This chapter is based on Drelichman and Voth 2010.
1 For an overview of Castile’s justice system, see Kagan 1981.
2 In particular, we rely on Dandelet 1995, 2001; De Lamar 1964, 1988; Koenigsberger 1951; Lynch 1961; Parker 1970, 1977, 1979, 1998, 2004; Tenace 1997, 2003; Thompson 1976, 1992.
3 Magnitudes are given in 1566 constant prices.
4 Because the cash flows are clearly specified, modifying this assumption by either fully front loading or fully back loading the interest does not significantly alter the results.
5 Asiento y Medio General de la Hacienda. AGS, Consejo y Juntas de Hacienda, Libro 42.
6 On this point, see Ruiz Martín 1965.
7 This stands in stark contrast to England’s debt conversions, also studied by Chamley (2011), which proceeded in an orderly fashion.
8 Figure 8 also speaks against the main conclusions of the “serial default” literature. Following the 1575 payment stop, we do not see the downward spiral of weakening fiscal institutions predicted by Carmen Reinhart, Kenneth Rogoff, and Miguel Savastano (2003). Rather, due to fiscal restraint and tax hikes, the primary surplus went up.
9 If we regress pst on MEt, we obtain a coefficient of −1.07 (t-statistic 9.4). This implies that the primary surplus moved almost exactly 1:1 with military spending.
10 Since some debt was issued below par, it is not quite clear which value one should use: the nominal value of debt or the amounts actually raised. If we used the nominal value of the debt, the burden would have exceeded 200 percent of GDP.
11 The only sense in which victory in Flanders could have worsened Philip’s fiscal position would have been a continuation of high-intensity warfare with England. While not impossible, we consider this unlikely.
12 Since there are no good estimates for the total value of debt issued by the rebellious provinces before their uprising, we abstract from the fact that the tax revenue of conquered territories would not have been entirely unencumbered after a Spanish triumph. For example, so-called Rentmeisterbriefe (long-term debts similar to juros) had been issued in Holland and Zeeland. The Fugger held a good number of these—and never received payment for them (Ehrenberg 1896). These debts would have been recognized and serviced after a Spanish victory, reducing the net gain in tax capacity.
13 This is a highly conservative calculation. Dutch revenues increased rapidly in the seventeenth century, averaging 2.6 million ducats per year between 1600 and 1650. This is four times the amount we use in our counterfactual.
CHAPTER 5
LENDING TO THE BORROWER FROM HELL
CREDIT AND THE KING’S SWORD
On Thursday, October 12, 1307, the Grand Master of the Order of Templars, Jacques de Molay, was carrying a heavy burden: he was a pallbearer at the funeral of Catherine of Courtenay, sister-in-law to Philip IV of France. The young noble woman had died at the age of thirty-two. Only nobles in favor at court were accorded the special honor of acting as pallbearers. Also present at the funeral were the king and every member of the royal family as well as most of the leading officials.
The next morning, bailiffs woke de Molay before dawn, bearing a warrant for his arrest. The Templars, or so the king alleged, were heretics who in their initiation rites denied Christ and spat on the cross, and then engaged in homoerotic rituals—“an abominable work, a detestable disgrace, a thing almost inhuman, indeed set apart from all humanity” (Barber 2001). At the same time, hundreds of Knights Templar were arrested all over France. Under torture, the grand master confessed to the Templars’ sins (before recanting). Seven years later, after a drawn-out legal process, he met his end on the Isle des Juifs in Paris, near the garden of the royal palace. He was burned at the stake on March 19, 1314, together with Geoffroi de Charney, the former master of Normandy for the Templars.
There can be no question that the charges against de Molay and the Templars were trumped up. Similar persecutions had already been brought against French Jews (expelled in 1306) and the Lombards. The reason for these judicial murders was the same: the Templars, Jews, and Lombards had lent money to the king. Philip IV was in desperate need of funds, having fought a major war with England only shortly before. Few regular tax revenues underwrote the king’s finances; he was largely expected to live off his demesne income. Loans regularly filled the gap between expenditure and revenue, especially at times of major financial demands such as during the campaigns in southwestern France against the English. After his attack on the Templars, Philip IV unsurprisingly found few willing lenders. Two hundred and fifty years later, the advisers of another king called Philip—who also happened to be strapped for cash while fighting expensive wars and borrowing heavily—voiced similarly murderous sentiments. In 1596, the Marquis of Poza, president of the Council of Finance of Castile, noted in writing that he would like to put the Genoese bankers lending to Philip II of Spain to the sword: “Every day, we discover new evidence against the Genoese…. I couldn’t possibly have enough of their blood” (Sanz Ayán 2004).1
Lending to kings could indeed be a perilous business.2 And yet in contrast to the events in Paris in 1307–14, the dark mutterings of the Marquis of Poza did not result in bloodshed. No banker to the Crown of Castile lost their life because of royally sanctioned violence. What had changed? Not only did the Genoese escape with their throats intact; they also made good money. A more enlightened approach to the lives of subjects or sudden aversion to killing was certainly not the cause; Philip II was perfectly capable of executing alleged troublemakers if it suited him. Even loyal subjects like the Counts Egmont and Horn, who opposed Spanish policy in the Netherlands, were summarily executed.
How can we explain the vast difference in outcomes—the violent persecution of the Templars, Jews, and Lombards, on the one hand, and prosperous security of the Genoese, on the other? Continued lending under Philip II of Spain stands in stark contrast to the collapse of royal credit under Philip IV of France. De Molay’s fate illustrates the heart of the problem of sovereign lending. Because the borrower is also the supreme judge and lawmaker, there can be no appeal to higher authorities if contracts are broken. Medieval kings repeatedly engaged in repudiations and confiscations on a grand scale. While some new lenders can sometimes be convinced that their fate is going to be different than that of the last group, few were willing to lend to Philip IV after the Templars’ end. In fact, in both medieval England and France, where monarchs had defaulted and turned on their creditors, royals found it hard to borrow afterward. Philip II had to reschedule his debts several times, but his creditors were never persecuted. As we will see, the spirit of loan agreements was upheld consistently (even if the king did not always fulfill them to the letter).
Sustaining sovereign borrowing on a vast scale—indeed, a modern scale—cannot be taken for granted. In this chapter we put forward our explanation: Philip II’s borrowing needs were so large that he could not do without future loans by Genoese lenders. Since these lenders were strongly connected among themselves, they “acted as one” in times of crisis; none cut a side deal, despite numerous offers from the royal camp. Attempts to lure in bankers from outside the circle of Genoese came to nothing. Faced with a dominant “coalition” that could effectively deny him the means to borrow and fight in the future, Philip II always came back to the negotiating table, offering solutions to his bankers that were fair and broadly in line with the original agreements. Sovereign lending—and the lives and limbs of his creditors—were effectively protected by their market power.
Our argument is based on a close analysis of our full database of asientos combined with bankers’ correspondence and the eventual settlement of the 1575 bankruptcy. This shows that bankers imposed effective lending moratoriums when the king failed to service old debts. Additional sanctions—though perhaps attempted—were ineffective. Genoese bankers provided two-thirds of short-term loans in overlapping partnerships, effectively forming a network or coalition. This lending structure created a web of multilateral obl
igations. As a result, lending moratoriums stopped the king’s access to credit: no network members broke rank; no preexisting lender from outside the network lent to him; and no new bankers supplied funds. The reason, we contend, is that bankers who “cheated” by lending during the moratorium would have faced severe penalties. Network members could hurt each other financially by seizing cross-posted collateral or failing to make payments due. Outsiders also did not enter since they feared being defaulted on by the king.
Philip II’s borrowing can be explained without punishments or banker irrationality. Instead, we document the importance of lenders’ incentive structures and banker collusion, using archival evidence. Our results are among the first to provide empirical support for models of sovereign lending that rely only on the borrower’s need for intertemporal smoothing and lateral enforcement among creditors. This finding is significant because it offers direct empirical proof for reputation-based models of sovereign lending where cheat-the-cheater incentives play a major role.
We begin our analysis summarizing how debt has worked over the last two centuries and how economic theory has approached the issue of sovereign debt in general. In the second part of this chapter, we demonstrate how lender collusion among the Genoese, cooperating in a network structure, kept the system of government borrowing going.
THE FUNDAMENTAL PROBLEM OF SOVEREIGN DEBT: THE ABILITY TO BORROW
An economist is someone who sees something that works in practice and wonders if it would work in theory.
—Ronald Reagan
For modern-day economics, the existence of sovereign (cross-border) debt contracts is a puzzle. Government spending can be financed by a mix of borrowing, taxes, and seigniorage. Borrowing today implies that revenue from the latter two sources of funding needs to increase in the future if the debt is to be repaid. Default on domestically issued debt can also be avoided, for example, by changing the tax treatment of bondholders. Within a country, government debt mainly has a redistributive effect, shifting the burden of funding the government between taxpayers, lenders, and holders of currency. The origins of government borrowing actually lie in a semicoercive structure: Italian city-states such as Venice initially forced their elites to lend. These loans could then be traded in secondary markets, and interest was paid on them (Kirshner 1996; Stasavage 2011).3
The defining problem of sovereign debt is that foreign governments are normally not subject to contract enforcement by courts in other countries (Panizza, Sturzenegger, and Zettelmeyer 2009). Cross-border borrowing is different from domestic transactions because the bondholders are not under the rule of the borrower.4 International loans are extended not because the alternative is taxation or expropriation; rulers have no direct power to coerce lenders into offering any funds. Foreign lenders need to be convinced that they will obtain a return on their investment.
HOW DEBT HAS WORKED
Over the last two centuries, sovereign debt has on the whole been profitable for lenders. Based on a comprehensive data collection exercise, Lindert and Morton (1989) analyze the returns for bondholders during the period 1850–1970. They find that on average, creditors made money, despite the fact that 106 countries defaulted a total of 250 times in the last two centuries (Tomz and Wright 2007). The excess return compared to loans to a benchmark “home” sovereign (the United Kingdom or United States) amounted to 0.42 percent. This is much less than the 1.81 percent return differential promised by the original loan contracts, but it implies risk-neutral investors clearly benefited from exposure to cross-border lending.5
This is not to say that there weren’t severe losses; many bondholders lost their shirts over the centuries. For instance, investors in Argentine debt lost on average 17 percent annually during the period 1992–2001 (Sturzenegger and Zettelmeyer 2008). Elsewhere, investors earned extremely rich returns. Holders of Egyptian debt in the last two centuries, for example, were promised 6.7 percent annually and received 6.2 percent—a full 253 basis points above the return on alternative benchmark bonds issued by the investors’ domestic government (Lindert and Morton 1989). Similarly, investors in Brazilian bonds between 1992 and 2001 earned an annual return of over 16 percent. Barry Eichengreen and Richard Portes (1989a) analyzed returns on sterling and dollar bonds during the 1920s and 1930s. They also find modest, though positive, compensation for risk overall.6
If sovereign debt on average “works,” how do we make sense of its success? The theoretical literature grapples with the question of why sovereign lending can exist at all. It can be divided into two broad approaches: sanctions and reputation. Papers in the reputation tradition argue that the need to smooth consumption is key: if a borrower fails to honor their contract, credit will dry up. The borrower will be markedly worse off, being unable to borrow in hard times. In contrast, the sanctions literature argues that without penalties above and beyond the mere exclusion from future borrowing, sovereign lending cannot exist. Sanctions range from trade embargoes to military intervention. Here we review the theory and evidence supporting each approach, and also discuss what happens when restructuring becomes necessary. Finally, we put the case of Habsburg Spain in the context of predictions from the modern literature.
SANCTIONS
Without sanctions, no sovereign debt can be issued. This is the argument made by Jeremy Bulow and Kenneth Rogoff (1989), who analyze a model in which the principal benefit of servicing debt is consumption smoothing. When a sovereign can sign insurance contracts, purchase assets, or make deposits in a bank, he can always make himself better off by first defaulting and then using “self-insurance” (ibid.). The implication of this reasoning is that no sovereign debt can exist in equilibrium, no matter how great the need to smooth is, if there are alternative ways to transfer resources across time. In Bulow and Rogoff’s setup, sanctions are the only factor that can make borrowing possible. If lenders receive help from their own governments—dispatching gunboats, say—then penalties can be so painful that loans will be repaid. Importantly, penalties have to go beyond the withholding of credit: punishment in addition to a moratorium by the existing lender is necessary to make sovereign borrowing feasible.
Over the last two hundred years, there are examples of both trade sanctions and armed intervention—but they are exceedingly rare. Carlos Díaz-Alejandro (1983) argued that Argentina did not default in the 1930s to avoid trade restrictions from lenders, but the claim is controversial (Tomz 2007). In other episodes, the mere threat of trade sanctions and expropriation of trade revenues may have improved outcomes for creditors. In the 1930s, for instance, the United Kingdom threatened to interfere with German trade if bondholders were not paid; Germany backed away from defaulting on its payments to British creditors (Eichengreen and Portes 1989a).
Direct military intervention constituted an effective punishment strategy. For example, Egypt first lost control of its customs revenues after suspending payments in the 1870s. Britain took over the running of the khedive’s finances, before making Egypt a part of the empire (Mitchener and Weidenmier 2010). So-called supersanctions—interventions with military force or the threat thereof—effectively reduced bond spreads. In 1904, the United States extended the Monroe Doctrine—the “Roosevelt Corollary.” It called for intervention in case of debt delinquency; Latin American bonds rallied sharply in response (Mitchener and Weidenmier 2005). Supersanctions were effective but uncommon. The British government generally held to the “doctrine that if investors choose to buy the bonds of a foreign country carrying a high rate of interest in preference to British Government Bonds … they cannot claim that the British government is bound to intervene in the event of default.”7
Andrew Rose (2005) argues that defaults coincide with a collapse in trade (contributing to lower output), and that this effectively constitutes a “punishment.” This is in line with Phillip Lane’s (2004) finding that countries more open to trade can support greater external debts. Rose shows that the volume of exports and imports between creditor and debtor countries s
hrinks substantially during a payment stop.8 Eduardo Borensztein and Ugo Panizza (2009) analyze a more comprehensive data set on trade at the industry level; they discover that exporters suffer—but only for relatively short periods. This effect is compensated for by a decline in competing imports for domestic industries (Lanau 2008).
The sanctions view receives support from calibrated models of sovereign borrowing. Even permanent exclusion from debt markets can only support relatively low levels of debt (Arellano and Heathcote 2010). Mark Aguiar and Gita Gopinath (2006) demonstrate that in their model, the value of smoothing consumption around a stable long-term trend is small and cannot sustain much lending.9 If there is an additional output penalty of 2 percentage points of GDP during a default, however, then feasible debt payments increase sharply. These payments can then reach 20 percent of GDP—implying a debt-to-GDP ratio above 100 percent.
Output losses in times of default can rationalize why positive and significant borrowing occurs. Calibrated models have more problems matching the frequency of default. The Aguiar and Gopinath model predicts that up to 50 percent of countries will restructure or be in default during peak years. In actual fact, the number is closer to 10–20 percent of countries (Reinhart and Rogoff 2009).10
The empirical literature typically finds significant declines in GDP during payment stops (Barro 2001; Cohen 1992). What is less clear is why they happen. Defaults may simply reflect “hard times.” In a paper studying the incidence of payments stops for the last two hundred years, Michael Tomz and Mark Wright (2007) show that countries typically default during downturns; more than 60 percent of the default episodes in their sample occurred while output was below trend. Yet the relationship is surprisingly weak: while some 32 percent of all defaults occurred in the 5 percent of observations in their sample with the biggest output declines, fully one-fifth happened in countries with mild downturns.11
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 18