)4)14 cag£. tœq THE INTERWAR DEBT CRISIS AND ITS AFTERMATH Barry Eichengreen Richard Portes L~ This article analyzes the sovereign defaults of the 1930s and reports nine major findings. (1) There is little evidence that financial markets of the 1930s were unsophisticated or that banks have a comparative advantage over the bond market in processing information. (2) Debt default in the 1930s depended on a combination of factors, including the magnitude of the external shock, the level of debt, and the economic policy response, as well as on a range of non- economic considerations. (3) Countries that interrupted service recovered more quickly from the Great Depression than did countries that resisted default. (4) There is little evidence that countries that defaulted in the 1930s later suf- fered inferior access to the capital market. (5) The readjustment of defaulted debts was protracted: the analogy with corporate bankruptcy proceedings is not applicable. (6) Although default led in some cases to a substantial reduc- tion of transfers from debtors to creditors, on balance returns on sovereign loans compared favorably with returns on domestic investments. (7) Creditor governments in the 1930s did act to accelerate the settlement process. (8) Glo- bal schemes analogous to the Baker plan were widely proposed but never im- plemented. (9) In contrast, debt buybacks played a useful role in the resolution of the crisis. T he history of foreign lending in the nineteenth and twentieth centuries offers a rich lode of evidence on the operation of international capital markets. For the historian, that experience provides an exceptional op- portunity to study the long-term evolution of international markets and their adaptation to repeated shocks. For the economist, it provides an opportunity The World Bank Research Observer, vol. 5, no. 1 (January 1990), pp. 69-94 C 1990 The International Bank for Reconstruction and Development/THE WORLD BANK 69 to reflect on how the current debt crisis may be resolved. It is not possible to extrapolate directly from historical experience, since institutional aspects of the lending process, including the relative importance of bank and bond finance, the rise of supranational agencies such as the World Bank and the International Monetary Fund, and the role of creditor governments in rescheduling, have changed fundamentally over the past century. But even though the extent of institutional variation makes it difficult to draw simple lessons from the past, history still offers the only evidence we have on the efficiency and distribution- al effects of different methods of organizing international lending and readjust- ing existing debts. In a series of papers, we have examined the lending of the 1920s, the defaults of the 1930s, and the debt readjustments of the 1940s and 1950s (see Eichengreen 1989a, 1989b; Eichengreen and Portes 1986, 1987, 1988, 1989a, 1989b, 1989c). This paper summarizes and extends the main conclusions of that research. The discussion is organized around nine findings. 1. Investors in the 1920s exhibited sophistication and foresight at the lend- ing stage. Our analysis suggests that a country's repayment record, its political circumstances, and its economic policies all figured in deter- mining the risk premiums on foreign bonds. There is little evidence that capital markets have grown more sophisticated over time or that banks have a comparative advantage in processing the relevant information. 2. No single economic variable-nor, for that matter, group of economic variables-adequately explains the incidence and extent of default. Although Diaz-Alejandro (1983) and Fishlow (1985) have pointed rightly to the magnitude of the external shock of the 1930s (proxied typically by the extent of deterioration in the terms of trade) as a leading indicator of default, our work reveals the importance of other economic variables, including the burden of the debt and the nature of the domestic policy response, as well as noneconomic variables, such as political links. 3. The implications of different debt management strategies for macroeco- nomic performance are difficult to isolate. In the 1930s, efforts to main- tain debt service tended to be associated with fiscal austerity, import compression, and export subsidies, whereas the decision to suspend pay- ments was often accompanied by fiscal expansion, monetary reflation, and import-substituting industrialization. This wholesale reorientation of a country's macroeconomic stance makes it difficult to pick out the effects of external debt management from the entire constellation of policies. Evidence points nonetheless to the conclusion that countries that interrupted service on their external debts recovered more quickly from the Great Depression than countries that resisted default. 4. There is little evidence that countries that defaulted in the 1930s in- curred a cost in terms of inferior access to the capital market after World War II. Following negotiated settlements with the creditors, countries that previously had suspended interest payments and amorti- 70 The World BankResearch Observer, vol.5,no. I (January 1990) zation were offered access to the capital market virtually identical to that of countries that had maintained debt service without interruption. This is not to suggest that default was without costs in relation to mar- ket access, only that those costs were not borne differentially by coun- tries that interrupted their debt service. Many of the costs were external to the defaulting countries: neither defaulting nor nondefaulting debtors had significant access to portfolio capital in the decades following World War II. 5. The readjustment of defaulted debts entailed a protracted process of ne- gotiation. The analogy with corporate bankruptcy proceedings, in which default and readjustment permit a clean break with the past, is not applicable. Default on sovereign debt has not been feasible in prac- tice. Although governments have from time to time suspended payment, in many cases the interruption was only temporary, and uncertainty over transfers lingered for decades. 6. Although default in some cases led to a substantial reduction of trans- fers from debtors to creditors, what we might call selective debt relief permitted a reasonable overall rate of return. The risk premiums charged sufficed to elevate the average rate of return on central govern- ment loans above the yields on British and U.S. treasury bonds. Losses to creditors on provincial, municipal, and corporate loans were more extensive, but risk premiums were still sufficient to yield positive returns to British investors. 7. Creditor governments often were intimately involved in debt negotia- tions. In recent years creditor governments have exerted continuous pressure on the debtors to maintain service on their external debts. In the 1930s and 1940s creditor governments pressured debtors and credi- tors alike. 8. Global schemes to short-circuit the protracted process of bilateral nego- tiation proved fruitless in the 1930s. Nearly every element of the global plans currently under consideration-a special international lending fa- cility, matching injections of private and public funds, conversion of ex- isting assets into new ones featuring different contingencies-was first suggested in the 1930s. Ultimately, those global schemes foundered on the question of who should fund and control them. 9. Market-based debt reduction made a useful contribution to resolving the debt crisis of the 1930s by reducing the debt overhang and eliminat- ing marginal creditors. There is little evidence that debt buybacks raised the market price of remaining debt sufficiently to leave the market value of the debt burden unchanged. Rather, prices in the secondary market seem to have been governed primarily by changes in the prospects for a negotiated settlement. In contrast to public statements of disapproval, creditor organizations were willing in private to entertain buybacks out of reserves as part of the readjustment process. Barry Eichengreen and Richard Portes 71 The Lending Stage International capital markets of the 1920s have been accused of all manner of excess. Issuers pushed questionable obligations on hesitant borrowers and questionable bonds on otherwise cautious investors (Skiles 1988, Darity and Horn 1988). New entrants into the market-large New York banks that "fore- saw a series of lean years in wholesale and industrial banking, and jumped on the new bandwagon of retail banking"-were particularly guilty (de Cecco 1985, p. 57). But they were not alone. Private investors failed to discriminate between good and bad credit risks (U.S. Congress 1932). Governments failed to monitor the industry adequately and to discourage dubious activities, even erring in the other direction as when, for example, they encouraged the flota- tion and purchase of German bonds (Schuker 1988). So it is alleged. Unfortunately, most of these assertions are difficult to test. It is tempting to cite the poor performance of these loans as proof of the va- lidity of the interpretation. There are obvious dangers, however, in drawing conclusions on the basis of 20/20 hindsight. There was no reason for investors in the 1920s to anticipate a macroeconomic crisis on the scale of the Great Depression. That their loans performed poorly given the exceptionally poor state of the global economy does not suffice to impugn the lending process. It is more informative to consider the market's ex ante assessment of the risks of foreign lending. Such assessment should be reflected in larger projected yields at purchase for higher-risk loans, which are reflected in spreads, that is, in the difference between the yields for high-risk and risk-free loans. We cal- culated spreads by comparing yields on two samples of foreign bonds floated in London and New York in the 1920s with yields on contemporaneous U.K. and U.S. government securities (consols and Treasury bonds, respectively).' Then, to determine if these spreads reflected consideration of the likely indica- tors of risk of default, their level was related to country characteristics, current economic policies, and capital market conditions. This method parallels that adopted by Edwards (1986), who sought to address Guttentag and Herring's (1985) contention that the rates banks charged foreign borrowers on bank loans could not have incorporated adequate country risk premiums because they varied so little across loans. Edwards attempted to test this hypothesis for the 1970s and 1980s for both bank loans and bonds by regressing spreads on indicators for country risk such as the magnitude of the debt burden, the level of investment, the policy stance of the debtor government, and other charac- teristics of the borrower and the loan. He found for the bond market that the spread rose with the ratio of debt to gross national product (GNP) and fell with the ratio of investment to GNP. Although this is consistent with the hypothesis that lenders distinguished among good and bad credit risks, his other coeffi- cients were insignificant, suggesting that investors paid little attention to other indicators of country risk.2 72 The World Bank Researcb Observer, vol. 5, no. 1 (January 1990) We seek to replicate this analysis for the 1920s. In our regressions (reported in appendix table 1) the spread on each loan is related to the debt-to-export ratio, to the category of borrower (national governments or others), to the year in which the loan was issued, and to the geographic location of the borrower. We also include two measures of economic policies and conditions in the bor- rowing country: the trade balance and the budget balance. For the 1920s it is not possible to construct a measure of the investment share of GNP like that used by Edwards for the 1970s and 1980s. But neither were investment statistics available to prospective bondholders. Instead, contemporary investment man- uals urged investors to focus on the trade and budget balances as the two most important indicators of a country's capacity to generate and mobilize the for- eign exchange required to service its external debt (Madden and Nadler 1929). Our analysis shows that the spread barely changed over time, after control- ling for other characteristics of the loan, which is contrary to what one would expect to find had foreign lending tended to come into or fall out of fashion.3 Compared with loans to national governments, higher spreads were charged on loans to states, municipalities, corporations, and banks. This is consistent with evidence, presented in the following, that loans to national governments outperformed these other loans.4 In the case of sterling loans there is similar evidence of informed behavior: higher risk premiums were charged on loans to local governments than to central governments and on politically unrelated for- eign issues than on issues on behalf of members of the British Commonwealth of Nations, or the Dominions. The results for dollar bonds suggest, not surprisingly, that Canada, Central America, and Western Europe enjoyed the best reputations. Canada and Western Europe had virtually unblemished records of servicing their foreign debts, whereas the small Central American republics economically or political- ly dependent on the United States had little choice in the matter.5 Conversely, the nations of Eastern Europe were charged the highest risk premiums. These geographical variations suggest bondholder sophistication, since the pattern of realized returns indicates that loans to Eastern Europe were relatively risky and those to Western Europe and Central America were relatively safe. With hind- sight, only the risk premium attached to German loans is anomalous. The spreads on loans to Germany, the leading borrower of American funds, were smaller on average than those on loans not only to Eastern Europe and South America but also to Australia and Japan. This is consistent with the warning by the U.S. State and Commerce departments, voiced as early as 1925, that American investors underestimated the risks associated with investing in German bonds (Eichengreen 1989a). The results for sterling bonds are basically consistent with the hypothesis that bondholders were relatively knowledgeable. The average spread is smaller (1.8 percentage points, compared with 2.6 for the United States), perhaps re- flecting the perception that British loans were less risky given their rather dif- ferent geographical destination. High risk premiums were charged on loans to Barry Eichengreen and Richard Portes 73 the Eastern European countries; Latin America, Greece, and Germany stood in the middle; and the Dominions and Japan enjoyed the lowest costs of borrow- ing. In a well-functioning market, investors should take into account current eco- nomic policies as well as reputation. It is logical, therefore, that we find for both sterling and dollar bonds some tendency for trade and budget surpluses to have reduced the cost of borrowing. Note, however, that only budget sur- pluses in the equation for dollar bonds differ significantly from zero at stan- dard confidence levels. Although absolute statements about the efficiency of a market are always problematic, comparative evaluations are more straightforward. In this case there is no compelling evidence that investors were less discriminating in the 1920s than in the 1970s. Nor is there compelling evidence that either banks or the bond market possess a comparative advantage in the pricing of foreign loans. Causes of Default The debt crisis of the 1930s unfolded in three stages (Eichengreen and Portes 1987). The first, spanning calendar 1931, was dominated by Latin American defaults. Interest and amortization payments were at least partially suspended by every South American country except Argentina. During the second stage, from 1932 to the middle of 1933, default spread to Southern and Eastern Europe. The precise amount by which payments were reduced varied across countries, but most countries were affected to some extent. The third stage, which coincided with the Monetary and Economic Conference of 1933, was dominated by Germany's default. The debt crisis did not have a single cause. Liberal foreign borrowing in the 1920s had increased the debtor countries' vulnerability to external shocks. By 1928 central government debt-to-export ratios in many Latin American coun- tries were over 100 percent, and interest and amortization on this component of the debt alone could require 10 percent of total export receipts (see table 1). Although central government debt-to-export ratios are not representative of the level of total debt, they are the only debt indicators available for a wide range of countries. Where state and local government debts were substantial, the total government debt-to-export ratio could be much higher (Schuker 1988, p. 65). Germany, the largest single foreign borrower, owed reparations as well as commercial debts; in 1928 these two categories required nearly 25 percent of gross export receipts (Harris 1935). The debtor countries were battered by a series of external shocks that in- creased their transfer burden and reduced their debt servicing capacity. First was the sudden decline in lending by the United States and other creditor na- tions. The mounting boom on Wall Street diverted American funds from for- 74 The World Bank Research Observer, vol. 5, no. I (January 1990) Table 1. Defaulters' Central Government Debt-to-Export Ratios, for Selected Years, 1929-35 (percent) Country 1929 1931 1933 1935 Heavy defaulters Brazil 153 163 215 127 Bulgaria 263 288 - 416 Chile 102 327 842 573 Colombia 58 101 123 67 Costa Rica 96 130 174 158 El Salvador 106 153 183 157 Germany 7 34 62 42 Greece 415 750 633 474 Guatemala 65 97 162 123 Hungary 124 251 347 256 Poland 135 212 472 355 Uruguay 147 185 212 148 Unweighted average 139 224 311 241 Light defaulters Argentina 49 73 113 81 Australia 162 684 621 585 Austria 78 157 310 274 Belgium 87 111 190 240 Canada 44 82 98 107 Czechoslovakia 26 48 119 98 Denmark 45 56 57 54 Finland 52 128 74 51 France - - 22 18 Italy 12 17 28 30 Japan 69 132 76 57 New Zealand 278 454 400 350 Nicaragua 30 42 53 46 Norway 107 165 133 118 Spain 43 95 137 157 Venezuela 5 1 1 1 Unweighted average 72 150 152 142 - not available. Note: War debts and reparations are excluded for France. Figures for Australia from 1931 include state debts assumed by the federal authorities. Those for Brazil and Chile include selected state and local debts assumed by the federal authorities. Source: League of Nations (various years). eign to domestic uses and like a powerful suction pump siphoned off liquidity from the rest of the world. At the beginning of 1928 approximately $800 mil- lion was required to meet foreign debt service payments on dollar debts. Net short- and long-term foreign lending by the United States had exceeded $1,000 million in 1927 and reached nearly $700 million in 1928. Thus, despite Barry Eichengreen and Richard Portes 75 the collapse of new lending in the summer of 1928, the funds provided between January and June nearly sufficed to finance that year's dollar debt service costs. In 1929 net short- and long-term lending by the United States turned neg- ative, and the $800 million bill came due. There could have been no less opportune time for the collapse of global commodity markets. Exporters of primary commodities had already come un- der strain. Their terms of trade had been deteriorating steadily throughout the 1920s and plunged with the onset of the Depression. Lewis (1949) provides the sad litany. "From 1929 to 1930 the average price of wheat fell by 19 per cent, cotton 27 per cent, wool 46 per cent, silk 30 per cent, rubber 42 per cent, sugar 20 per cent, coffee 43 per cent, copper 26 per cent, tin 29 per cent; the index of prices of commodities entering world trade fell from 1929 to 1932 by 56 per cent for raw materials, 48 per cent for foodstuffs, and 37 per cent for manu- factures" (p. 56). It is these developments that Diaz-Alejandro (1983) and Fishlow (1985) stress as determinants of default. They suggest that the collapse of real export revenues was so severe that countries could avoid default only through the most decisive, concerted action. They argue that the magnitude of the decline in the terms of trade determined whether the necessary steps were still feasible. Argentina and Australia continued to service their debts, for ex- ample, because as wheat exporters they were less affected than countries such as Brazil, which exported coffee, or Bolivia, which exported tin. The effect of the collapse of export prices was reinforced by the contraction of the volume of exports. Between 1929 and 1932, world trade in foodstuffs fell 11 percent, and trade in raw materials dropped 19 percent (Lewis 1949, p. 58). Analysis of the incidence and extent of default suggests, however, that the "commodity lottery" (Diaz-Alejandro's phrase) was only one of several deter- mining factors. Eichengreen and Portes (1986) report a regression analysis re- lating the percentage of government and government-guaranteed debt in default (for all levels of government) in the 1930s to a set of country charac- teristics. These results confirm that sharp declines in commodity prices in- creased the incidence and extent of default. But in addition, two other sets of variables conditioned debt management strategy. First, the domestic response played an important role. Countries that reined in government budget deficits, through either increased taxes or reduced public spending, were less likely to default. To put it another way, the governments least willing or able to re- trench fiscally were least able to avoid default. Second, political aspects of the decision not to suspend debt service payments appear to retain a role, even al- lowing for economic circumstances. For example, our equation significantly underpredicts Australia's level of debt service. That Australia avoided default on even a portion of its external debt we attribute to its political ties to Great Britain, to which the vast majority of the debt was owed. This analysis provides a more nuanced picture of the debt servicing deci- sions of borrowing countries, in which the severity of the external shock, the 76 The World Bank Research Observer, vol. 5, no. 1 (January 1990) vulnerability of the economy, the domestic policy response, and a broader range of political factors combined to influence the decision. Macroeconomic Repercussions of Default Efforts to maintain external debt service payments required the compression of imports and current expenditures. A surplus on the noninterest current ac- count of the public authorities was needed to mobilize funds for service of gov- ernment and government-guaranteed debt. The drop in export revenues and international reserves forced officials to adopt increasingly stringent monetary measures to strengthen the balance of payments. In combination, these policies depressed domestic demand in general and investment demand in particular. In contrast, governments willing to suspend interest payments were better able to adopt monetary and fiscal policies conducive to recovery. For the 1930s, it is possible to compare the economic performance of coun- tries that adopted very different policies toward external debts (see figures 1 and 2). Both GNP and industrial production appear to have expanded more quickly after 1931 in those countries that were heavy defaulters. This compar- ison suggests that countries that opted for default recovered more successfully from the ravages of the Great Depression. There are several reasons to treat the comparison with caution. First, poli- cies influencing growth may have differed across countries in ways that are not directly attributable to different strategies for managing debts. There is also the possibility that because production declined more dramatically before 1931 in defaulting countries, higher levels of unemployment and excess capacity may have offered more scope for output to snap back. Finally, declines in the terms of trade were more severe for the heavy defaulters. Thus any subsequent re- bound in the terms of trade also could have induced a more favorable growth performance after default. As shown in figure 3, however, the terms of trade of the heavy defaulters actually recovered less fully than did those of the light defaulters, and except for a short period around 1934, their (lower) terms of trade then generally moved in parallel with those of the light defaulters. We used regression analysis to control for other factors likely to have affect- ed the growth of industrial production in the 1930s (the analyses are reported in full in Eichengreen and Portes 1989b). For example, we added movements in the terms of trade as an additional determinant of growth performance. In general it appears that countries that suffered more severe external shocks, as measured by movements in their terms of trade, grew more slowly, as one would expect. Countries with larger shares of debt in default still appear to have recovered more quickly in the 1930s even after controlling for this addi- tional determinant of growth. Of course, there is the possibility that causality ran in the other direction: that growth performance determined the extent of default rather than the other Barry Eichengreen and Richard Portes 77 Figure 1. Rates of Growtb of Real National Income, 1929-3 7 (percent) 15 10 Heavy defaulters 5 0 . / Light defaulters -5 -10 I I I I I 1929 30 31 32 33 34 35 36 37 Note: Unweighted averages of country data are used. Heavy defaulters: Brazil, Bulgaria, Chile, Colombia, El Salvador, Germany, Greece, Guatemala, Hungary, Poland, and Yugoslavia. light defaulters: Argentina, Australia, Austria, Belgium, Czechoslovakia, Denmark, Finland, France, Italy, Japan, New Zealand, Nicaragua, Norway, and Spain. Source: Eichengreen and Portes 1986, pp. 636-37. way around. But this should only reinforce the conclusions drawn from the analysis. If rapid growth in output makes default less likely, this reduces the probability that we will observe more rapid growth in countries that did de- fault, as in figures 1 and 2. Two-stage least squares regressions intended to control for feedback from growth to default are consistent with the hypothesis that continued debt service led to slower economic growth. To test the "rubber band effect" (that all we are observing is that countries that suffered the largest declines in output in the early stages of the Depression had the greatest scope for recovery), we related the rate of growth of produc- tion over the recovery period (1931-36) to the percentage of debt in default, the percentage change in the terms of trade, and the percentage change in pro- duction in the preceding period (1929-31). Although we found evidence of the rubber band effect, it remained true that the rate of growth in the recovery period was positively associated with the percentage of debt in default. The decision to continue external debt service was only one of a number of policy responses in the 1930s. There was little conscious manipulation of bud- 78 The World Bank Research Observer, vol. 5, no. 1 (January 1990) Figure 2. Rates of Growtb of Industrial Production, 1929-3 7 (percent) 15 10 Heavy defalulters 5 -5 -10 1929 30 31 32 33 34 35 36 37 Note: Unweighted averages of country data are used. Heavy defiulters: Brazil, Bulgaria, Chile, Colombia, El Salvador, Germany, Greece, Guatemala, Hungary, Poland, and Yugoslavia. Light defaulters: Argentina, Australia, Austria, Belgium, Czechoslovakia, Denmark, Finland, Prance, Italy, Japan, New Zealand, Nicaragua, Norway, and Spain. Source: Eichengreen and Portes 1986, pp. 636-37. getary instruments, although there were independent expenditure programs that had fiscal effects, such as schemes to stabilize commodity prices and pro- grams of public assistance. Monetary policy was more widely used, although monetary authorities were constrained by fears of inflation, by the desire to defend the gold standard, and by doubts about the effectiveness of monetary reflation. The most widely adopted measures were those referred to as import substitution policies (see Fishlow 1972, Thorp 1984). With the collapse of ex- port markets after 1929, governments adoptedf policies designed in varying de- grees to reduce their dependence on exports and to shift resources to industries that could produce substitutes for imports. To raise the price of exports, exchange rates were devalued, first in Australia and Latin America and subsequently in other parts of the world. Tariffs and quotas, often supplement- ed by exchange controls, were used to limit imports. In some countries gov- ernment credit was made available on favorable terms to manufacturers of products that competed with imports. The use of such policies may well have Barry Eichengreen and Richard Portes 79 Figure 3. Index of International Terms of Trade, 1929-36 (1929= 100) 105 100 95 \ Light defaulters 90 _N 85 Haydfutr 80 75 1929 30 31 32 33 34 35 36 Note: Heavy defaulters: Brazil, Bulgaria, Chile, Colombia, El Salvador, Germany, Greece, Guatemala, Hungary, Poland, and Yugoslavia. Light defaulters: Argentina, Australia, Belgium Czechoslovakia, Denmark, Finland, France, Italy, Japan, New Zealand, Nicaragua, Norway, and Spain. Source: Eichengreen and Portes 1986, pp. 636-37. been related to the choice of debt management strategy. Countries such as Argentina which continued to service their debts did so by retaining resources in the export sector and by relying on imports for other goods. Other coun- tries, such as Brazil, which reduced or suspended debt service payments, had neither as urgent a need to generate foreign exchange receipts nor as favorable access to certain creditor country markets, and had reason to shift resources to domestic producers. In an attempt to capture the effects of this range of trade policies, we added to our regressions the import share of GNP as a determinant of the growth rates of income and industrial production. The variable is positively associated with the rate of economic growth, suggesting that countries that most rapidly raised their capacity to import recovered most quickly from the Great Depression. The coefficient on percentage of debt in default continues to indicate that heavy defaulters recovered more quickly from the Depression. 80 The World Bank Research Observer, vol. 5, no. I (January 1990) Default and Access to the Capital Market The immediate way for creditors to retaliate against defaulters was to exclude the debtors from the capital market. From 1825, the London Stock Exchange refused to list new loans of governments in default on outstanding obligations and in extreme instances refused to quote all loans of the offending government. Although New York had no comparable arrangement, it was still possible for individual creditors to refuse to accommodate borrowers in de- fault. That the capital markets reacted strongly to the defaults of the 1930s is be- yond question. For years-in fact decades-the volume of portfolio lending re- mained depressed. Only in the late 1960s and early 1970s did portfolio capital flows recover as a result of the entry of money center banks into international retail banking. A generation of American and European investors had conclud- ed that purchases of foreign bonds were to be avoided. An official of the U.S. National Association of Manufacturers said in 1949, "After the experience of the thirties and the serious balance of payment difficulties now plaguing most of the world, the superiority of equity over loan financing has, we believe, a universal appeal" (Eichengreen 1989a, p. 137). Few deceived themselves into believing that the risk of nationalization was negligible; that it was thought to be significantly less than the danger of default on foreign portfolio investments is an indication of the scars left by the experience of the 1930s. But there is scant evidence that defaulting debtors were differentially affect- ed. On the basis of a study of six large Latin American debtors, Jorgensen and Sachs (1989) find no observable capital market penalty in the period 1950-64 for defaults in the 1930s. Lindert and Morton (1989, pp. 63-65) conclude from their study of more than a century of international lending experience that de- faulting governments were seldom punished with direct sanctions or discrimi- natory denial of credit. Eichengreen (1989a) analyzes borrowing by a cross section of thirty-two countries in the first postwar decade, relating foreign bor- rowing to both demand-side variables (as proxied by country characteristics such as openness, export variability, and inherited debt) and supply-side vari- ables, notably past performance in servicing debt. The results show no appar- ent relationship between the severity of default and the ability to borrow immediately after World War II. This is not to deny that lenders were aware of defaults. But the reaction against portfolio investment abroad was general rather than selective. "A gen- eral sigh of resolve was to be heard over the United States," Herbert Feis wrote. "Never again should we lend or invest our money in foreign lands" (Feis 1950, p. 1). American investors, having been rudely reminded of the spe- cial risks of foreign loans, revised their assessment of the desirability of lending abroad. The British and others reacted similarly, and they were significantly constrained by controls on capital movements until the end of the 1950s. Hence the market for portfolio investments remained becalmed for more than a third Barry Eichengreen and Richard Portes 81 of a century. Sovereign default may have had costs in the form of disrupting access to capital markets. But many of these costs were external to the default- ing countries and affected the market access of defaulters and nondefaulters alike. Two caveats to this conclusion are in order. First, the reaction of the capital markets may have been a function of the fact that default was so widespread in the thirties. It therefore led to a comprehensive reassessment of the risks of international lending. Even countries that had kept their debt service up to date suffered loss of access to capital markets. In contrast, in such periods as the 1920s, when the only countries in serious default were Ecuador, Mexico, and the USSR, the market may have been more inclined to focus on the creditworthiness of particular countries and less inclined to reassess the entire enterprise. Second, the market's tendency to discriminate among countries according to their past record on servicing debt may differ during periods of stagnation and buoyancy. Ozler (1988) suggests that during the boom in bank finance in the 1970s, larger spreads over the risk-free rate were demanded of countries that had defaulted on loans. Still, there is little evidence that defaulters were kept out of the market. The Settlement Process Observers of the current debt crisis deplore its protracted nature. Investment and growth remain low, while uncertainty about the resolution of the crisis continues to depress bank share prices and to increase the vulnerability of the money center banks to destabilizing shocks. Often a contrast is drawn with the era of bond finance, when debtors were able to make a clean break with past problems and divert scarce resources to productive domestic uses and creditors were able to get on with their business. In fact, even in the era of bond finance the readjustment of defaulted debts often entailed protracted renegotiation. Service might be suspended in part or in full, restarted for some years, suspended again, and so forth. Negotiations with the creditors were frequently on-again, off-again, and required as much as a quarter of a century to complete. Bolivia, the first country to default, in 1931, was also the last major debtor to settle, in 1955. Brazil's experience is illustrative (Abreu 1978, Cardoso and Dornbusch 1989, Eichengreen and Portes 1989c). In October 1931 Brazil suspended interest pay- ments on most of its external debts. In March 1932 the government announced that it would issue twenty- and forty-year bonds to capitalize interest arrears and would resume normal interest payments no later than 1934. But in 1934, on the advice of British financial experts, Brazil announced a plan to readjust the debt. The Aranha Plan, designed to run through 1937, limited debt service to roughly half of Brazil's net export receipts. Bonds were divided into seven 82 The World Bank Researcb Observer, vol. 5, no. I (January 1990) grades, with funding loans and other select obligations to receive full interest, other federal, state, and local loans partial interest (17.5 to 50 percent of con- tractual levels), and certain state and municipal loans no interest. At the end of 1937, with the external situation little improved, debt service payments were suspended again. In 1940, following sporadic negotiations with its creditors, Brazil announced another temporary, four-year settlement that lowered the in- terest rates on the seven-categories of bonds. Finally, in 1943 Brazil negotiated a permanent agreement with its creditors, giving bondholders the option of choosing between two plans. Under the first, interest rates would be cut by 30 to 70 percent, and under the second, the creditors would surrender 20 to 50 percent of the bond's capital in exchange for a cash payment of 6 to 60 per- cent of par value and somewhat higher interest rates on the remainder. Thus, although there may be some truth to the notion that in the era of bond finance debtors were better able to jettison their debts and redirect re- sources to domestic uses, the extent of the difference should not be exaggerat- ed. The effects of the debt overhang were substantial. Even when payments were in suspension, there was often a real possibility that they would be re- started. Investors in debtor countries had reason to be wary about committing funds to domestic investment on the grounds that officials might tax the re- turns and devote them to debt service. The central difficulty for negotiators was the problem created by the exist- ence of a multitude of bondholders. It was even harder than in the era of syn- dicated bank lending to solicit the opinions of the creditors on a settlement proposal. Yet the British and, after 1933, the Americans came up with a re- markably efficient solution. Committees were appointed to negotiate with the debtor. Bondholders sometimes were asked to lodge their bonds with the com- mittee, other times simply to register as interested investors. Readjustment plans were signaled by the publication of an offer or simply by an announce- ment that bond covenants were henceforth modified. If the plan was judged to be fair or at least to be the best that could be expected, the committee would recommend its acceptance by the bondholders. The latter signified their ap- proval by cashing a coupon or, when requested, by exchanging the old certif- icate for a new one. The only option available to dissident bondholders was to hold out for better terms. The recommendation of a reputable bondholders' committee was the seal of approval on an offer. In Britain, where the Corporation of Foreign Bondhold- ers had been in existence since 1868 and was universally acknowledged to rep- resent the creditors, that recommendation carried considerable weight. When the corporation recommended that an offer be accepted, any stock exchange sanctions could be expected to be withdrawn, and dissident bondholders had little hope of obtaining better terms. In the United States, a comparable organization, the Foreign Bondholders Protective Council, came into operation only in 1934. Before that time, bondholders relied on ad hoc committees cre- ated to deal with individual defaults. The ad hoc nature of such committees Barry Eichengreen and Richard Portes 83 tended to inflate administrative expenses, and the existence of rival commit- tees, as well as committees of dubious reputation, made it difficult for both debtors and creditors to determine who best represented the interests of the bondholders. Terms of Settlement In some cases default led to a substantial reduction in net resource transfers from debtors to creditors (Lindert and Morton 1989, Jorgensen and Sachs 1989, and Eichengreen and Portes 1986, 1989a). Based on two large samples of foreign bonds issued between 1920 and 1929, Eichengreen and Portes (1989a) calculated the nominal own-currency internal rate of return on more than 300 issues (see table 2). For dollar bonds, at one extreme the return marginally ex- ceeded the contractual rate-as in the cases of Norway and Canada-because of early debt retirement. At the other extreme, the internal rate of return reached levels as low as -7.4 percent for Brazil, -9.8 percent for Bolivia, and -14.8 percent for Hungary, indicating that not just interest but a substantial fraction of the principal had been written off by the creditors. On average, however, both British and American investors recovered their principal. For creditors with diversified foreign bond portfolios who were will- ing to hold out for final settlement, the defaults of the 1930s were not as di- sastrous as typically portrayed. The average nominal internal rate of return (weighted by issue value) was roughly 4 percent on dollar bonds and about Table 2. Rates of Return on Overseas Bonds Issued in the 1920s, by Region (percent) Dollar Sterling Location of borrower bonds bonds Central America 1.46 - South America 3.50 1.44 Germany 1.12 3.61 Other Western Europe 4.83 4.81 Eastern Europe 2.04 1.45 Canada 5.08 5.18 Australia 5.97 5.26 Japan 6.20 5.30 Other Asia - 5.92 Africa 5.62 - not available. Note: Rates of return on sterling bonds are calculated on the assumption that repurchases occurred at market prices. Returns on dollar bonds use market price or par retirements as specified in the bond covenants. Source: Based on samples of 207 dollar bonds and 125 sterling bonds issued between 1920 and 1929 (Eichengreen and Portes 1989a). 84 The World Bank Research Observer, vol. 5, no. 1 (January 1990) 5 percent for sterling issues. Although dollar bondholders settled for approxi- mately half of the contractual interest and sterling bondholders settled for only slightly more, on balance the former did only slightly worse than if they had held U.S. Treasury bonds, and the latter did slightly better. Thus risk premi- ums were nearly sufficient to compensate American bondholders for the risks of foreign lending, and for British bondholders they more than sufficed. Three factors contributed to the rather different experiences of investors in dollar and sterling bonds. First, loans to different categories of debtors fared very differently in the 1930s and 1940s. For both sterling and dollar issues, loans to national governments yielded higher returns than loans to states and municipalities; returns on loans to states and municipalities in turn were higher than those on loans to foreign corporations. And American investors pur- chased a disproportionate share of the speculative bonds issued on behalf of foreign municipalities and corporations. Second, London and New York chan- neled funds in different geographical directions. London specialized in new overseas issues on behalf of the Commonwealth, and it devoted a smaller share of capital to loans to Latin America and Central Europe. It is difficult to assess the extent to which London's behavior is attributable to the sophistication of a more experienced market, to longstanding political and financial ties with the Commonwealth, to preferential British tax treatment of colonial issues, or to the Bank of England's intermittent embargoes on foreign issues. But the combined result of these factors was a significantly lower incidence of default on sterling than on dollar issues. This difference, however, does not by itself account for international differ- ences in rates of return. British creditors also recovered from defaults more suc- cessfully. For the bonds in our sample, the average default on a dollar issue cost the creditors more than four percentage points on the rate of return, whereas the average default on sterling issues cost them only about two per- centage points. Part of the difference may be attributable to the greater effec- tiveness of the British Corporation of Foreign Bondholders; part may also be attributable to the different positions taken by the British and American gov- ernments (see Eichengreen and Portes 1989a). Government Intervention One might think that in the 1930s and 1940s, when banks held only a small share of their portfolios in foreign bonds and hence were not at risk in the event of foreign default, creditor governments might have felt less compulsion to intervene (Eichengreen and Portes 1987, pp. 23-24). This view is supported by the tendency of British officials between the wars to cite the remarks of their well-known nineteenth-century predecessors to the effect that the British gov- ernment was not a debt collector. American officials also argued that the high- er returns on foreign loans represented an implicit acknowledgment of and Barry Eichengreen and Richard Portes 85 compensation for the special risks of lending abroad, and that investors had no justification for seeking assistance. Governments were in fact intimately involved in debt readjustments in the 1930s, though the British were more willing to intervene than the Americans. The British used the Ottawa Agreements of 1932, which provided preferential British market access to the Commonwealth and Empire at the expense of for- eign exporters, to secure favorable treatment of Argentina's sterling debts, in return for tariff preferences under the provisions of the 1933 Roca-Runciman Treaty (see Abreu 1984). The Americans were more reluctant to link trade and debt, especially once President Franklin D. Roosevelt was converted to the ar- guments of Secretary of State Cordell Hull, and the United States began to move toward freer trade with the adoption of the Reciprocal Trade Agree- ments Act in 1934. The links between U.S. Export-Import Bank loans and com- mercial debts varied over time. On some occasions the United States extended Eximbank loans despite the existence of unsettled defaults; on others officials made it clear that Eximbank decisions might be torpedoed by Congress unless progress was made on the debt. Britain and several European governments, however, threatened to block Germany's foreign exchange earnings after its de- fault in 1933, prompting Germany to resume partial or full debt service on its European bonds. The United States declined to adopt a similar stance, and no significant interest was paid to Germany's American creditors for the remain- der of the decade. Until the outbreak of World War II, government interven- tion in trade and finance was limited largely to instances in which public officials had actively promoted the original issues (as in the case of Germany's Dawes and Young Plan loans) or in which the debtor discriminated against one class of national creditors in favor of others. With the approach of the war, pressure was applied to creditors as well as debtors; governments sometimes intervened directly in negotiations if they felt that progress was inadequate. A favorable settlement for the creditors was not viewed as a high priority relative to international trade concessions and inter- national security agreements. Creditors were urged to settle, in no uncertain terms, if their obstinacy stood in the way of these goals. Global Plans In addition to bilateral negotiations, officials in the 1930s suggested calling on the newly established Bank for International Settlements (BIS) to readjust defaulted debts. Other observers advocated an independent facility under the control of private creditors or proposed converting existing debts into new ob- ligations, not unlike the debt-equity swaps and seniority provisions so fashion- able today. Still others emphasized the need to index payments to export revenues or to service the debt in local currency (compare Dornbusch 1988). 86 The World Bank Research Observer, vol. 5, no. I (January 1990) Four global initiatives can be distinguished (Eichengreen 1989b). The first proposed to endow the BIS with the resources needed to resolve the crisis. The BIS had been established in 1930 as part of the Young Plan rescheduling of German reparations; it was logical to propose that it might also take charge of other debt problems. Hubert Henderson, a highly placed British government economic adviser, suggested that the BIS issue unbacked International Certifi- cates to exporting countries to finance debt service payments and other eco- nomic needs. The second scheme, the 1931 Kindersley-Norman Plan, named after two of- ficials of the Bank of England, proposed establishing a new international facil- ity to make loans to countries and corporations unable to obtain them through normal channels. This new international entity was to be capitalized by the leading creditor governments and further financed through the sale of bonds to private investors. The agency would extend new loans to indebted countries unable to obtain them in the market, "reestablishing the credit of the foreign Governments, corporations, etc., to whom the money is lent ... improving the price of their securities ... and the purchasing power of their nationals" (Eichengreen 1989b). Like a mutual fund, it would in effect resell packages of these claims to private investors in the creditor countries. A third scheme, the Beyen and Crena de Jongh plans, offered by two Dutch bankers at the Standstill Conference in 1931-32, addressed the problem of short-term debt. Beyen proposed that short-term debts frozen by debtor gov- ernments be converted into long-term obligations repayable in installments over as long as twenty years. Debtors without enough foreign exchange to meet their current obligations might be permitted to make payments in local currency and to extend preferential treatment to creditor countries that were their best export customers. Crena de Jongh took the idea of payments in do- mestic currency a step further. He proposed a central administrator to issue foreign currency-denominated bonds; the administrator would accept repay- ment of short-term obligations in domestic currency, invest the currency at home, and pay the creditors out of the proceeds. Finally, a variety of proposals were mooted during the preparatory meetings that preceded the World Economic Conference of 1933. In a series of meetings with U.S. officials, Britain proposed a $2,000 million "normalization fund" provided by creditor governments to channel capital to countries that needed foreign funds for debt service and other purposes. Representatives of the debt- or countries endorsed variants of the plan. It is a pessimistic commentary on the global plans under consideration in the 1980s that none of their predecessors bore fruit. Implementing those schemes would have required a serious commitment on the part of the creditor governments. But at each juncture domestic problems diverted their attention from the international debt crisis. In 1931 Britain was battling increasingly in- tense balance of payments difficulties, which ultimately forced it to aban- don the gold standard in September of that year. In 1933 the Roosevelt Barry Eichengreen and Richard Portes 87 administration devalued the dollar and rejected international policy coordina- tion precisely when the World Economic Conference was poised to take up the debt problem. As a result, that conference limited its efforts to addressing the increasingly turbulent exchange rate situation. Even when governments and banks were willing to entertain the possibility of an international debt facility, there remained insoluble problems of finance and control. Under the provisions of the Kindersley-Norman Plan, for exam- ple, the bulk of new finance was supposed to come from private investors. But large investors such as J. P. Morgan and Co. were unwilling to contribute un- less control of the new organization was in private hands. For their part, gov- ernments insisted that control be allocated according to the nationality of the finance. In 1931, for example, the Bank of France insisted that it should have control on the grounds that Paris was the largest single contributor to the fund. Finally, there were debilitating disputes over what countries and what obli- gations to include. At the World Economic Conference, Washington declared all discussion of war debts off limits. The realization that war debt relief was unlikely to be forthcoming weakened the resolve of the Europeans, who would have preferred to discuss the whole interlocking set of debts. Market Solutions In contrast to global plans, repurchases by the issuer of bonds on the mar- ket, often at prices substantially below par, made a useful contribution to the resolution of the debt crisis. Critics argue that buybacks out of reserves may leave the debtor worse off: when bonds are repurchased, the price of the re- maining debt may rise if the country's overall debt servicing capacity is un- changed. Thus the overall value of debt may be unchanged, whereas reserves have been depleted. This position is difficult to reconcile with the preference of debtor governments, many of which opted for bond repurchases. In 1939 the Foreign Bondholders Protective Council estimated that twelve countries in default had repatriated between 15 and 50 percent of their bonds since the be- ginning of the decade. Jorgensen and Sachs (1989) estimate that Bolivia repur- chased 5 percent of its defaulted debt at an average price of 16 cents on the dollar, Chile retired 18 percent at 59, Colombia 22 percent at 22, and Peru 31 percent at 21. Buybacks were controversial. Dornbusch (1988) points to creditors' opposi- tion, while Skiles (1988) notes that "because creditors objected so strenuously to the practice, most of these bond repurchases were carried out through in- termediaries" (p. 33). Another interpretation of the use of intermediaries is that debtors wished to avoid driving up bond prices by signaling their policy through a first-person appearance in the market. But the time-series behavior of bond prices suggests little systematic reaction to repurchases of securities in default. There is scant evidence that debtors sought to depress prices immedi- 88 The World Bank Research Observer, vol. 5, no. t (January 1990) ately before repurchases, or that buybacks had a major impact on price trends. Figure 4, for example, shows high and low market prices for a sterling loan issued by the Chilean government in the 1920s, along with estimates of the vol- ume of repurchases of bonds (valued at face value). The data for a variety of other bonds are similar. The Chilean example is interesting since the country was one of the first to engage in extensive buybacks and was heavily criticized by creditors. Buybacks in the late 1930s, late 1940s, and early 1950s were not preceded by unusual declines in bond prices, nor did they result in major price increases, although in the 1930s and 1940s there was some upward movement in prices. The private attitude of the bondholders' committees toward repurchases dif- fered significantly from their public position on the question (Eichengreen and Portes 1989c). In public they complained that any available foreign exchange should be allocated to the resumption of debt service and repayment of prin- cipal rather than to repurchases of defaulted bonds at a discount. They object- ed to the potential for debtors to manipulate bond prices by declaring their Figure 4. Repurchases and Market Prices for a Sterling Bond Issued by Cbile in the 1920s, 1929-75 (thousands of pounds sterling) 180 170 - 160 - 150 140- 130- 120- 110 Repurchases 100 I - 90 80 A\4. 4o 70 Bar0y Eichengre.n andRichardPo9Lowprice 20 10 1930 35 40 45 50 55 60 65 70 75 Note: Repurchases are quantity timies face value. Calculations on the quantity of buybacks are based on figures in Council of the Corporation of Foreign Bondholders (various years) on the face value of issues still outstanding. Barry E3icben green and Ricbard Portes 89 e inability to service their debts and then to turn around and retire the debt as soon as the prices fell. In private, though, the committees were much more re- ceptive to repurchases, especially if they were accompanied by the resumption of at least partial debt service payments and when it was apparent that the creditors were unlikely to receive a better offer. Numerous examples can be cited. In December 1936, following Chile's de- cision to limit repayments and interest, representatives of U.S. bondholders re- fused to accept the reduced interest payments that had been deposited in New York. Chile then borrowed from Schroeders in London on this security in order to purchase bonds in the market. The Council of the Corporation of For- eign Bondholders (various years; dates of minutes serve to locate references) did not object; its minutes for December 17, 1936, note that "the Finance Minister so far has obtained and used $4 million for purchases." As part of the 1940 temporary settlement recommended by the council, the minutes of March 14, 1940, report that the Brazilian authorities devoted "at least $400,000 in each of four years" to purchases in the English market. In the 1941 Colombian negotiations, the council was asked to agree that all the sterling debt be repurchased in the market. Although it objected to the "disastrous pre- cedent" that might be set, its members admitted on December 14 that "we have long become acclimatized to the idea of a debtor being allowed to devote some sums to amortization, provided that he is paying an agreed percentage on the principal of his debt." Typically, the bondholders argued that purchases below par should take place only if at least partial service was being paid. Thus on May 31, 1944, the council decided "to refuse to consider a settlement of the Ecuadorean debt by means of a purchase offer at below par" unless at the same time the bondholders were offered an option of partial service on the debt. Perhaps the most revealing exchange surrounded the 1937 hearings of the U.S. Securities and Exchange Commission investigation into defaulted bonds (U.S. Securities and Exchange Commission 1937). The commission had recom- mended that in order to curb the repatriation of bonds by defaulters, measures be introduced to restrain bankers and brokers from dealing on behalf of gov- ernments in default. The Council of the Corporation of Foreign Bondholders, meeting on May 31, 1937, noted that such restraints on repurchases would be met with "strong and... effective criticism on the ground that, by limiting the market in such bonds, it would act detrimentally to the bondholders." Conclusion The debt crisis of the 1930s sheds light on a number of aspects of the current debt situation. It highlights the advantage, from the viewpoint of financial sta- 90 The World Bank Research Observer, vol. 5, no. 1 (January 1990) bility, of dispersing foreign obligations across large numbers of private inves- tors rather than concentrating them in the hands of financially vulnerable commercial banks. While reminding us of the sophistication of the bond mar- ket, it shows that bondholders historically have been no more adept than banks in avoiding loans to countries with a repeated tendency to default. It alerts us to the central role creditor governments can play in promoting or impeding a negotiated resolution of the crisis. Finally, it reminds us of the serious obstacles to the implementation of any global plan for resolving the debt crisis and un- derscores the useful contribution to resolution offered by market-based debt reduction schemes. Notes Barry Eichengreen is a professor of economics at the University of California at Berkeley and a research fellow of the Centre for Economic Policy Research (CEPR), London. Richard Portes is a professor of economics at Birkbeck College, University of London, and director of the CEPR. This is a nontechnical summary of a paper presented to the World Bank Conference on the Debt Crisis (January 1989) and published in full as Eichengreen and Portes 1989b. We thank Andrew Brociner, Tony Humm, Mara Leonaki, Anita Santorum, and Carolyn Werley for re- search assistance; Albert Fishlow, Marilyn Skiles, and Steven Webb for helpful comments; the Council of the Corporation of Foreign Bondholders for permission to cite materials from their archives; and the Institute of Business and Economic Research at the University of California at Berkeley for logistical support. 1. For Britain we constructed estimates for all 125 overseas bonds offered for subscription and listed by the Stock Excbange Yearbook, whereas for the United States we drew a stratified sam- ple of 250 foreign bonds from the list of more than 1,400 such bonds published in U.S. Depart- ment of Commerce (various years). After adjusting for the actual price paid by the purchaser, we then tracked interest and amortization payments until the bond issue was extinguished and calculated the nominal own-currency internal rate of return (see Eichengreen and Portes 1988). The number of bonds included in the regression analysis was limited by the availability of ex- planatory variables. 2. Edwards's results for bank loans also are basically consistent with these results. In addition, see Ozler (1988), who finds evidence that spreads decline with the amount of time the borrower has been in the market. 3. In an analysis using different data (see Eichengreen 1989a) there was some apparent ten- dency of the spread to rise over the course of the 1920s, suggesting that investors recognized the increasingly risky nature of foreign loans. 4. Only the coefficient on loans to corporations differs significantly from that on national gov- ernment loans, however. 5. Under the provisions of the Hay-Bunau Varilla Treaty of 1904, the U.S. government was permitted to supervise the expenditure of Panamanian government loans placed in the United States. The Platt Amendment gave Washington the right to object to "improvident or otherwise objectionable fiscal policy" in Cuba. The Dominican Republic was under U.S. military admin- istration until 1924, and after that the United States retained the right to object to changes in tariffs and public debt. Haiti was under U.S. martial law from 1916 to 1931 (Angell 1933, pp. 8-27). Barry Eichengreen and Richard Portes 91 Appendix Table 1. Relation of Loan Spreads to Loan Characteristics, 1921-29 Dollar bonds Sterling bonds Category coefficient coefficient Economic indicator Debt-to-export ratio -0.07 (0.24) 0.0007 (0.0007) Trade surplus -0.20 (0.25) -0.56 (0.40) Budget surplus -0.80 (0.40) -0.12 (0.41) Kind of borrower Municipal 0.11 (0.18) n.a. n.a. State 0.07 (0.17) n.a. n.a. Corporate 0.76 (0.16) n.a. n.a. National bank 0.04 (0.34) n.a. n.a. Other bank 0.14 (0.28) n.a. n.a. a Dominion central government n.a. n.a. -1.52 (0.38) a Dominion local government n.a. n.a. -1.25 (0.69) Foreign corporate stocks n.a. n.a. -0.45 (0.25) Year of issue 1921 -0.12 (0.26) n.a. n.a. 1922 0.04 (0.25) n.a. n.a. 1923 -0.01 (0.36) n.a. n.a. 1924 0.17 (0.29) 0.18 (0.48) 1925 0.19 (0.27) 0.39 (0.52) 1926 0.40 (0.26) 0.09 (0.44) 1927 0.38 (0.26) -0.01 (0.45) 1928 0.07 (0.27) 0.22 (0.45) 1929 0.06 (0.28) 0.20 (0.48) Location of borrower Western Europe -0.73 (0.18) 0.33 (0.26) Eastern Europe 1.21 (0.26) 1.15 (0.33) Canada -1.38 (0.19) -0.69 (0.85) Central America -0.79 (0.29) n.a. n.a. South America 0.50 (0.21) 0.38 (0.67) Japan 0.05 (0.36) -0.03 (0.38) Australia -0.91 (0.45) -0.29 (0.67) Econometric constant 2.61 (0.31) 1.97 (0.46) n.a. = not applicable. Note: A spread is the difference between the projected yields of a bond or loan at time of purchase and the projected yield of a risk-free loan. Figures in parentheses are standard errors of the estimates. Dependent variable is spread. For dollar bonds 1920 is the omitted year; for sterling bonds 1923 is omitted. The omitted country is Germany; the omitted category of borrower is national government. For dollar bonds, the R-squared is 0.69, the standard error of regression is 0.00654, and the number of observations is 207. For sterling bonds, the R-squared is 0.91, the standard error of regression is 0.3876, and the number of observations is 43. a. Dominion borrowers are self-governing member countries of the British Commonwealth of Nations. Source: Authors. 92 The World Bank Research Observer, vol. 5, no. 1 (January 1990) References Abreu, Marcelo de Paiva. 1978. "Brazilian Public Foreign Debt Policy, 1931-1943." Brazilian Economic Studies 4:105-40. - 1984. "Argentina and Brazil during the 1930s: The Impact of British and American In- ternational Economic Policies." In Rosemary Thorp, ed., Latin America in the 1930s. London: Macmillan. Angell, James W. 1933. Foreign Financial Policy of the United States. New York: Council on Foreign Relations. Cardoso, Eliana, and Rudiger Dornbusch. 1989. "Brazilian Debt Crises: Past and Present." In Barry Eichengreen and Peter Lindert, eds., The International Debt Crisis in Historical Perspec- tive. Cambridge, Mass.: MIT Press. Council of the Corporation of Foreign Bondholders. Various years. Annual Report. London. Darity, William, and Bobbie Horn. 1988. The Loan Pushers: The Role of Commercial Banks in the International Debt Crisis. Cambridge, Mass.: Ballinger. De Cecco, Marcello. 1985. "The International Debt Problem in the Interwar Period." Banca Na- zionale del Lavoro Quarterly Review 152:45-64. Diaz-Alejandro, Carlos. 1983. "Stories of the 1930s for the 1980s." In Pedro Aspe Armella and others, eds., Financial Policies and the World Capital Market. Chicago: University of Chicago Press. Dornbusch, Rudiger. 1988. "Mexico: Stabilization, Debt and Growth." Economic Policy 7:231-83. Edwards, Sebastian. 1986. "The Pricing of Bonds and Bank Loans in International Markets: An Empirical Analysis of Developing Countries' Foreign Borrowing." European Economic Re- view 30:565-89. Eichengreen, Barry. 1989a. "The U.S. Capital Market and Foreign Lending, 1920-1955." In Jeffrey Sachs, ed., Developing Country Debt and Economic Performance. Chicago: University of Chicago Press. - 1989b. "Resolving Debt Crises: An Historical Perspective." In Sebastian Edwards and Fellipe Larrain, eds., Debt, Adjustment and Recovery. Oxford: Blackwell. Eichengreen, Barry, and Richard Portes. 1986. "Debt and Default in the 1930s: Causes and Con- sequences." European Economic Review 30:599-640. 1987. "The Anatomy of Financial Crises." In Richard Portes and Alexander Swoboda, eds., Threats to International Financial Stability. Cambridge: Cambridge University Press. - . 1988. "Les prets internationaux dans l'entre-deux-guerres: le point de vue des porteurs de titres." Economie appliquee 41:741-74. - . 1989a. "After the Deluge: Default, Negotiation and Readjustment in the Interwar Years." In Barry Eichengreen and Peter Lindert, eds., The International Debt Crisis in Histor- ical Perspective. Cambridge, Mass.: MIT Press. - . 1989b. "Dealing with Debt: The 1930s and the 1980s." In Ishrat Husain and Ishac Diwan, eds., Dealing with the Debt Crisis. Washington, D.C.: World Bank. . 1989c. "Settling Defaults in the Era of Bond Finance." World Bank Economic Review 3, no. 2 (May): 211-39. Feis, Herbert. 1950. Diplomacy of the Dollar: First Era, 1919-1932. Baltimore: Johns Hopkins University Press. Fishlow, Albert. 1972. "Origins and Consequences of Import Substitution in Brazil." In L. de Marco, ed., International Trade and Development. New York: Academic Press. . 1985. "Lessons from the Past: Capital Markets during the 19th Century and the Interwar Period." International Organization 39:383-439. Guttentag, Jack, and Richard Herring. 1985. "Commercial Bank Lending to Developing Coun- tries: From Overlending to Underlending to Structural Reform." In G. W. Smith and Barry Eichengreen and Richard Portes 93 J. T. Cuddington, eds., International Debt and the Developing Countries. Washington, D.C.: World Bank. Harris, C. R. S. 1935. Germany's Foreign Indebtedness. London: Oxford University Press. Jorgensen, Erika, and Jeffrey Sachs. 1989. "The Costs of Default: The Settlement of Interwar Loans to Latin America." In Barry Eichengreen and Peter Lindert, eds., The International Debt Crisis in Historical Perspective. Cambridge, Mass.: MIT Press. League of Nations. Various years. Annual Statistical Yearbook. Geneva. Lewis, W. Arthur. 1949. Economic Survey, 1919-1939. London: Allen & Unwin. Lindert, Peter, and Peter Morton. 1989. "How Sovereign Debt Has Worked." In Jeffrey Sachs, ed., Developing Country Debt and Economic Performance. Chicago: University of Chicago Press. Madden, John T., and Marcus Nadler. 1929. Foreign Securities. New York: The Ronald Press. Ozler, Sule. 1988. "Have Commercial Banks Ignored History?" Working Paper 498. University of California, Los Angeles, Department of Economics. Schuker, Stephen A. 1988. "American 'Reparations' to Germany, 1919-33: Implications for the Third-World Debt Crisis." Princeton Studies in International Finance 61. Skiles, Marilyn. 1988. "Latin American International Loan Defaults in the 1930s: Lessons for the 1980s?" Research Paper 8812. Federal Reserve Bank of New York. Thorp, Rosemary, ed. 1984. Latin America in the 1930s. London: Macmillan. U.S. Department of Commerce. Various years. American Underwriting of Foreign Securities. Washington, D.C.: GPO. U.S. Securities and Exchange Commission. 1937. Report on the Study and Investigation of the Work, Activities, Personnel and Functions of Protective and Reorganization Committees, Part V: Protective Committees and Agencies for Holders of Defaulted Foreign Bonds. Washington, D.C.: GPO. U.S. Congress. Senate. 1932. Hearings before the Committee on Finance, Sales of Foreign Secu- rities. 72nd Cong. Washington, D.C.: GPO. 94 The World Bank Research Observer, vol.5, no. 1 (January 1990)