age of Philip II (The Princeton Economic History of the Western World)

by Mauricio Drelichman

  By effectively selling “shares” in loans made to the king of Spain, Genoese bankers could achieve a dual objective. They continued to monopolize access to the short-term lending market. This was necessary for lending to be sustainable, as we argued in chapter 5. It also cannot have been bad for profitability. At the same time, selling on parts of the loans reduced the principal lender’s risk. Securitization was thus remarkably successful: it provided funds to the Spanish monarchy at the height of its powers, and the system weathered the effect of temporary, negative shocks such as the 1596 bankruptcy.

  This chapter is based on Drelichman and Voth 2011b, 2012.

  1 AGS, Contadurías Generales, Legajo 92. “Los dichos Francisco y Pedro de Maluenda. Asiento tomado con ellos en 13 de julio de 1595 sobre 439,500 ducados que han de proveer en Lisboa.”

  2 The summary on the front page of the contract describes the principal as consisting of 349,500 ducats. These small discrepancies, in all likelihood introduced for rounding convenience, are not uncommon. The relevant amounts, which we use throughout our empirical exercises, are those in the specific clauses.

  3 Bankers could request juros yielding a maximum of 7.14 percent. Under our discount rate assumption (also 7.14 percent), the present value of lifetime juros of any allowed yield would have been lower than their face value.

  4 For a full description of all the assumptions we used, see the appendix in Drelichman and Voth 2011b.

  5 In chapter 4, we used the MIRR as a measure of asiento rates of return.

  6 The obvious choice for a reinvestment rate is the juro yield of 7.14 percent. Juros were relatively safe investments that could be traded on a fairly liquid market. Bankers could possibly do better, in which case our estimate produces a lower bound on profitability. Specifying the finance rate is trickier. We bias the results against finding profitability by specifying the finance rate at 5 percent for our benchmark estimates. This was the lowest yield of any juro that was not part of a forced conversion and clearly below the average yield of long-term debt. We also conduct sensitivity analysis by lowering the finance rate all the way to zero. Since intermediate negative cash flows are substantially smaller than intermediate positive ones, the impact of any finance rate assumption will be limited.

  7 Note that it is not realistic to assume that any one banker could have earned a return as low as the one implied by this scenario; the banker would in all likelihood not have lent again after 1575.

  8 We use the dates of arrival of the fleets in Morineau 1985.

  9 Because the clause structure in this particular contract is detailed, it is possible to calculate that the outstanding capital at the time of the default was 177,000 ducats. Its MIRR therefore would have been −4.6 percent. Few contracts contain similar detail. Thus we apply the “capital amortization first” methodology uniformly.

  10 While the Maluenda brothers lost money on this particular contract, their overall relationship with the king was profitable. They lent over 4.3 million ducats to Philip II, realizing a MIRR of 20.6 percent after taking into account the effects of the defaults.

  11 We describe the terms of each medio general in chapter 4.

  12 Both quantile and robust regressions show an excess return of 2.5 to 3.2 percent for contracts that rescheduled earlier obligations (t-statistics 1.5 and 1.6, respectively—marginally below the level required for significance at the 10 percent level).

  13 This validates our choice of the juro rate as an upper bound for the opportunity cost of funds.

  14 Note that families that were not affected by the defaults have the same rate of return under each of the three scenarios.

  15 We calculate profitability using the net disbursements as weights for each individual contract. Curiel de la Torre’s disbursements were timed to coincide with repayments from the king. Even though the contracts were nominally for large amounts, Curiel de la Torre’s actual net exposure was low and hence his returns on capital at risk were high. The effect is particularly noticeable because he did not lend large amounts.

  16 The MIRR assumes that all intermediate positive cash flows are reinvested at the exogenously assumed reinvestment rate until the end of the contract. For long loans, this biases the estimated profitability toward the reinvestment rate. The PI is independent of loan maturity.

  17 If we control for other characteristics in a multivariate setting, this premium rises to 6.5 to 9.3 percent (Drelichman and Voth 2011b).

  18 The regression exercise does not allow us to determine if the higher returns for Spaniards and Genoese attest to their “insider” role, or reflect the fact that German lenders only participated in particularly safe loans.

  19 Allen Berger and Gregory Udell (1995) and Mitchell Petersen and Raghuram Rajan (1994, 1995) find R-squared values of around 0.06 to 0.15 in modern-day data. See also Eugene White 2001a.

  20 This was calculated by adding up the disbursements that actually took place in the context of each contract and assigning the total disbursed amount to the year in which the contract was signed.

  21 When a cash disbursement at the court was requested, the specific language was “en esta corte en reales de contado.” Because bankers or their agents resided at and collected their payments whenever the court was stationed, this type of transaction would have carried the lowest transaction costs. When a disbursement was needed at a payments fair, either in Castile or abroad, the language was “en feria de [specific fair], en banco con cinco al millar”—that is, as a bank draft with a five per thousand surcharge.

  22 The standard language is that the king will pay the bankers for “hasta lo que se acostumbra entre hombres de negocios.”

  23 This would only be problematic for our results if the bankers systematically undercharged the king for these services. There is no evidence to suggest this.

  24 We do not know to what extent bankers sold the loans that they issued. If they sold the loans at lower rates of interest, their profits would have been higher still. In the absence of reliable information, we refrain from speculation.

  25 These contracts are located in the AGS, Contadurías Generales, Legajo 92.

  26 We report cumulative sums actually disbursed by the Spinola and De Negro families rather than contracted ones. For example, in the second contract we examine in this section, Spinola and De Negro agreed to lend over 1 million ducats, but only disbursed 127,000 before the payment stop of 1596 put a premature end to the contract. We use the latter amount in our calculations.

  27 It was not unusual for disbursements and payments to predate the actual signing of an asiento. The contracts carried the date on which they were signed by the king. Bankers and royal officials, however, might have come to an agreement weeks or months earlier, and several of the promised cash flows might have already happened by the time the documents were formally signed.

  28 As discussed in chapter 3, the three graces were three income streams (cruzada, subsidio, and excusado) that the church collected on behalf of the Crown and forwarded to the royal treasury. The servicios were direct taxes approved by the Cortes.

  29 Since this transaction would have taken place in March 1596, the bankers would have collected the entire yearly interest of the new juro in November 1596, although they would have held it only for nine months. This would have increased the present value of the operation from 485 to 502 ducats. The small amount of this transaction suggests that the bankers already had a nonperforming juro worth exactly 485 ducats in hand and took advantage of this asiento to dispose of it.

  30 For accounting purposes, juros were valued as perpetual streams. Lifetime bonds, however, stopped performing at the death of their holders and hence had a lower present value than perpetuities.

  31 This would have likely yielded between 1 and 2 percent of the 60,000 ducats under license, and hence perhaps enhanced the overall profitability of the contract to the order of 0.2 to 0.4 percent annually, depending on the timing and actual yield of the transaction. Since the actual return would ha
ve depended on the relative conditions of the Spanish and Italian money markets, which we do not observe, we refrain from including this additional profit in our calculations.

  32 We calculate the profitability of asientos using the MIRR, with a finance rate of 5 percent and reinvestment rate of 7.14 percent. For a detailed discussion of the properties of the MIRR and justification of our choice of parameters, see Drelichman and Voth 2011b.

  33 Because the MIRR incorporates the opportunity cost of funds, its value depends on the terminal date of the contract. We use the terminal date originally specified in the contract to calculate the losses sustained in the restructurings. The reason is that the bankers expected to have their funds tied up until that time, and would have made their original investment decisions based on that terminal date. This also ensures comparability between the expected and actual rates of return.

  34 The additional 245 ecus are a rounding error due to the specific unit of account used.

  35 Quiroga y Vela was a towering figure in the Spanish ecclesiastical hierarchy. He held its two most coveted posts—those of inquisitor general and archbishop of Toledo. He enjoyed large rents and possessions, many of which reverted to the Crown on his death in November 1594. This contract shows that the Crown did not transfer these assets entirely to the new archbishop but instead chose to use part of them to satisfy its financial obligations.

  36 This contract illustrates how both parties to the contract bore risks. The bankers were cash flow positive for ten months, as they gradually transferred to Flanders the large sum that the king had given them up front. Had Spinola and De Negro gone bankrupt, the king would have lost money.

  37 The king had received net inflows for 3.1 million ducats between January and October 1596. In November and December, he was expected to have net outflows of 1.5 million ducats. The payment stop, declared just before the end of November, froze these disbursements.

  38 For example, on July 1, 1572, the king entered into a contract with Pablo de Grimaldo for 800,000 ducats, to be delivered in October 1573. The agreed-on repayment structure shows that the king was to make the first repayment of 125,000 ducats in July 1573, three months before the banker made his initial disbursement. This practice was not uncommon in large contracts, particularly those involving international transfers.

  39 ADG, Inventario Doria 193. For a detailed study of this book, see Felloni 1978. Our description closely follows his account.

  40 When referring to bankers based in Genoa, we use the Italian spelling of their names.

  41 The asientos at Simancas only allow us to observe the promised cash flows, not the actual ones. The Pichenotti–Di Negro account book thus provides a rare window into what actually transpired after the contracts had been signed. This example makes it clear that deviations from the letter of the contracts did not always harm the bankers. The early arrival of the fleet meant that they collected a portion of their debts ahead of schedule, hence mitigating the impact of the bankruptcy.

  42 In the Pichenotti–Di Negro account book, the ecus are valued at the exchange rate agreed to between the king and Madrid bankers rather than at their metallic content. This suggests that the Madrid bankers did not pass through the profits obtained in the exchange operation.

  43 To obtain conservative estimates, we assume the intermediation fee was front loaded.

  44 “Pues la una le yes la palbra de v.m. y me la dio de que esto se haria assi conmigo por mis muchos servicios y los que tengo de hazer.” Instituto Valencia de Don Juan, Envío 22, Caja 33, TB 144.

  45 “Oltre la rovina degli Assentisti, hanosi questi ritirato a dietro molti, che gli soccorevano di rivelantissime partite, e fra gli uni, e gli altri, sono restate esterminate molte ricche famiglie, e molte Vedove, e pupilli insiememente ridotti a miserabile povertà.”

  46 ADG, Inventario Doria 192.

  47 This is consistent with the society’s book, which records the final settlement in 1600.

  48 This contract is identified as the assiento del millione—a common name given to contracts for 1,000,000 ducats or ecus. Since there were four different asientos for that amount open at the time of the 1596 suspension, it is not possible to identify the exact one that Di Negro had invested in.

  49 ADG, Ambrogio Di Negro, Libro Mastro, Inventario Doria 342.

  50 To compare like with like, we should remember that the plotted interest rates are gross rates of return, prior to any haircuts, defaults, or restructurings. Since most were arguably safer than lending to the king of Castile, we also give the net rate of return on loans to Philip II. Even with this standard of comparison, business with Madrid appears an attractive proposition.

  51 A full comparison would have to take into account that Castile’s juros were not defaulted on at all in the sixteenth century.

  CHAPTER 7

  RISK SHARING WITH THE MONARCH

  In early October 1591, lookouts near Cadiz could see the sails of the Spanish treasure fleet on the horizon. After crossing the Atlantic from Havana, the galleons’ final leg of their journey saw them sailing up the Guadalquivir River. They put into harbor under the walls of Seville and unloaded their well-guarded cargo: many tons of silver, mined in modern-day Bolivia using the latest chemical processes as well as forced labor. At the Casa de la Contratación, the value of imports was assessed in detail; one-fifth had to be paid as tax.

  The previous year, two rich fleets had reached Spain, bringing up to 2.5 million ducats with them (Morineau 1985). Whenever the fleets did not sail or came in with smaller silver cargoes than expected, the royal income was also lower; the king, strapped for cash, had to take even more short-term loans from his Genoese bankers. In some years, much more silver came in than anyone expected. In 1595, for example, the fleet arriving in March carried more silver and gold than the Casa de la Contratación could store. The overflow had to be stockpiled in the courtyard (Deforneaux 1979). Nor could anyone predict the fleet’s date of arrival with any accuracy; news traveled with the same ships that brought the silver from the Americas.1 Also, prior to Piet Heyn’s successful raid in 1628, pirates never captured an entire Spanish treasure fleet. The principal determinant of the size and timing of the galleons’ arrival was weather in the Atlantic in the summer and fall.

  One person who waited for the news of the fleet’s arrival in 1591 with particular interest was Tomás Fiesco. A Genoese banker who would eventually rise to the role of factor general, he had agreed to provide 300,000 ecus (some 293,000 ducats) to Philip II.2 These funds were to be paid out to the military commander in Flanders, the Duke of Parma. The king had promised to repay a substantial part of the loan with the proceeds of the 1591 silver fleet.

  FIGURE 22. Frequency of fleet arrivals, 1587–1600. Source: Morineau 1985.

  Typically, Spanish silver fleets arrived in September or October, as shown in figure 22. The contract specified that one month after the convoy’s arrival, the king had to repay the loan. If he failed to do so within thirty days, the banker could stop all future disbursements. The king had the right to delay payment if the fleet did not reach Seville by October in exchange for a penalty of 1 percent additional interest per month. Eventually, the banker could demand repayment in the form of long-dated juros.

  In this chapter, we ask what the sighting of the silver fleet off Cadiz and the king’s contract with Fiesco can teach us about history of sovereign debt in early modern Europe, and about the nature of cross-border lending more generally.

  THE PROBLEM OF NONCONTINGENT DEBT

  Philip II and his bankers effectively solved a problem that defeats bonds market today: how to link repayment terms for sovereigns to a borrower’s financial situation while avoiding perverse incentives. We first summarize the nature of the concern in order to appreciate how remarkable this sixteenth-century feat of “financial engineering” was. We then show how the contractual structures worked and what they can tell us about cross-border lending in the age of monarchs.

  The issuance of noncontingent sove
reign debt can be destabilizing. It requires procyclical fiscal policies, which aggravate recessions (Eichengreen 2002). In the extreme, outstanding noncontingent debts can no longer be serviced in bad times. After defaults, GDP typically falls, trade plummets, and banking systems have to be recapitalized (Eaton and Fernandez 1995; Rose 2005). Economists and policymakers alike have argued that the issuance of debt indexed to GDP (or export prices) could reduce the risk of bankruptcy and smooth consumption (Borensztein and Mauro 2004; Borensztein et al. 2004; Kletzer, Newbery, and Wright 1992). While state-contingent debt is conceptually attractive, few GDP-indexed bonds have actually been issued (Griffith-Jones and Sharma 2006).3 Most instances—such as Argentina’s and Greece’s GDP-linked bonds—occurred in the aftermath of defaults.4 Overall, there is substantial skepticism that the problems with governments issuing state-contingent debt can be overcome.

  Why did a sixteenth-century monarch and his financiers succeed where modern states and investment banks fail? We argue that two factors were key. First, the king’s need to spend ahead of revenue was particularly large. Expenditure—dominated by war financing—fluctuated wildly from year to year; revenue was broadly stable.5 The need for intertemporal barter (Kletzer and Wright 2000) was acute. In this environment, the principal risk for the monarch was a shortfall of liquidity. Risk sharing was especially valuable, and negative shocks were most likely to be temporary.6 Second, the sixteenth-century fiscal environment generated easily observable, verifiable state variables reflecting the strength of the monarch’s finances. The arrival and size of silver fleets from the Americas as well as the yield from individual tax streams controlled by third parties, for instance, served as a ready reference for payments due.