Comments on de Fontenay and Jorion

D. Folkerts-Landau, and Marcel Cassard
Published Date:
July 2000
  • ShareShare
Show Summary Details
Gregory Makoff

“Foreign Currency Liabilities in Debt Management” addresses two main issues. First, how much foreign currency debt a country should have (if any). And second, if a country has foreign currency debt, what its composition should be. The first part discusses the selection of the currency composition of foreign currency debt from the point of view of countries that have foreign currency debt by necessity. The second part jointly addresses the issues of how much foreign currency debt a country should have and what its composition should be from the point of view of developed countries that have sufficient local markets to fund government liabilities, but that have the option to borrow in foreign currencies. The third part of the paper applies modern portfolio theory to discuss the selection of an optimal amount and denomination of foreign currency debt for a developed country, Australia.

Comments on First Part

The first part of the paper takes a largely macroeconomic approach. The authors suggest that the currency composition of debt can hedge the volatility of a country’s terms of trade or debt-service ratio.

This approach importantly offers a clear procedure for relating balance of payments and financial market data to the selection of the foreign currency composition of a country’s debt: “The currency composition of the external liabilities depends on the covariances among exchange rates and the covariances between exchange rates and net foreign exchange receipts.”

However, as emphasized in the paper, “the practical limitations of this approach result from the fact that the relationship between exchange rates (or borrowing costs) and terms of trade changes are for many countries weak or unstable.”

The authors’ concern can be rephrased as the following question: Can a borrower perform a regression analysis of historical trade flows and borrowing costs and mathematically define an effective terms of trade hedge?

The technical difficulties are:

  • Data may not provide statistically significant measurements of key parameters.

  • Data may indicate parameters that are unstable—which change dramatically period by period.

  • Finally, world economic relations are continuously changing, so historical terms of trade data may not accurately indicate future terms of trade relations.

Some observers also question whether a terms of trade hedge is an appropriate objective for a sovereign. These observers point out that a government should focus on hedging its own cash flows and allow private sector entities to take or hedge foreign exchange risks according to their own objectives. Some observers additionally question whether a sovereign could effectively implement a terms of trade hedge. To the extent the private sector of a country actively hedges exchange rate risks in transactions that are not publicly disclosed, the government may create risk by trying to hedge trade-related currency exposures, which have already been hedged.

I would draw as a general lesson of this section that, regardless of the actual use of macroeconomic figures in the determination of a liability portfolio structure, it is essential that a country’s portfolio analysts have a clear understanding of balance of payments issues and how they directly or indirectly relate to government finances.

Comments on Second and Third Parts

The authors point out four rationales that countries may use to justify foreign currency borrowings:

  • Access to a wider range of debt instruments and maturities.

  • Macroeconomic hedging or budgetary hedging.

  • Providing a policy signal that authorities will not attempt to reduce the value of domestic debt through inflation.

  • Reducing borrowing costs and risks.

The authors suggest that the last point is the most relevant and important consideration for most developed countries. The authors further suggest that the mean-variance approach is an important analytical method by which borrowers can simultaneously calculate the amount of foreign debt a borrower should have and its currency composition.

The mean-variance approach suggests a portfolio structuring process that “uses as inputs the cost of borrowing in various financial markets and the covariances between the various interest rates, exchange rates, and the whole debt portfolio.” The analysis then seeks to use these inputs to calculate an “efficient set” or “efficient frontier” representing the locus of portfolios that minimize risk for a given cost level. As the authors show, a specific objective function including both cost and risk terms can be defined, which accounts for the relative risk aversion of the borrower.

The authors show that such analysis can be performed using either historical data or the liability manager’s interest rate and foreign exchange rate forecasts. It is emphasized that constraints can be added and that it can be generalized to include macroeconomic hedging. As this method is well known, I limit my remaining comments to two major concerns I have about the appropriate application of this technology.

First Concern

In most cases, I doubt the value of mean-variance analysis to sovereign portfolio managers in deciding the amount of external as opposed to domestic currency debt. My concern relates to the fact that domestic currency debt is distinguished from borrowings in external currencies by a government’s ability to print its own currency and a government’s authority to tax in its own currency. This qualitative difference between domestic currency liabilities and foreign currency liabilities challenges the validity of a portfolio optimization method that treats them on an equal footing.

From a risk management perspective, I emphasize that the decision to have external debt (and the amount of external debt) is a fundamental policy decision that should be determined by a thorough analysis of worst-case outcomes and risk tolerances, as well as the possible impact of international borrowings on the value of the country’s currency and domestic interest rates.

Second Concern

I have a more limited concern that there are a reasonable number of situations in which a borrower will want to avoid the mean-variance approach as a method of determining the currency composition of external debt. As an example, consider the selection of the currency composition of a portfolio consisting of deutsche marks, French francs, Dutch guilders, and Belgian francs, and possibly Spanish pesetas and Portuguese escudos. The process of EMU makes it likely that these currencies will be unified in the next few years as the euro. I would be concerned about the following issue: A risk analysis or portfolio optimization using the risk/return parameters of the past five years is likely to be inappropriate for estimating the risks and relative costs going forward. One could try to handle this situation within the confines of the suggested mathematical model by adjusting correlation coefficients and volatilities using projected values, but as there are distinct scenarios for EMU from success to meltdown, it might end up that this technology is more cumbersome than useful in defining key portfolio parameters.

I offer these criticisms of the application of portfolio theory to debt portfolio structuring without meaning to suggest that the principle of portfolio diversification be ignored. A portfolio should in all cases have a well-diversified set of market risks. A sovereign should avoid making a big bet on the appreciation or depreciation of a specific currency, or on the rise or fall of rates in a specific market. The results of portfolio theory are sensitive to inputs, typically historical or projected risk/return parameters—which leads to the important result that a large speculative position could be justified by applying apparently view-neutral calculations. Conclusions based on portfolio theory should be tested under a wide range of assumptions, and significant portfolio constraints should be imposed both to guarantee a sufficient level of diversification and to enforce consistency with other portfolio objectives.

See Claessens (1988) and (1992); and Kroner and Claessens (1989) and (1991). So-called “natural hedges,” such as royalty receipts in a specific currency, can easily be integrated into the model.

This section is based in part on de Fontenay, Milesi-Ferretti, and Pill (1995).

If the local currency depreciates versus the dollar to the same extent as the other nondollar currencies, the value (in domestic currency) of debt-service payments in those currencies remains unchanged, while it increases for the dollar-denominated debt. The increase would accentuate the worsening of the current account, resulting from the deterioration in the terms of trade.

See, for example, Kritzman (1993).

A number of authors have emphasized the benefits of International diversification. See, for example, Odier and Solnik (1993).

If the instrument is a long-term currency swap that exactly matches the cash flows of the foreign currency bond, the bond effectively becomes a domestic currency bond, and there is no residual exposure to foreign interest rates. Such currency swaps do not represent views on foreign interest rates or currencies, but rather take advantage of market conditions to lower capital costs. In the case of short-term swaps, which are effectively forward contracts, we know from covered interest rate parity that differences in capital costs should be small.

This number was selected from the typical risk aversion of equity investors. U.S. stocks have returned about 8 percent a year with a risk of 16 percent. This implies a risk aversion of about 0.08/(0.16*0.16)=3.

Because this is an ex post optimization, the efficient set must dominate any single investment. Therefore, there may he a bias resulting from optimization, which has been noted by Michaud (1989).

Similar results are obtained when considering only the more recent 1986—95 period.

See, for instance, Froot and Thaler (1990). There is abundant academic literature on the topic that has attempted, so far without success, to explain the anomaly in terms of a rational risk premium.

    Other Resources Citing This Publication