In sum, we find that sixteenth-century Spanish defaults occurred in bad times; they did not adversely influence the pricing of loans, and the Crown’s ability to borrow did not suffer. Combined with the evidence on the extensive use of contingency clauses, this makes it likely that the defaults were excusable in Grossman and Van Huyck’s (1988) sense.
THE TIMING OF DEFAULT
Why did Philip II default when he did? There can be no question that he suspended payments when revenue was unusually low and expenditure was high. Here we examine an additional factor that entered the king’s calculus: the structure of maturing debts. So far, the implication of our reasoning was that Spain simply could not pay its debts at a particular moment because of a combination of pressing spending needs and revenue shortfalls. Before such a toxic blend leads to a default, though, one more problem typically needs to materialize: an inability to roll over maturing debts. We do not know with certainty that this contributed to the timing of the 1575 default, but the archival record provides some tantalizing hints that point in this direction.
Few borrowers can cope with a sudden need to repay maturing debt instead of merely paying off old creditors with new borrowings. Spain in the age of Philip II was no different. In a classic paper, Kehoe and Cole (2000) argue that rollover crises will be more likely the shorter the maturity profile of debt is. A short maturity profile raises the risk that repayments will be high in the absence of new debt contracts being signed. Interestingly, the borrower’s optimal decision in this setting depends on the possibility of rolling over the debt. If the market offers a chance to repay old bonds with new borrowing, it is typically best to repay. In contrast, if the maturing debt needs to be paid by reducing expenditure or increasing taxes, the cost may be too high. Cole and Kehoe (1996) point out that Mexico’s crisis in 1994–95 occurred with relatively low outstanding debt, but after a massive shortening of the maturity profile of the debt outstanding. If all loans had been contracted with a single lender, that lender would always roll over the debt to avoid the cost of default; a larger group of lenders cannot easily coordinate to the good equilibrium, since it is individually optimal to “run” on the sovereign and demand repayment.
We now show that this logic at least partly applied to Spain’s famous default of 1575. As a result of reconstructing all cash flows, we have a good idea of how high the repayment burden implied by scheduled payments was on the eve of the decree—and how much worse things became as few contracts were signed as the debt burden mounted. Table 26 shows the pattern of inflows and outflows per year as specified in the loan covenants:
Table 26. Short-term debt inflows and outflows (in ducats)
Positive and negative net flows alternated over time, as expenditures and tax revenues ebbed and flowed. In a growing economy and with rising tax revenue, this does not indicate a fiscal problem. The balance remained positive for the first eight months of 1575. For the last three months of the year, however, the outflows would have been massively negative. The total repayments implied by the signed contracts amounted to 4.4 million ducats; at the same time, there were contractual arrangements for inflows to the tune of more than 2 million ducats. This was a much faster rate of net outflow than in 1574, when the entire year saw a liquidity drain from asiento borrowing of 2.9 million.
The fact that lots of debt was maturing itself offers only limited information about the nature of the 1575 crisis. Earlier years had seen sustained periods when the gross monthly liquidity drain was above 0.5 million ducats—but matching inflows had reduced the net figure substantially. In the last quarter of 1575, there were few commitments for fresh funds. This suggests that some bankers at least had begun to suspect that a decree suspending payments was in the offing. Indeed, in an undated memorandum of one royal official, Francisco Gutiérrez de Cuéllar, there is a hint of a building crisis of confidence among the bankers during the run-up to the default. The memorandum observes that there is “gran rumor” and growing concern about a coming suspension, and that Nicolao de Grimaldo and Lorenzo Spinola in particular had begun preparations in advance of a default.30
Gutiérrez himself apparently only learned about the decree later than even the bankers concerned—a sign of the great secrecy surrounding the decision, which was largely taken by the king himself with only minimal advice from officials (Lovett 1980). It also shows that some information had begun to leak out. Plans about how to solve the problem of the floating debt had been circulating since the 1573 discussions about an increase in the sales tax, and many of them included measures against the bankers. While bankers were willingly rolling over maturing debt, these plans came to naught; once some of the better-informed bankers like Lorenzo sought to salvage their own position, the cost of paying out maturing contracts at a time of general cash shortage became too big—a classic rollover crisis created by a scramble for the exit by the biggest and most-informed lenders.
If our interpretation is correct, one question arises: Why could bankers coordinate during defaults (as argued in chapter 5), but not in the run-up to a default? We argue that the structure of incentives was different. Lending during a moratorium would be easily observed and could possibly lead to a permanent ostracism in the community of Genoese lenders. Failing to lend, on the other hand, may have had many reasons, including a lack of available funds. Since lenders did not have to step forward on a fixed schedule, the failure to join in a common enterprise was less visible, and at the best of times, coordination was more indirect. Finally, the balance of risks and returns was arguably different; lending in the run-up to a default offered only average returns in exchange for high risks. Lending during a default combined sky-high risks with substantial rates of return. Lenders could be perfectly rational in having an appetite for one but not the other combination of risks and returns.
CONCLUSION
Over the last eight hundred years, many periods of debt accumulation have been followed by default (Reinhart and Rogoff 2009). Despite these disruptions, the market for sovereign debt did not disappear. What accounts for this resilience? We argue that at least in the case of asiento lending to Philip II, excusable defaults were an important factor. Studying the loan documents directly, we show that a significant share of short-term loans contained contingency clauses. We analyze the different types of loan modifications along with their impact on cash flows and loan maturity. These modifications allowed effective risk sharing between king and bankers—an institutional solution that offered many of the desirable properties that contingent debt would have today (Borensztein and Mauro 2004).
Genoese bankers effectively offered insurance to the king. They charged relatively high interest rates in normal times, but offered a service that the king valued when times were bad: de facto insurance. Loan contracts often included clauses that modified loan terms if financial conditions worsened in an unforeseen manner. One contingency that featured prominently in the documentation for short-term loans was fleet arrival. If the king’s ships put into harbor later than expected (or not at all in a single year), then the repayment clauses could automatically be adjusted. Similar clauses covered low tax receipts and the like. In addition, there were numerous possibilities to extend payment horizons at the king’s discretion.
The system of state financing under Philip II was not without shortcomings. The 1575 bankruptcy was probably caused by a combination of negative fiscal shocks—high spending coupled with low revenue—and a “run” by bankers on the treasury. There is some indirect evidence that financiers stopped offering fresh funds as the Crown’s financial situation deteriorated. The short maturity profile of asiento borrowing then made it optimal for the king to stop servicing his debts, since a high share of outstanding debts would have to be redeemed from other sources (instead of simply rolled over). In this sense, a combination of coordination failure and a negative liquidity shock may well account for the timing of the 1575 default.
The fact that an early modern monarch and his financiers could effectively w
rite state-contingent debt contracts—providing insurance—is remarkable in its own right. This insight can also inform our understanding of Philip’s famous defaults. In all likelihood, they were excusable.31 They occurred in a context that involved loans that were individually rescheduled with some frequency, they happened in verifiably bad states of the world, and they reflected events beyond the control of the sovereign. The early Spanish defaults were far from synonymous with a one-sided abrogation of contracts, as Reinhart and Rogoff (2009) imply.32 Instead, the defaults healed a form of market incompleteness: not all possible states of the world can be contracted on. When some unspecified—or perhaps unspecifiable—contingency arises, it is rational for the borrower to default. Since lenders realize this before the fact, the default—while violating the letter of the contract—does not go against the spirit of the original agreement. Payment stops thus are not followed by higher interest rates or lengthy exclusion from loan markets. “Business as usual” should resume quickly once the temporary, adverse shock that caused the default in the first place has dissipated. This is what happened in the case of both the 1575 and 1596 bankruptcies.
1 Sometimes a faster ship (a frigate or sloop) would sail ahead of the main fleet and inform the authorities in Seville of its imminent arrival (Perez-Mallaina 1998).
2 The factor general was in charge of raising loans to meet the king’s financing needs and often provided part of those loans themselves.
3 Some scholars have argued that all sovereign debt is de facto contingent (Grossman and Van Huyck 1988).
4 Greece avoided an outright default through a “voluntary” restructuring.
5 Silver revenues were highly volatile; we describe them below.
6 Elsewhere, we have shown that Philip II’s famous defaults do not reflect insolvency but instead were caused by liquidity crises (Drelichman and Voth 2010).
7 We were not able to do so for 41 contracts. This is mostly due to material damage to the documents. In a few cases, the clauses were too vague to allow an accurate estimation of cash flows and rates of return.
8 This asiento also shows how loans were combined with transfers. The bankers first disbursed 38,000 ducats, the king next repaid the principal and interest on 95,000 ducats, and only afterward did the bankers disburse the final 57,000 ducats in Flanders. Therefore, the latter disbursement is a transfer rather than a loan, since the bankers had already received the money from the king.
9 The deeper reason for collateralizing with juros is that fiscal centralization in Castile was limited—the king could sometimes not pay the bankers directly, but the city of Seville, say, would still pay holders of juros. Hence, the fragmentation of fiscal authority facilitated the continuation of lending. The incentives that supported the collateralization of one type of debt instrument with another are similar to those in Broner, Martin, and Ventura 2010.
10 As mentioned earlier, the excusado is a tax levied on church revenue, one of the so-called three graces, introduced in 1567.
11 This example also shows the importance of weak tax-collecting powers in determining lending arrangements, with the king effectively outsourcing the right to access taxes already collected.
12 AGS, Contadurias Generales, Legajo 90.
13 By comparison, the median return of loans that do not have contingent clauses is 19 percent.
14 The increase is a mere 0.4 percent; the difference in cost compared to the baseline is not statistically significant.
15 The king, however, could manipulate the order in which lenders were paid. Contracts were therefore quite specific in establishing the collection priority of individual lenders with respect to specific tax revenues.
16 Juro interest was paid twice yearly. If the banker received juros in October, he would be allowed to collect the entire December interest payment rather than the portion corresponding to the three months he had held the bond. This increases the profitability of the contract relative to a cash payout in December.
17 The maturity of the loan for the king changed differently from the one for the bankers. The king would now only repay through tax revenue foregone, which means at a slow pace.
18 The exception was the 1576–80 period, when most lending came from the 1577 settlement.
19 Shocks arising from military defeat are an obvious case in point. It would be hard, for instance, for the Crown to contract on the possibility of the Armada sinking.
20 Strictly speaking, in chapter 5 we surmised that defaults were excusable; here, we actually argue the case based on a closer reading of loan and fiscal conditions.
21 If there had been changes in the actual loan conditions, these could also be rationalized by Bayesian updating (about, say, the strength of the Spanish navy). In this sense, demonstrating that rates did not change requires the “strong” version of our hypothesis to hold.
22 For a detailed analysis of Philip’s military strategy, see Parker 1998.
23 All figures are from the analysis in chapter 4 and the appendix in Drelichman and Voth 2010.
24 As Alfaro and Kanczuk (2005) argue, rising interest rates after a default act as a punishment for borrowers who violated the original loan contract in a context of contingent lending.
25 We obtain a value of 0.68 (p-value 0.49).
26 A t-test has a value of 0.48 (p-value 0.63).
27 Our data for these regressions end in 1596, as no fiscal indicators are available after that date. As a result, we are unable to explore the effects of the 1596 bankruptcy.
28 These data are discussed in chapter 4.
29 It should be noted that neither the debt-to-revenue ratio nor the fiscal balance were readily available statistics. Estimating the latter is substantially more difficult.
30 Instituto de Valencia de Don Juan, E22, C33, TB, 142. Lovett (1980) argues that Spinola even sold juros de resguardo in the run-up to the crisis.
31 In the sense of Grossman and Van Huyck (1988).
32 Reinhart and Rogoff (ibid., 86) treat early French and Spanish defaults as equivalent, and use the terms abrogate and default interchangeably.
CHAPTER 8
TAX, EMPIRE, AND THE LOGIC OF SPANISH DECLINE
Non sufficit orbis—the world is not enough. When Sir Francis Drake’s men stormed the Spanish governor’s palace in Santo Domingo in 1586, they found a coat of arms displaying a map of the world. It was adorned with a horse rising triumphantly on its hind legs; the Latin motto was prominently displayed above.1 What sounds like hubris to modern ears (and will remind readers of a James Bond film) was not an exaggeration for contemporaries: in the days of Philip II, Spain’s empire had no equal. After the takeover of Portugal, Philip II ruled virtually all of the Americas, numerous trading posts along the African coast and in India, the Philippines, and a string of possessions in Italy and the Low Countries. Philip II’s Portuguese ships dominated the spice trade and Indian Ocean; Spanish ships led the fleet that defeated the seemingly unstoppable Ottoman expansion at the Battle of Lepanto; Spanish armies won glittering victories against France. Spain’s armed forces, numbering 163,000 troops, were three times larger than that of the next European power (Karaman and Pamuk 2010). No other power in the world controlled a territory as large, armed forces as powerful, or financial resources as vast as those of Habsburg Spain.
And yet within little more than a century, Spain’s empire was a shadow of its former self. By 1700 already, its armed forces numbered no more than 63,000 troops—a third of their former size. France, in contrast, had raised the number of soldiers under arms from 57,000 to 342,000—from one-third of the Spanish figure to a number almost seven times greater. Britain was not far behind, increasing the size of its army and navy more than threefold, from 66,000 to 191,000 (ibid.). While still controlling vast territories, Spain’s standing had taken a beating. Portugal broke away in 1640; northern Catalonia and the Franche-Comté had to be ceded to France as a result of peace treaties in 1659 and 1678. Rebellions in Catalonia and Sicily along with conspiraci
es against the Crown in Andalusia and Aragon underlined the imperial center’s fragile grip on its possessions in Europe and the Iberian Peninsula. Gone were the trading outposts in India and control of the spice trade.
Table 27. Europe’s great powers: Measures of might
Source: Karaman and Pamuk 2010.
* in thousands
** in tons of silver per year
Revenues had declined—from more than four hundred tons of silver per year in the heyday of imperial Spain to barely half that figure. Other powers had succeeded in raising more revenue. England, where the total tax income was a mere 15 percent of the Spanish figure in 1600, raised revenue almost tenfold. It now commanded financial resources that were 150 percent greater than Spain’s. France similarly went from 68 to 400 percent of Spanish tax revenue.
There can be no doubt that by 1800, in the European concert of powers, Spain had become a bit player; it “failed” where England succeeded. England came to control dominions every bit as vast as those held by Philip II; its frigates and ships of the line ruled the waves more absolutely than Spanish galleys and galleons ever had.
Two of the great empires during the early modern period ended up on radically different trajectories. Spain under the Habsburgs, in the sixteenth and seventeenth centuries, is today a byword for poor governance, profligacy, economic stagnation, and military decline, from which Spain even now has not fully recovered. Eighteenth- and nineteenth-century Britain serves as a paragon of good institutions, fiscal probity, economic growth, and military prowess.
One of the main factors underlying Spain’s poor performance as a great power was economic decline. Since Earl Hamilton’s (1938) famous essay, many studies have examined and documented the slowdown after 1600. What was a vigorously expanding economy in the fifteenth and sixteenth centuries, with rising per capita incomes and expanding population, first stagnated and then began to shrink. Hamilton attributed much of the decline to monetary mismanagement following the influx of American silver. Elliott (1961) later reexamined the evidence, and found it overwhelmingly in favor of eventual decline—but stressed that Castile, not Spain as a whole, was its principal victim.2
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 29