Barry Eichengreen and Harry Bank exposure, market discounts country de - PDF document

Barry Eichengreen and Harry Bank exposure, market discounts country debt. Cambridge, Mass.: National Bureau Economic Research. This paper growing body literature exploring the determi- into and market countries. One the main points over when what extent “pull” irrationality operate the rapidly for developing the fundamental characteristics the issuing country. this case could assume stable appetite for emerging careful evaluation past, present, investors turning asset class certain threshold. Here the bonds depends greater extent the strength demand for the market bond global liquidity the risk-free rate market bonds. These studies focus varyingly payments, trade balances, actual measured, and other Variation in launching. This emphasis should make the Street, but exciting as predict variation spreads as bonds the secondary market. The The paper interesting findings follow-up work. model includes behavior. These variables shaping credit worthiness such as the debt growth. These fundamentals are motivate investors market investment. also included control for changes in the risk-free investment rate and the explains cross-national variation better differences between Sylvia Maxfield is research associate Latin American What Explains Changing Spreads bago, Tunisia, included in Mauritius, Saudi John A. Penicook Jr. market debt: Practi- cal portfolio considerations. Presented Center Conference Market Capital balance-of-payments crisis: land, Department Economics. Mimeo. and Frank The credit rating Bank of New York Quarterly Review sovereign credit ratings. York Economic Policy Review 2 (October): Cline, William International debt reexamined. Cline, William R., and market lending. national Finance. and bank loans in international European Economic Review Eichengreen, Barry, tion, and the interwar international debt crisis historical perspective, Barry Eichengreen bridge, Mass.: secondary markets believe in life and I. World Bank. Haque, Nadeem, and Donald The economic section bias as a specification error. been affected? Dealing with the debt D.C.: World van Kleist. 1997. Settlements. Mimeo. and Ashoka Mody. 1997. usefulness of private for foreign investment decisions. Policy Research Working D.C.: World sovereign risk. Moody’s Barry Eichengreen Ashoka Mody reported quarterly sources for these variables World Bank’s the International tional Financial edt WBDT series called gnp WBDT series called 4, Debt Indicators IFS, series number series number “99b.p” IFS, series number “99b” IFS, series dres WBDT table A3.3 sue; table A4.4, agreements with mercial banks, exact series specific sources included: Israel, Korea, Singapore, obtained from the Central Brazilian Ministry busch in Papers on the Census Department, Hong Kong, August for Taiwan December 1996/January District, the the Central are Argentina, Bahrain, Barbados, Bolivia, Brazil, Chile, Colombia, Congo, Costa Rica, Croatia, prus, Czechoslovakia, Czech Republic, Estonia, Guatemala, Kong, Hungary, India, Indonesia, Poland, Qatar, Romania, Russia, Singapore, Slo- What Explains Changing Spreads on Market Debt? The bond obtained from Capital includes (a) (in basis a per- maturity; (b) the amount issue (millions the maturity in years; (d) a dummy variable taking the “privately placed” private party. following variables from Bondware. LAC Latin Eastern Europe national product foreign reserves minus gold domestic product domestic product budget deficit to denote a debt previous year domestic product treasury bond have studied the determinants spreads on ket debt using a framework pricing decisions controls for the spread issue while the spread results confirm the importance results for different, especially that the American issuers market conditions. There this regional dif- as the signs of a maturing market well-defined yield exploiting scale economies most striking changes in market sentiment ously related in observable issuer characteristics explanation for changes over time in value of In important periods, such as crises, blanket shifts in the domi- The obvious governments should exercise economic policy strategy inflows of capital intermediated bond market. foreign credit may available when shifts in their also shift against beyond their control, making finance large adjustment. There insuring against the bond market include foreign direct investment, insure against the sudden evaporation foreign financing a painful adjustment burden limit the current account empirical work for this of large-scale dramatic spread compression. There but a wilderness warning that this affairs could sudden end any time. Changing Spreads Market Debt? generate within-sample forecasts for forecasts for actual spreads (in basis presented in table 4.9. in table that the market growing increasingly exuberant. Actual observation that primary launch spreads secondary market prices. Following decline in issues as secondary spreads primary issues to the at higher second half of 1996, divergence between predicted spreads narrowed, the “irrational of 1997. next repeated (as in table in-sample predictions for predictions for fall in evident for Latin America. predicted with the gap increasing up of 1997, falls. With the East Asian crisis spreads. For East Asia, crisis, with actual spreads rising rapidly above predicted second half Sample for Sample for Latin East Latin East America Asia America Asia represent differences between predicted Barry Eichengreen 1995, when only high-quality issues to the that secondary spreads skyrocketed in sentiment against developing behavior, as discussed period preceding Mexican crisis immediately following secondary market spreads launch spreads first panel table 4.8 (again, according to our positive change decline in This decline in broad factors. First, market sentiment moved strongly against emerging markets: the spread equation primary spreads consider- ably (in parallel with markets). Second, this a change rates. For the market was willing the market. Third, the fundamentals issuers coming the market the post- issuers in the market more benign crisis. An exception more negative American issues private issues deutsche mark-denominated issues market spreads in 1996 first half of 1997, a trend was viewed exuberance.” But primary spreads crease. But once this increase table 4.8). Overall improves. Again, as credit worthy issuers the market, this shift market sentiment reduced sensitivity changes in est rates, higher premium the poorer risks entering the Market Irrationally Exuberant second half at least first half some observers the market issuers entered estimates allow with predictions. first estimated the equation for Latin up to Changing Spreads the contribution the contribution markets regard credit worthiness of changes in impact on spreads blanket changes in of favor, changes in the independent independent (- @*)I, which can be thought of as changes in characteristics (as, for example, issues of risky borrowers be regarded distinguish changes average inverse Mills of as a measure sample selectivity) versus changes average value the other ables. Finally, distinguish changes in effect of inverse Mills be included) changes in effect of regressors. Intuitively, that the The decompositions for two after the Mexican crisis, during the spread com- in log declined following Mexican crisis; The Mexican (1991: Ql-1994: treasury rate Inverse Mills Irrational Exuberance? Inverse Mills ratio Barry Eichengreen Ashoka Mody The supply and demand vary noticeably that the coefficients for issues in yen significant negative signs for Thus, for East Asia, the deutsche the East Asian issuers time their error terms the spread probit equations. For Latin behaves much credit rating probability of associated with For East Asia, in An interpretation Asian bonds to the with higher spreads. to obtain investors, for reasons of East Asian are not A central question in the literature whether changes in changes in whether there been changes pricing behavior over time. credit rating, service ratios, the reductions spreads that occurred between be reason pricing behavior about the otherwise explicable shift in pricing in favor of developing country debt, then there a priori a sudden a corresponding curtailment Fundamentals and between two periods, Oxaca decomposition: right side of the contribution their economic the contribution What Explains Changing Spreads For East Asia, on the rate persists after correcting this finding, the probit the spreads equation together, for this leads to an For Latin America, higher interest rate reduces spreads (after correction selection bias). interest rates the demand their price therefore declines (equivalently, the spread East Asia, interest rates appears to supply curve the spread. that appear both the spreads equations, it is again useful whether they work in directions. In regressions using full sample bonds, a larger credit residual (a credit rating, probability of the spread, with inferior credit ratings find it Latin America, higher debt In Latin has a weak positive rescheduling debt are apt to the capital market same time raising the spread that successful issuers dummy variable for Latin America behaves similarly: Latin American borrowers issue more bonds but although the magnitude and the coefficient of the errors two equations-varies sample, it that unob- to the also reduce as unobserved coefficients work in offsetting influence mainly bonds. For countries that recently rescheduled tend therefore supply the corre- sponding outward supply reduces the spread. regional dummies suggest continent-specific characteristics, unusually high volume this works, drive down their issues the spreads invoke this for in a market with imper- possible for the demand bend back ments along the demand along a discussion in demand, not rationing. The of only terminological importance, but its significance a “backward rather than a “backward bending supply curve.”) Barry Eichengreen Ashoka Mody the Selectivity Correction Latin America Latin and East Asia America Private placement Debt service/ Private issuer Latin America Log likelihood (1 3.17) Numbers in parentheses are reduces observed this interpretation correct, then tive sign be significantly comparing the coefficients in tables that the treasury rate significant at idea that selection bias role in driving observed result. What Explains Changing Spreads, 1991-96, America Latin East Eastern America Asia treasury rate for debt Public issuer Latin America 14.1 18) (-2.1 11) parentheses are interest rate market spreads (see, e.g., results presented with this finding. Among the negative sign interest rates Since higher interest rates issuance (as in probit above), it that the the market some unobserved credit features them attractive Barry Eichengreen Ashoka Mody (henceforth referred as Latin America), East Asian, treasury rate high, the volume of issues declines This finding “search for yield” market-the idea interest rates fall there a greater for higher-yielding emerging bonds. This the strength relationship varies across decline in issues following treasury rates evident for East Asia, that the East Asian supply of interest rates rise their borrowing costs go up.’* The higher the credit rating residual, all regions except Asia, where effect is insignificant. Higher probability of issue especially in Latin America, For East Asia, higher debt service probability of the supply side, higher deficits, Eastern Europe, full model with once again reveal differences pricing across regions. For Latin indicates plausibly larger issues (consistent with and the greater liquidity larger issues the secondary market). Latin American private placements spreads equation positive sign, with the fact issued in markets with stringent disclosure requirements. Disclosure require- for private placements investors capable associated risk. for this reason are private placement because infor- issuer characteristics in that spread. Latin American issues typically having been relatively little evidence well-defined yield curve until The picture for East Asian issuers, who higher spreads. large negative coefficient in spreads equation dummy variable for Israel, reflecting its issues are guaranteed ernment. Supranational bonds (guaranteed entities beyond have lower Recent studies somewhat surprisingly, negative influence demand and consider the issues decisions together. the Probability America Latin East Eastern Asia America Asia for debt Public issuer Private issuer Latin America Public issuer Private issuer American Interactions Latin America Asia Latin America Asia Eastern Private placement rating residual growth rate Latin America Deutsche mark 319.10 384.25 152.18 265.82 174.77 165.96 195.95 257.47 6.71 5.81 6.79 6.50 6.78 6.49 6.51 6.75 37.12 44.46 40.42 37.10 28.97 62.15 63.14 40.67 33.34 2.05 4.87 -2.15 16.41 0.13 0.22 0.30 0.28 0.19 0.38 0.22 0.48 0.28 0.35 0.62 0.28 0.38 0.42 0.51 0.00 0.00 0.00 0.00 0.02 0.71 0.66 0.00 0.00 0.26 0.34 0.47 0.32 0.31 0.44 0.37 0.30 The spread its relationship spreads are observed defined by: determines the desire willingness of lend (which the error terms a standard the fitted probabilities the selection equation probit requires information on those address this problem For each for three types of issuers: sovereign, For each types did the market, zero. Table highlights the character- likely when interest rates better quality credit (with credit rating residuals when reserves public-sector demand for estimate equations using maximum maximum-likelihood estimates verge nicely. performing sensitivity tests using maximum-likelihood estimates converge; in this case perform the conventional two-step procedure, probit model ratio (with full-information are obtained using difference in derivative of the probability distribution function to a the independent average values facilitate interpretation separate equations for for the Latin American estimate a similar model Barry Eichengreen these ratings on a international bankers, numerical value ranging indicating zero probability default). For each observation the most recent credit rating the bond's date rather than include the the residual the credit rating on utilize only the orthogonalized component because the credit rating included separately as and Packer et al. preliminary look at the issuers in the Caribbean, Latin American Asian issues separately as linear relationship the form the dependent the logarithm the spread, a random bond characteristics amount, and global economic conditions the ten-year rate credit rating residual, the GDP, the ratio the previous growth rate; including the whether the borrower other public, the currency the bond conditions under this linear equation of issuer points in The estimated equation credit rating 12.15 (reserveslGNP) (debt reschedule) t-statistics are in parentheses Changing Spreads Market Debt? points for spreads; numbers for East Eastern Latin Middle Africa Asia Bonds with Bonds with Bonds with 166 94 205 226 Sovereign issues the other issues, which although the private issues bonds’ greater for all regions except confirm this regression analysis. variety of proxy for (at time of issue). proxy for credit from the Global Development Finance World Economic the ratio external debt national product (GNP), exports, a dummy variable for whether country con- restructuring agreement previous year, budget deficit as a Sovereign credit 9. This with the rating agencies, are reluctant private borrowers higher the grounds with transfer problems private domestic foreign exchange; Launch Spreads, (basis points) 1991 1992 1994 1995 1996 1997 197 151 140 94 173 145 125 149 178 152 332 266 Latin America 448 410 369 413 216 219 South Asia 190 233 356 288 289 Bonds with 52 118 163 170 302 165 332 202 203 342 369 1,697 blank indicates that no first calculating using the the time of issue then subtracting the “riskless” sovereign security the currency source provides the period issued grew before falling Mexican crisis market conditions in 1995). Bond issuance rose sharply in a banner over 1,300 bonds. publicly available for ancillary information country conditions on data for which 1,033 usable use 277 The main Latin a distant second with bonds (Korea a leading Eastern Europe in this especially active year for the region. America has also highest spreads. years from first time In con- trast, East Asian spreads always less they rose their highest typically higher for Latin America. Eastern consistently been between those East Asia. summarizes sample characteristics distinguished: sovereign, private borrowers, public entities in bonds. Sovereign Eb86 ib86 eb16 1b16 €696 EbE6 5 IbE6 B 0 8 E P 0 088O 0 0 52 W Ti- cu 8 0 2zc!z 0 0 0 0 EbZ6 3 IbZ6 8 EbI6 $ E Ib16 ' 6 a c) C 8 v ._ 7 w E Barry Eichengreen proportionately, for example, because high-risk borrowers discouraged from coming the market. the determinants impact of factors on the the market therefore be of selectivity next sections methodology designed these problems. Sample Characteristics The bonds income securities with face value, emerging market borrowers, placed on international markets and are dollars). Although consists of market), over period a growing number tries floated market for foreign issuers private placement market (under provisions Rule 144a); there also some growth The bonds are who commit with investors. trade on secondary market Our data “launch” spreads for bonds countries, obtained IMF. That launch spreads launch spreads move differently over time secondary markets. While there some tendency for primary spreads secondary spreads four quarters, the series frequently In poor primary spreads indeed, some- times they fall. This tendency for the most credit worthy borrowers the market. investing in emerging secondary market spreads, may the opposite launch spreads riskier borrowers leaving only low-risk, low-spread borrowers This makes it different classes of borrowers. Numbers, Spreads, international bond issues with convertible rise slightly exclude these from the grounds and relationship warrant a separate What Explains Changing Spreads variables for private issues in Brady plan debt reduction schemes). Because his sample his estimates economic variables, only growth differ significantly from zero at 1997) uses data and somewhat longer list of otherwise subject European bor- emerging markets, raising questions tional bonds last time bond markets for international lending. But their sample includes industrial countries the informa- the market has time limits this study for present purposes. bond spreads in 1976-80, since there in developing-country for only thirteen countries should expect pricing same over time.6 Edwards analyzes: all emerging market debt the end than two-thirds mid-1990s (Bernstein van Kleist spreads on and Standard a summary measure industrial country interest rates (as a Latin American than on comparable Asian result suggestive segmentation. Surprisingly they find that the on industrial country interest rates insignificantly different country participation the bond market has time suggests least squares ity bias. same changes in economic traders often that a rates raises emphasized above accentuating the trade-off between acquisition emphasized Barry Eichengreen Section 4.5 basic results for both the issue decision pricing behavior. Section 4.6 the contributions pricing behavior over time. Shifts in pricing behavior also examined estimated sample and out sample. Section analysis for market debt. explosive growth market debt, there have been systematic studies the secondary market for bank which there significant number studies (reflecting the dominance But there of risk therefore pricing behavior differ between bonds. Models that banks may have a comparative advantage in assembling their clients into the secondary market their claims. Pecking-order finance suggest claims with different degrees levels of and that prices should bear a different relation- Bonds typically have senior status, more variable. Thus, the studies of in bank to the bonded debt. the market for developing-country bonds are Cantor and spreads on for forty-nine tries in income, gross domestic growth, inflation, external balance, ternal debt, economic development to the of Moody’s this study include it considers developed as developing countries, it analyzes only and not also private that none the macroeconomic statistically significant credit ratings included. Cline rate as as government borrowers. he studies only debted countries, whose representativeness siders only four economic determinants rate spreads: per capita with dummy Huizinga (1989) Secondary Spreads Interest Rate Barry Eichengreen regard the post-1995 surge portfolio capital its equally sudden halt (along with toward the end of 1997. If increased lending spread compression half of reflected improved the borrowing countries, then there long as governments are rewarded for putting national on a firmer footing, there are the trend toward for emerging market mainly the credit conditions in the money centers, then there effects of the eventual tightening financial market con- in the advanced industrial countries. more troubling the increased capital reduced spreads period centered case there sudden shift back corresponding curtailment on about developing-country bonds the years period that spans the recent episode bonded debt. In contrast previous studies, sample selection. sue decision debtors and the pricing decision minimize selectivity treating the Overall, the results confirm higher credit quality leads spreads. This the presumption market discriminates same explanatory variables the principal debt- America, East Asia, Eastern Europe). And shifts in market sentiment rather than break point fourth quarter 1994, when Mexican crisis lower levels with 1995:Ql Mexican crisis changes in spreads adverse shifts market sentiment more fundamentals. The same the subsequent compression, favorable shifts market sentiment. the literature. Section tional Monetary allowed them Mexican crisis scot-free. is likely again come have little effort needed among bonds that the process. Investors, in incomplete knowledge of countries’ economic a practice conducive powerful incentives be informed evidence they exist between with different advanced their decline in emerging market spreads that took place between early 1997. secondary-market spread between developing Advocates of functioning markets” rededicated themselves Mexican cri- to put their fiscal, monetary, affairs in order. Better policies, implied reductions market debt tals improved sufficiently decline in investors in their exuberance may have been relationships between that the costly information for portfolio diversification, the information individual investor about a particular country. (This depends on the assumption that the cost of acquiring information about a country size of disproportionately affect funds across countries prices of particular bonds portfolio diversification facilitates these circumstances, investors might fail on a particular bond a gradual economic conditions, could lead one country could lead investors their expectations the prospects with superficially similar characteristics. money” results from “frictions” in information leading herd-like behavior and out countries on the are not obviously attractive definitely “no-no’s.’’ their model with diversify limits incentives for information acquisition. Kehoe argue their model predicts types of countries that predict only that hot flows will occur. Herd behavior observed for foreign direct investment Spreads continued the third which they One oft-heard justification for this Mexican rescue for investors concern themselves with a potential credit worthiness, at least left them with this package engineered and the Interna- Syndicated Loans Markets (gross, 1991 1992 1993 1994 1995 1996 1997 Portfolio equity Western Hemisphere Europe and Central Middle East Syndicated loan 7.2 11.9 11.2 16.4 13.9 24.3 62.7 57.6 101.9 7.1 12.9 28.8 18.0 23.1 47.1 25.3 43.1 1.9 1.6 42.5 43.0 74.9 79.7 11.9 14.3 82.4 129.5 What Explains on Emerging Ashoka Mody value of the course a major for developing significant implications for the international capital markets. The the bonds billion in billion in 1992, billion per year in billion in of much market’s ability Some observers Barry Eichengreen Pardee Professor Political Science University of California, research associate Bureau of Economic Research, the Centre Ashoka Mody visiting professor and man- agement at School, University this paper Specialist for Finance with Development Prospects authors are Dunaway, Anne Jensen, Ken Wood of who helped assemble Ananda Chanda and for sterling research assistance; early informal discussion of results at the International discussant. Robert Hill, David Shapiro also provided useful comments. are those authors and preliminary estimate for three quarters anticipate a the turmoil Asian financial markets. Barry Eichengreen and Harry Bank exposure, market discounts country debt. Cambridge, Mass.: National Bureau Economic Research. This paper growing body literature exploring the determi- into and market countries. One the main points over when what extent “pull” irrationality operate the rapidly for developing the fundamental characteristics the issuing country. this case could assume stable appetite for emerging careful evaluation past, present, investors turning asset class certain threshold. Here the bonds depends greater extent the strength demand for the market bond global liquidity the risk-free rate market bonds. These studies focus varyingly payments, trade balances, actual measured, and other Variation in launching. This emphasis should make the Street, but exciting as predict variation spreads as bonds the secondary market. The The paper interesting findings follow-up work. model includes behavior. These variables shaping credit worthiness such as the debt growth. These fundamentals are motivate investors market investment. also included control for changes in the risk-free investment rate and the explains cross-national variation better differences between Sylvia Maxfield is research associate Latin American What Explains Changing Spreads bago, Tunisia, included in Mauritius, Saudi John A. Penicook Jr. market debt: Practi- cal portfolio considerations. Presented Center Conference Market Capital balance-of-payments crisis: land, Department Economics. Mimeo. and Frank The credit rating Bank of New York Quarterly Review sovereign credit ratings. York Economic Policy Review 2 (October): Cline, William International debt reexamined. Cline, William R., and market lending. national Finance. and bank loans in international European Economic Review Eichengreen, Barry, tion, and the interwar international debt crisis historical perspective, Barry Eichengreen bridge, Mass.: secondary markets believe in life and I. World Bank. Haque, Nadeem, and Donald The economic section bias as a specification error. been affected? Dealing with the debt D.C.: World van Kleist. 1997. Settlements. Mimeo. and Ashoka Mody. 1997. usefulness of private for foreign investment decisions. Policy Research Working D.C.: World sovereign risk. Moody’s Barry Eichengreen Ashoka Mody reported quarterly sources for these variables World Bank’s the International tional Financial edt WBDT series called gnp WBDT series called 4, Debt Indicators IFS, series number series number “99b.p” IFS, series number “99b” IFS, series dres WBDT table A3.3 sue; table A4.4, agreements with mercial banks, exact series specific sources included: Israel, Korea, Singapore, obtained from the Central Brazilian Ministry busch in Papers on the Census Department, Hong Kong, August for Taiwan December 1996/January District, the the Central are Argentina, Bahrain, Barbados, Bolivia, Brazil, Chile, Colombia, Congo, Costa Rica, Croatia, prus, Czechoslovakia, Czech Republic, Estonia, Guatemala, Kong, Hungary, India, Indonesia, Poland, Qatar, Romania, Russia, Singapore, Slo- What Explains Changing Spreads on Market Debt? The bond obtained from Capital includes (a) (in basis a per- maturity; (b) the amount issue (millions the maturity in years; (d) a dummy variable taking the “privately placed” private party. following variables from Bondware. LAC Latin Eastern Europe national product foreign reserves minus gold domestic product domestic product budget deficit to denote a debt previous year domestic product treasury bond have studied the determinants spreads on ket debt using a framework pricing decisions controls for the spread issue while the spread results confirm the importance results for different, especially that the American issuers market conditions. There this regional dif- as the signs of a maturing market well-defined yield exploiting scale economies most striking changes in market sentiment ously related in observable issuer characteristics explanation for changes over time in value of In important periods, such as crises, blanket shifts in the domi- The obvious governments should exercise economic policy strategy inflows of capital intermediated bond market. foreign credit may available when shifts in their also shift against beyond their control, making finance large adjustment. There insuring against the bond market include foreign direct investment, insure against the sudden evaporation foreign financing a painful adjustment burden limit the current account empirical work for this of large-scale dramatic spread compression. There but a wilderness warning that this affairs could sudden end any time. Changing Spreads Market Debt? generate within-sample forecasts for forecasts for actual spreads (in basis presented in table 4.9. in table that the market growing increasingly exuberant. Actual observation that primary launch spreads secondary market prices. Following decline in issues as secondary spreads primary issues to the at higher second half of 1996, divergence between predicted spreads narrowed, the “irrational of 1997. next repeated (as in table in-sample predictions for predictions for fall in evident for Latin America. predicted with the gap increasing up of 1997, falls. With the East Asian crisis spreads. For East Asia, crisis, with actual spreads rising rapidly above predicted second half Sample for Sample for Latin East Latin East America Asia America Asia represent differences between predicted Barry Eichengreen 1995, when only high-quality issues to the that secondary spreads skyrocketed in sentiment against developing behavior, as discussed period preceding Mexican crisis immediately following secondary market spreads launch spreads first panel table 4.8 (again, according to our positive change decline in This decline in broad factors. First, market sentiment moved strongly against emerging markets: the spread equation primary spreads consider- ably (in parallel with markets). Second, this a change rates. For the market was willing the market. Third, the fundamentals issuers coming the market the post- issuers in the market more benign crisis. An exception more negative American issues private issues deutsche mark-denominated issues market spreads in 1996 first half of 1997, a trend was viewed exuberance.” But primary spreads crease. But once this increase table 4.8). Overall improves. Again, as credit worthy issuers the market, this shift market sentiment reduced sensitivity changes in est rates, higher premium the poorer risks entering the Market Irrationally Exuberant second half at least first half some observers the market issuers entered estimates allow with predictions. first estimated the equation for Latin up to Changing Spreads the contribution the contribution markets regard credit worthiness of changes in impact on spreads blanket changes in of favor, changes in the independent independent (- @*)I, which can be thought of as changes in characteristics (as, for example, issues of risky borrowers be regarded distinguish changes average inverse Mills of as a measure sample selectivity) versus changes average value the other ables. Finally, distinguish changes in effect of inverse Mills be included) changes in effect of regressors. Intuitively, that the The decompositions for two after the Mexican crisis, during the spread com- in log declined following Mexican crisis; The Mexican (1991: Ql-1994: treasury rate Inverse Mills Irrational Exuberance? Inverse Mills ratio Barry Eichengreen Ashoka Mody The supply and demand vary noticeably that the coefficients for issues in yen significant negative signs for Thus, for East Asia, the deutsche the East Asian issuers time their error terms the spread probit equations. For Latin behaves much credit rating probability of associated with For East Asia, in An interpretation Asian bonds to the with higher spreads. to obtain investors, for reasons of East Asian are not A central question in the literature whether changes in changes in whether there been changes pricing behavior over time. credit rating, service ratios, the reductions spreads that occurred between be reason pricing behavior about the otherwise explicable shift in pricing in favor of developing country debt, then there a priori a sudden a corresponding curtailment Fundamentals and between two periods, Oxaca decomposition: right side of the contribution their economic the contribution What Explains Changing Spreads For East Asia, on the rate persists after correcting this finding, the probit the spreads equation together, for this leads to an For Latin America, higher interest rate reduces spreads (after correction selection bias). interest rates the demand their price therefore declines (equivalently, the spread East Asia, interest rates appears to supply curve the spread. that appear both the spreads equations, it is again useful whether they work in directions. In regressions using full sample bonds, a larger credit residual (a credit rating, probability of the spread, with inferior credit ratings find it Latin America, higher debt In Latin has a weak positive rescheduling debt are apt to the capital market same time raising the spread that successful issuers dummy variable for Latin America behaves similarly: Latin American borrowers issue more bonds but although the magnitude and the coefficient of the errors two equations-varies sample, it that unob- to the also reduce as unobserved coefficients work in offsetting influence mainly bonds. For countries that recently rescheduled tend therefore supply the corre- sponding outward supply reduces the spread. regional dummies suggest continent-specific characteristics, unusually high volume this works, drive down their issues the spreads invoke this for in a market with imper- possible for the demand bend back ments along the demand along a discussion in demand, not rationing. The of only terminological importance, but its significance a “backward rather than a “backward bending supply curve.”) Barry Eichengreen Ashoka Mody the Selectivity Correction Latin America Latin and East Asia America Private placement Debt service/ Private issuer Latin America Log likelihood (1 3.17) Numbers in parentheses are reduces observed this interpretation correct, then tive sign be significantly comparing the coefficients in tables that the treasury rate significant at idea that selection bias role in driving observed result. What Explains Changing Spreads, 1991-96, America Latin East Eastern America Asia treasury rate for debt Public issuer Latin America 14.1 18) (-2.1 11) parentheses are interest rate market spreads (see, e.g., results presented with this finding. Among the negative sign interest rates Since higher interest rates issuance (as in probit above), it that the the market some unobserved credit features them attractive Barry Eichengreen Ashoka Mody (henceforth referred as Latin America), East Asian, treasury rate high, the volume of issues declines This finding “search for yield” market-the idea interest rates fall there a greater for higher-yielding emerging bonds. This the strength relationship varies across decline in issues following treasury rates evident for East Asia, that the East Asian supply of interest rates rise their borrowing costs go up.’* The higher the credit rating residual, all regions except Asia, where effect is insignificant. Higher probability of issue especially in Latin America, For East Asia, higher debt service probability of the supply side, higher deficits, Eastern Europe, full model with once again reveal differences pricing across regions. For Latin indicates plausibly larger issues (consistent with and the greater liquidity larger issues the secondary market). Latin American private placements spreads equation positive sign, with the fact issued in markets with stringent disclosure requirements. Disclosure require- for private placements investors capable associated risk. for this reason are private placement because infor- issuer characteristics in that spread. Latin American issues typically having been relatively little evidence well-defined yield curve until The picture for East Asian issuers, who higher spreads. large negative coefficient in spreads equation dummy variable for Israel, reflecting its issues are guaranteed ernment. Supranational bonds (guaranteed entities beyond have lower Recent studies somewhat surprisingly, negative influence demand and consider the issues decisions together. the Probability America Latin East Eastern Asia America Asia for debt Public issuer Private issuer Latin America Public issuer Private issuer American Interactions Latin America Asia Latin America Asia Eastern Private placement rating residual growth rate Latin America Deutsche mark 319.10 384.25 152.18 265.82 174.77 165.96 195.95 257.47 6.71 5.81 6.79 6.50 6.78 6.49 6.51 6.75 37.12 44.46 40.42 37.10 28.97 62.15 63.14 40.67 33.34 2.05 4.87 -2.15 16.41 0.13 0.22 0.30 0.28 0.19 0.38 0.22 0.48 0.28 0.35 0.62 0.28 0.38 0.42 0.51 0.00 0.00 0.00 0.00 0.02 0.71 0.66 0.00 0.00 0.26 0.34 0.47 0.32 0.31 0.44 0.37 0.30 The spread its relationship spreads are observed defined by: determines the desire willingness of lend (which the error terms a standard the fitted probabilities the selection equation probit requires information on those address this problem For each for three types of issuers: sovereign, For each types did the market, zero. Table highlights the character- likely when interest rates better quality credit (with credit rating residuals when reserves public-sector demand for estimate equations using maximum maximum-likelihood estimates verge nicely. performing sensitivity tests using maximum-likelihood estimates converge; in this case perform the conventional two-step procedure, probit model ratio (with full-information are obtained using difference in derivative of the probability distribution function to a the independent average values facilitate interpretation separate equations for for the Latin American estimate a similar model Barry Eichengreen these ratings on a international bankers, numerical value ranging indicating zero probability default). For each observation the most recent credit rating the bond's date rather than include the the residual the credit rating on utilize only the orthogonalized component because the credit rating included separately as and Packer et al. preliminary look at the issuers in the Caribbean, Latin American Asian issues separately as linear relationship the form the dependent the logarithm the spread, a random bond characteristics amount, and global economic conditions the ten-year rate credit rating residual, the GDP, the ratio the previous growth rate; including the whether the borrower other public, the currency the bond conditions under this linear equation of issuer points in The estimated equation credit rating 12.15 (reserveslGNP) (debt reschedule) t-statistics are in parentheses Changing Spreads Market Debt? points for spreads; numbers for East Eastern Latin Middle Africa Asia Bonds with Bonds with Bonds with 166 94 205 226 Sovereign issues the other issues, which although the private issues bonds’ greater for all regions except confirm this regression analysis. variety of proxy for (at time of issue). proxy for credit from the Global Development Finance World Economic the ratio external debt national product (GNP), exports, a dummy variable for whether country con- restructuring agreement previous year, budget deficit as a Sovereign credit 9. This with the rating agencies, are reluctant private borrowers higher the grounds with transfer problems private domestic foreign exchange; Launch Spreads, (basis points) 1991 1992 1994 1995 1996 1997 197 151 140 94 173 145 125 149 178 152 332 266 Latin America 448 410 369 413 216 219 South Asia 190 233 356 288 289 Bonds with 52 118 163 170 302 165 332 202 203 342 369 1,697 blank indicates that no first calculating using the the time of issue then subtracting the “riskless” sovereign security the currency source provides the period issued grew before falling Mexican crisis market conditions in 1995). Bond issuance rose sharply in a banner over 1,300 bonds. publicly available for ancillary information country conditions on data for which 1,033 usable use 277 The main Latin a distant second with bonds (Korea a leading Eastern Europe in this especially active year for the region. America has also highest spreads. years from first time In con- trast, East Asian spreads always less they rose their highest typically higher for Latin America. Eastern consistently been between those East Asia. summarizes sample characteristics distinguished: sovereign, private borrowers, public entities in bonds. Sovereign Eb86 ib86 eb16 1b16 €696 EbE6 5 IbE6 B 0 8 E P 0 088O 0 0 52 W Ti- cu 8 0 2zc!z 0 0 0 0 EbZ6 3 IbZ6 8 EbI6 $ E Ib16 ' 6 a c) C 8 v ._ 7 w E Barry Eichengreen proportionately, for example, because high-risk borrowers discouraged from coming the market. the determinants impact of factors on the the market therefore be of selectivity next sections methodology designed these problems. Sample Characteristics The bonds income securities with face value, emerging market borrowers, placed on international markets and are dollars). Although consists of market), over period a growing number tries floated market for foreign issuers private placement market (under provisions Rule 144a); there also some growth The bonds are who commit with investors. trade on secondary market Our data “launch” spreads for bonds countries, obtained IMF. That launch spreads launch spreads move differently over time secondary markets. While there some tendency for primary spreads secondary spreads four quarters, the series frequently In poor primary spreads indeed, some- times they fall. This tendency for the most credit worthy borrowers the market. investing in emerging secondary market spreads, may the opposite launch spreads riskier borrowers leaving only low-risk, low-spread borrowers This makes it different classes of borrowers. Numbers, Spreads, international bond issues with convertible rise slightly exclude these from the grounds and relationship warrant a separate What Explains Changing Spreads variables for private issues in Brady plan debt reduction schemes). Because his sample his estimates economic variables, only growth differ significantly from zero at 1997) uses data and somewhat longer list of otherwise subject European bor- emerging markets, raising questions tional bonds last time bond markets for international lending. But their sample includes industrial countries the informa- the market has time limits this study for present purposes. bond spreads in 1976-80, since there in developing-country for only thirteen countries should expect pricing same over time.6 Edwards analyzes: all emerging market debt the end than two-thirds mid-1990s (Bernstein van Kleist spreads on and Standard a summary measure industrial country interest rates (as a Latin American than on comparable Asian result suggestive segmentation. Surprisingly they find that the on industrial country interest rates insignificantly different country participation the bond market has time suggests least squares ity bias. same changes in economic traders often that a rates raises emphasized above accentuating the trade-off between acquisition emphasized Barry Eichengreen Section 4.5 basic results for both the issue decision pricing behavior. Section 4.6 the contributions pricing behavior over time. Shifts in pricing behavior also examined estimated sample and out sample. Section analysis for market debt. explosive growth market debt, there have been systematic studies the secondary market for bank which there significant number studies (reflecting the dominance But there of risk therefore pricing behavior differ between bonds. Models that banks may have a comparative advantage in assembling their clients into the secondary market their claims. Pecking-order finance suggest claims with different degrees levels of and that prices should bear a different relation- Bonds typically have senior status, more variable. Thus, the studies of in bank to the bonded debt. the market for developing-country bonds are Cantor and spreads on for forty-nine tries in income, gross domestic growth, inflation, external balance, ternal debt, economic development to the of Moody’s this study include it considers developed as developing countries, it analyzes only and not also private that none the macroeconomic statistically significant credit ratings included. Cline rate as as government borrowers. he studies only debted countries, whose representativeness siders only four economic determinants rate spreads: per capita with dummy Huizinga (1989) Secondary Spreads Interest Rate Barry Eichengreen regard the post-1995 surge portfolio capital its equally sudden halt (along with toward the end of 1997. If increased lending spread compression half of reflected improved the borrowing countries, then there long as governments are rewarded for putting national on a firmer footing, there are the trend toward for emerging market mainly the credit conditions in the money centers, then there effects of the eventual tightening financial market con- in the advanced industrial countries. more troubling the increased capital reduced spreads period centered case there sudden shift back corresponding curtailment on about developing-country bonds the years period that spans the recent episode bonded debt. In contrast previous studies, sample selection. sue decision debtors and the pricing decision minimize selectivity treating the Overall, the results confirm higher credit quality leads spreads. This the presumption market discriminates same explanatory variables the principal debt- America, East Asia, Eastern Europe). And shifts in market sentiment rather than break point fourth quarter 1994, when Mexican crisis lower levels with 1995:Ql Mexican crisis changes in spreads adverse shifts market sentiment more fundamentals. The same the subsequent compression, favorable shifts market sentiment. the literature. Section tional Monetary allowed them Mexican crisis scot-free. is likely again come have little effort needed among bonds that the process. Investors, in incomplete knowledge of countries’ economic a practice conducive powerful incentives be informed evidence they exist between with different advanced their decline in emerging market spreads that took place between early 1997. secondary-market spread between developing Advocates of functioning markets” rededicated themselves Mexican cri- to put their fiscal, monetary, affairs in order. Better policies, implied reductions market debt tals improved sufficiently decline in investors in their exuberance may have been relationships between that the costly information for portfolio diversification, the information individual investor about a particular country. (This depends on the assumption that the cost of acquiring information about a country size of disproportionately affect funds across countries prices of particular bonds portfolio diversification facilitates these circumstances, investors might fail on a particular bond a gradual economic conditions, could lead one country could lead investors their expectations the prospects with superficially similar characteristics. money” results from “frictions” in information leading herd-like behavior and out countries on the are not obviously attractive definitely “no-no’s.’’ their model with diversify limits incentives for information acquisition. Kehoe argue their model predicts types of countries that predict only that hot flows will occur. Herd behavior observed for foreign direct investment Spreads continued the third which they One oft-heard justification for this Mexican rescue for investors concern themselves with a potential credit worthiness, at least left them with this package engineered and the Interna- Syndicated Loans Markets (gross, 1991 1992 1993 1994 1995 1996 1997 Portfolio equity Western Hemisphere Europe and Central Middle East Syndicated loan 7.2 11.9 11.2 16.4 13.9 24.3 62.7 57.6 101.9 7.1 12.9 28.8 18.0 23.1 47.1 25.3 43.1 1.9 1.6 42.5 43.0 74.9 79.7 11.9 14.3 82.4 129.5 What Explains on Emerging Ashoka Mody value of the course a major for developing significant implications for the international capital markets. The the bonds billion in billion in 1992, billion per year in billion in of much market’s ability Some observers Barry Eichengreen Pardee Professor Political Science University of California, research associate Bureau of Economic Research, the Centre Ashoka Mody visiting professor and man- agement at School, University this paper Specialist for Finance with Development Prospects authors are Dunaway, Anne Jensen, Ken Wood of who helped assemble Ananda Chanda and for sterling research assistance; early informal discussion of results at the International discussant. Robert Hill, David Shapiro also provided useful comments. are those authors and preliminary estimate for three quarters anticipate a the turmoil Asian financial markets.