Information about Western Hemisphere Hemisferio Occidental
Chapter

Mexican Debt Exchange: Lessons and Issues

Editor(s):
Peter Wickham, Jacob Frenkel, and Michael Dooley
Published Date:
March 1989
Share
  • ShareShare
Information about Western Hemisphere Hemisferio Occidental
Show Summary Details
Author(s)
DAVID FOLKERTS-LANDAU and CARLOS ALFREDO RODRIGUEZ* 

Mexico undertook its first debt-exchange programan exchange of discounted restructured loan debt for new bonds with improved debt-service characteristicsin early 1988. The rate at which old restructured debt was exchanged for the new bonds was determined in an auction; this note analyzes the outcome of the auction and the valuation of the new bonds with a simple asset pricing model and data from the secondary market for restructured loans.

DEBTOR COUNTRIES that experience unexpected changes in current or anticipated cash flows1 may find it beneficial to modify the stream of debt service payments mandated under existing contractual debt obligations by exchanging outstanding debt for new debt with different debt service characteristics. Such a voluntary debt exchange, frequently employed in U.S. capital markets, offers the debtor the opportunity to match the contractual stream of debt service payments more closely with current and anticipated cash flows. In addition, the debtor can affect a buy-back of outstanding debt at market prices, by issuing credit-enhanced, that is, collateralized, debt in return for outstanding debt.

These considerations led Mexico to undertake its first debt-exchange program in early 1988. Under this program, Mexico was prepared to issue up to $10 billion of a new negotiable bond debt in exchange for rescheduled bank debt. The bond obligations were to have a higher coupon than recently rescheduled bank debt, and the bond principal was to be fully collateralized. The rate at which old restructured debt was exchanged for the new bond obligations was determined in an auction in which banks tendered restructured loans for new bonds at specified ratios and amounts. In the event, Mexico issued $2.6 billion of the new bonds in exchange for $3.7 billion of restructured bank debt.

This note analyzes the outcome of the auction with a simple asset pricing model which relies on data from the secondary market for restructured bank loans to obtain a valuation of the new bond obligations. By comparing the predicted prices for the new bond obligations with actual prices prevailing in the secondary market for the new bonds since the auction, it is possible to obtain an indication of the validity of the model and of the efficiency of the secondary market for bank debt of indebted developing countries. We then compare the debt exchange to cash buy-backs and raise some general issues associated with the securitization of external debt.

I. Debt Exchange Program

Mexico was prepared to issue up to $10 billion of 20-year marketable international bonds to its bank creditors at a spread of 1

percent above the 6-month London Interbank Offered Rate (LIBOR—double the spread on rescheduled Mexican bank debt), payable semiannually. The new bond will be amortized in a single payment at the end of 20 years (corresponding to the maturity of eligible rescheduled bank debt), but will have call provisions permitting earlier amortization at the discretion of the Mexican authorities. The new bond obligations are listed on the Luxembourg Stock Exchange and are traded and cleared through the Eurobond market system. They are fully marketable on issuance subject to normal Euromarket practices. Any tranche of the new bond obligations issued within the United States will be on a private placement basis and may be traded in the private placement market in the United States, but the issue will not be registered with the Securities and Exchange Commission (SEC) under the U.S. Securities Act and thus cannot be publicly offered or traded in the United States.

In addition to a higher interest spread, the principal of the new bond is collateralized with a nonmarketable foreign series zero-coupon U.S. Treasury security maturing in 20 years. The zero-coupon collaterial is held in escrow by the Federal Reserve Bank of New York in the form of a book entry. The actual face value of the zero-coupon U.S. Treasury security was determined once the volume and price of bank claims to be exchanged became known, while the amount of Mexican foreign exchange reserves required to purchase the zero coupon bonds depended on the prevailing market interest rates.

Bank debt eligible to be included in the program consisted of claims under the 1985 debt restructuring agreement and the 1983 and 1984 credit agreements, but the 1986/87 new money package, short-term credits, and private sector debt are excluded. The total eligible volume was estimated at about $53 billion.

Before the exchange could take place, the sharing procedures relating to prepayment of loans, the pari passu clauses relating to the priority ranking of payments, and the negative pledge clauses under existing agreements all had to be waived. Since the restructuring agreements covering the eligible debt permitted such waivers to be based on the simple majority of the eligible bank exposure to Mexico, only 50 banks, compared with about 500 banks that originally signed these documents, were needed to amend some 92 loan agreements. Negative pledge clauses contained in the agreement with the World Bank and the Inter-American Development Bank also were waived. To avoid negative pledge problems with other holders of existing Mexican bonds, these international bonds would have had to have their creditworthiness similarly enhanced.2 The new bonds carry a pari passu clause that requires them to be treated on similar terms as all other external debt including obligations to international organizations, such as the Fund.

II. Pricing New Bond Obligations

The basic assumption of the valuation model is that the banks holding Mexican obligations are risk neutral. This assumption implies that the market value of Mexican debt will equal the expected present value of principal and interest payments. We assume there exist two states of the world.3 In the first state, the debt service payments are made in full; in the second state no debt service payments are made. If the probabilities that interest and principal are paid in full are denoted by π1, and π2, respectively, then the market value of Mexican debt of N period maturity, face value of FV, and an interest spread of s percent above ri, the LIBOR expected to prevail in period i, will be given by:

where Ri=j=1j(1+rj).

Since LIBOR is the banks’ cost of funds, it is used as a discount rate for the expected debt service payments. The bonds that are being priced are floating-rate obligations, that is, the coupon is reset at regular intervals by adding a spread to the then prevailing LIBOR. The forward LIBOR rates ri, used for discounting, are obtained by adding a constant spread to the forward interest rates extracted from the zero-coupon U.S. Treasury obligations, with the spread equal to the anticipated difference between the U.S. Treasury obligations and LIBOR.4 Since payments can be designated arbitrarily as payments for interest or principal, by the debtor as well as creditor, when such payments fall short of full contractual requirement payments, we can assume that π1 = π2 = π.

We can solve for π(= π1 = π2) in equation (1)

where Ri=j=1i(1+rj).

Since principal payments of the new bonds are guaranteed, we set π2 = 1, while π2 = π. Thus, given the spread above LIBOR on the new bond as 2s, the market value of the new bonds will equal:

where Ri=j=1i(1+rj).

An alternative, but equivalent, method of pricing the new bond is to assume that the discount rate applied to the uncertain interest and principal payments is equal to the internal rate of return (IRR) that equates the contractual debt service payments with the observed market value, that is,

Since the principal payments of the new bonds are certain, the bonds will be valued in this model at:

where ρ(IRR) is the discount rate for interest and principal payments.

The valuation model developed above relies on four assumptions. First, Mexico’s ability to service external debt has increased by the amount necessary to pay for the collateralization of principal. In the case at hand, the increase in Mexico’s foreign exchange reserves prior to the debt exchange is thought to justify this assumption. If instead Mexico’s ability to service external debt had remained unchanged, the increase in expected payment of principal would merely have reduced the expected payment of interest, and collateralization would be ineffective.

Second, investors’ assessment of the probability of debt service payments being made in full has not changed as a result of the assumed increase in its ability to make such payments.5 This assumption is clearly valid only as long as the share of debt service payments that is collateralized does not become large relative to total debt service payments.

Third, investors perceive the likelihood of interest payments being made on the new secured debt to be the same as the likelihood of interest payments being made on outstanding rescheduled bank loans (that is, the new bond obligations are not perceived to be senior claims).

Fourth, arbitrage will ensure that yields on unsecured restructured bank loans and unsecured bonds of equal structure (coupon, maturity) will be equalized in the secondary markets (in particular, once the stock of secured bonds outstanding reaches a significant proportion of the stock of syndicated loans outstanding).6

III. Auction Outcome

In January 1988, 20-year floating-rate Mexican restructured debt had a coupon of about 8.75 percent (LIBOR was 7.94 percent), a spread of

percent, and traded at a discount of about 50 percent (yield-to-maturity of 18.2 percent). A 50 percent discount suggested that the expected present value of total debt service payments on a bank loan of $100 face value was $50 (equal to its market value). The prevailing LIBOR of 7.94 percent implied that the present value (PV) of full principal repayment and full interest payment was $21.70 and $86.31, respectively.7 The credit market’s view about the probability of full payment being made on the outstanding bank loans is equal to the ratio of the market value of the old restructured bank loans and can be obtained from equation (2), that is, as the ratio of the market value of the restructured loans to the present value of its total contractual payments (0.463). The expected PV of principal and interest payments were then $10.05 and $39.95, respectively, totaling to a market value of $50.

The new bond was to have a coupon of LIBOR + 1.625 percent (about 9.56 percent in February 1988), and principal repayment was secured with a zero-coupon bond of equal face value. The present value of principal repayment was $21.70, and the present value of interest payments was $94.32 on $100 face value of the new bond. Collateralization of principal implied that the expected present value of principal payment was equal to the PV of full principal payment, $21.70. Hence, the expected value of the interest payments on the new bonds was $43.65, and the predicted market value of the new bond, that is, the sum of the expected value of principal and interest payments, was $65.35; or equivalently, the predicted market discount on the new bond was 34.65 percent. The relatively higher market value and yield of the new bond obligations compared with restructured bank loans are the result of the higher coupon and the collateralization of principal repayment.

The amount of outstanding bank debt that could be expected to be offered in exchange for a given amount of new secured debt can be calculated from the predicted market discounts on these two types of debt. In particular, the market discount of 50 percent on restructured bank loans and the predicted discount of 34.65 percent on the new collateralized bond suggested that an average of $1.31 of face value of the old syndicated loans would be offered for $1 of face value of the new secured debt. If the entire proposed $10 billion of new bond obligations were issued, the reduction in total indebtedness would have been $3.1 billion. The higher the market discount on old bank loans, the larger the amount of syndicated loans that was offered in exchange for a given amount of secured debt and greater the reduction in total indebtedness for various assumptions about the magnitude of the market discount of syndicated loans.

The reduction in market discounts and in total indebtedness made possible with the collateralization of principal was limited by the fact that at the prevailing nominal U.S. dollar interest rate levels, the PV of interest payments on debt of 20-year maturity was about 80 percent of the PV of all payments, while the PV of principal was only about 20 percent of all payments.

The Mexican auction was completed on February 26, 1988, with 139 banks making bids covering $6.7 billion in restructured debt. Mexico accepted bids from 95 banks for $3.7 billion in claims. Among successful tenders, Japanese banks offered the largest amount followed by U.S. and Canadian banks. Only two or three money-center banks participated out of about 30 U.S. banks that tended bids. These claims were exchanged for $2.6 billion in new bonds, at an average exchange rate of $1.42 of restructured debt for $1 of new bond obligations. Mexico’s debt will be reduced immediately by $1.1 billion and contractual interest savings amount to about $65 million a year.

Mexico ranked accepted bids in descending order of bid ratio (amount of eligible debt tended in exchange for new bonds) to maximize the amount of eligible debt retired, and the accepted bid was at the bank’s bid ratio. The Mexican Government thus determined the amount and price at which such offers were accepted. The dispersion of bids was such that the highest bid was $2.08 and the lowest was $1.12 with about three quarters of bids concentrated in the range of $1.33 to $1.67. Within this band, most exchange offers lay between $1.44 and $1.54 of old bank debt per dollar of face value of the new bank debt. This dispersion in bid prices was the result of differing beliefs about the seniority of the new bonds and about the ability of the secondary market for restructured bank debt to absorb large amounts of debt.

The auction procedure adopted by Mexico thus made it possible for Mexico to accept bids in descending order of price down to its reservation price of $1.33 of old bank loans per dollar of new bonds and to achieve an average exchange rate of $1.42. According to the model developed above, Mexico’s reservation or limit price of $1.33 of old bank debt per dollar of new bank loans corresponded closely to the discount of 50 percent then prevailing in the secondary market for restructured bank debt.8 This type of auction thus made it possible for Mexico to accept only bids above the current market price and to exchange a modest amount of new collateralized bond debt at “better-than-market” terms for restructured bank debt.

IV. Market Valuation of New Bonds

Trading in the new bonds began shortly after the auction results were announced. Initially in April, the bid-ask spread for the price of the new bond with a face value of $100 ranged from $64 to $70. By July-August the bid-ask spread had narrowed, and transactions fell in the range of $62–68. The price of the new bond predicted by the model in March was about $65, while a higher LIBOR and lower price for restructured debt resulted in a lower predicted price of about $62 during the June-August 1988 period. Thus, the predicted value of the new bonds falls within the actual bid-ask range for the new bond (see Table 1).

Table 1.Valuation of Restructured Bank Loans and Collateralized Bond Debt
Market Price ofMarket Price ofPredicted Price
Restructured BankNew Bond Debtof New Bond Debt
Loans (In dollars per(In dollars per(In dollars perLIBORLoanBond
$100 of face value)$100 of face value)$100 of face value)(In percent)CouponCoupon
March 198849–5164–70657.98.79.6
June 198848–5063–6863.88.59.310.1
August 198847–4862–666199.810.6
Source: International Financing Review, various issues.
Source: International Financing Review, various issues.

The results in Table 1 appear to indicate that the simplified valuation model with risk-neutral investors and unchanging probabilities of debt service payments predicts the market value of the new bond reasonably well. The results also indicate that the secondary market for restructured bank loans and the market for the negotiable new bond obligations are being well arbitraged according to the valuation model employed above. Thus, the fact that the new bond is fully negotiable and not encumbered by the same kind of regulatory and fiscal constraints as are bank loans has not influenced the pricing very much. In other words, there appear to be no special features present in the market for bank loans that should lead us to doubt the usefulness of the prices observed in this market as indicators of country risk.

V. Comparison of Debt Exchanges With Cash Buy-Backs

As was shown above, a market discount of 50 percent on restructured loans (LIBOR of 7.9 percent) would on average result in $1.30 of restructured debt being exchanged for $1.00 of new bond obligations. If Mexico had bought about $2 billion of U.S. Treasury zero-coupon bonds as collateral for $10 billion of new bonds, then these new bonds could be exchanged for $13 billion of restructured loans. Since the new bonds would trade at a discount of about 35 percent, Mexico can buy back the new bonds with $6.5 billion of reserves. Thus Mexico would have retired $13 billion of restructured bank loans with a cost outlay of $8.5 billion, of which $2 billion are invested in the zero-coupon bonds. After selling the zero-coupon bond Mexico’s net expenditure is $6.5 billion, that is, the same result as could have been achieved through a direct cash buy-back of restructured bank debt. Hence debt exchanges and cash buy-backs are equivalent in their economic outcome.

In actual fact, the collateralization of principal cost Mexico about $500 million in foreign exchange reserves, and Mexico recovered about $1.1 billion in par value of rescheduled bank debt. In addition, Mexico does not have to fulfill the obligation to pay $3.7 billion of old bank loans in 20 years since this payment is covered by the maturing U.S. Treasury zero-coupon bond. If the discount on old Mexican debt is 50 percent, then the market’s expected present value of Mexico’s promise to pay $3.7 billion of bank loan principal in 20 years is about $300 million. Thus, the net present cost to Mexico of reducing its nominal debt by $1.1 billion was about $200 million. A direct cash buy-back of $1.1 billion of old bank debt would have required $550 million of foreign reserves at the currently prevailing discount on bank loans of 50 percent. Thus, by accepting only bids offering more than $1.33 of old bank debt for $1 of new bond obligations, Mexico has been able to keep the average exchange ratio above the exchange ratio implied by the market discount of 50 percent of old bank debt, and hence was able to outperform a direct cash buy-back. If, however, Mexico had accepted a volume of bids sufficiently large that the average exchange ratio of old bank debt for new bonds coincided with the ratio implied by the prevailing market discount, the debt exchange would have been equivalent to a direct cash buy-back.

VI. Systemic Implications of the Mexican Debt Exchange

The systemic implications of the Mexican proposal derive almost entirely from the creation of a liquid secondary market in the new bonds with price quotations published daily in the financial press. For the first time since the 1920s, there exists a market for external obligations of a major developing country, which can provide continuous data on the market’s assessment of Mexico’s willingness to repay such obligations. Currently, Mexico’s outstanding syndicated loans are being exchanged among banks (and some nonbank investors) attempting to modify their exposure, and the prices at which such exchanges take place are frequently thought to be distorted by the peculiarities of the participants. for example, tax considerations, absence of a significant number of nonbank participants, new money obligations, high transactions costs, and thin markets. Thus, while these prices undoubtedly reflect the banks’ assessment of Mexico’s creditworthiness, they have not attained the credibility and information content that are normally associated with Eurobond prices of major issuers.

Credible daily price quotations in the external obligations of a major debtor have three implications for the overall debtor-creditor relationship. First, the pricing of new money, in whatever form, will have to respect the price information available from secondary markets. It would, for example, expose the directors of U.S. banks to liability arising from shareholder suits if a bank participates in new money packages at interest rates significantly below those observable in the market for the new bonds (adjusted for the collateral).9 If the new Mexican bond continues to trade at yields of about 18 percent, any new financing for Mexico will have to be at a cost of about 18 percent. Similarly, new money for other developing country debtors would have to be priced in relation to the yield on the Mexican bonds. The market pricing of the credit risk associated with developing country borrowers may in the long run lead to a more efficient allocation of credit.

Second, it is likely that the market prices of Mexico’s external obligations will decline further from the level currently observed in the market for restructured loans. The yields currently obtainable in the U.S. high-yield corporate bond markets are generally in excess of 15 percent. The nonbank investors in the new Mexican bonds can be expected to ask for a yield in excess of U.S. corporate high-yield bonds because of the difficulty in claiming any residual assets. Therefore, to be competitive the new bond may well need to have a yield in excess of 20 percent in the absent of major policy adjustments. One reason why market prices of the external obligations of developing countries remained above those of high-yield corporate bonds is that the concentration of these obligations in the major banks gave these creditors greater influence over debtors than dispersed bondholders would have. However, the recent indication that bank holdings of developing country debt may be sold (e.g., Citicorp reserving) has already led to significant declines in market prices. For example, the price of Mexico’s obligations fell from about 65 in April 1987 to 50 in January 1988.

Third, the implications for systemic stability of the banking system are largely positive. U.S. banks have nearly doubled their primary capital since 1982 and, with some exceptions, are able to cope with the implications of lower market prices of the external obligations of developing countries. The share prices of U.S. banks are thought to reflect the market value of banks’ assets, rather than the book value of such assets. In any event, U.S. banks are being given an option to carry the new bonds at face value or to treat the bonds as a security and mark its value to market regularly. In the longer run, the ability to dispose of these assets, even at a substantial discount, may well serve to help the banking system to reduce its own cost of funds and thus improve its competitive position vis-à-vis the securities markets by reducing the amount of disintermediation that has taken place in recent years.

Thus the three systemic implications of the Mexican debt proposal are: (1) the cost of new financing for developing countries will be based on price information from bond markets; (2) further decline will occur in currently observable prices of syndicated loans; and (3) systemic stability in banking markets will improve. The Mexican proposal may therefore be regarded as the first step toward the system of development finance that existed before the 1930s, that is, development finance being provided through a broad-based bond market at risk-related yields, while collateralized trade finance is being provided by banks.

VII. Conclusion

According to the asset-pricing model employed in this paper, about $1.30 of restructured Mexican bank debt exchanges for $1.00 of the new Mexican bond. By accepting only bids of more than $1.30 of restructured bank loans for each $1.00 of the new bond, Mexico was able to exchange $2.6 billion of the new bond for $3.7 of restructured bank debt, that is, an average exchange rate of $1.40.

If Mexico had accepted a sufficient volume of bids so that the average exchange rate of restructured debt for the new bond had coincided with the equilibrium rate of $1.30, then the debt exchange would have been equivalent to a cash buy-back. That is, if the amount of foreign exchange needed to collateralize the new bond would have been used to buy back restructured debt at market prices, the reduction in external indebtedness would have been equivalent to the reduction achieved under the debt exchange.

The post-auction market valuation of the new Mexican bond obligation appears to be broadly in line with the observed prices for restructured bank debt according to the asset-pricing model with risk-neutral investors. Since the new bonds are fully negotiable Eurobonds, we conclude that prices observed in the secondary market for bank loans are not greatly distorted by special features of bank loans arising from their non-negotiability and regulatory or fiscal considerations.

*Mr. Folkerts-Landau is the Assistant Division Chief of the Financial Studies Division of the Research Department. He is a graduate of Harvard University and holds a doctorate from Princeton University. He was an Assistant Professor of Economics and Finance in the Graduate School of Business, University of Chicago, before joining the International Monetary Fund.Mr. Rodriguez is a staff member of the Argentine Center for the Study of Macroeconomics (CEMA) in Buenos Aires.
1Such cash-flow variations may arise with changes in the country’s trade balance or external financial flows.
2The credit enhancement of Mexico’s outstanding international bonds is estimated to have cost about $100 million.
3Since bank creditors are assumed to be risk neutral, the specifications of the probability model for debt service payments affect the valuation only through expected values. Hence, we can simplify the exposition without loss of generality by assuming that only two states can occur.
4This is the so-called TED (U.S. Treasury Eurodollar) spread.
5For an analysis of how debt conversions might affect the probability of debt service payments see Michael P. Dooley, “Buy-Backs, Debt-Equity Swaps, Asset Exchanges, and Market Prices of External Debt,” above.
6Negotiability of the bond debt instrument by itself would tend to reduce its yield because of larger institutional demand and increased liquidity, but negotiability may also reduce the ability of the creditors to enforce the bond contract, and this would tend to raise yields. In the case at hand, the enforceability of negotiable bond contracts may well be significantly worse than that of the existing syndicated loan contract. The bank syndicates, with the backing of financial authorities concerned about systemic financial problems arising from the erosion of bank capital, generally have a greater ability to increase the likelihood of debt service than bondholder councils. We assume that the combined effects on yield of negotiability and enforceability will offset each other. Finally, the possibilities of exiting from credible new money obligations by exchanging syndicated bank debt for negotiable collateralized bonds might act to increase the attractiveness of the new bonds to some banks, thus permitting a lower yield to be offered initially.
7The present value is computed using the currently prevailing LIBOR as discount rate, instead of using the rates implied in the term-structure as indicated in equation (1).
8The reservation price predicted was $1.31 of old bank debt per $1.00 of new bank debt.
9In particular, concern about shareholder suits forced Citicorp to postpone establishing reserves against its developing-country obligations until there was a clear deterioration of conditions (which was provided by Brazil’s moratorium on interest payments).

    Other Resources Citing This Publication