Foreign Currency Bonds

Generally speaking, across a range of currencies, foreign bonds as an asset class have very attractive statistical properties. Foreign bonds’ low correlation of returns with domestic equities and bonds is of particular interest. Research has shown that the correlation of returns between domestic and foreign equity markets tends to vary over time, increasing when they are declining, and decreasing when they are rising. This has caused a number of authors to conclude that the "effective" amount of diversification benefits one receives from investing in foreign equities is lower than it first appears. A key question is whether or not this is also the case with bonds. A recent research paper ("Asymmetric Dynamics in the Correlations of Global Equity and Bond Returns" by Cappiello, Engle, and Sheppart) shows that bonds behave very differently from equities in this regard. They found that the linkages across bond markets were much weaker than the linkages across equity markets, and the lowest correlations in their study were between equity returns in one region (e.g., Asia, North America, and Europe) and bond returns in another. Finally, they noted how the "flight to quality" phenomenon tended to maximize diversification benefits just when they most needed, as equity-bond correlations tend to decrease during periods of financial turmoil.

To check this point, we looked at how foreign currency bonds had performed during the same periods we used for our analysis of investment grade bonds.

As you can see, foreign currency bonds as an asset class have performed well, across a range of home currencies, time periods, and market conditions. In particular, real returns for this asset class generally display a slightly negative correlation with inflation rates across the full 1971 to 2002 period. The highest correlation across our six currencies is only .13 in Australia.

In spite of these apparent advantages, a recent study ("Diversification, Original Sin, and International Bond Portfolios" by Berger and Warnock of the Federal Reserve) has noted that the home bias in bond portfolios is often quite severe. Consider the following shares of foreign bonds in the Global Market Portfolio (i.e., the market capitalization weighted portfolio that contains all the world’s liquid bonds and equities), as seen from different countries’ perspective:

What could account for this? While many theories have been offered, we suspect that confusion over currency hedging may have a lot to do with investors’ reluctance to hold foreign currency bonds. Let’s start with three leading theoretical arguments in favor of including foreign currency bonds in your portfolio.

First, because local currency returns on foreign bonds have a low correlation with the home currency returns on domestic bonds, there is a potential diversification/risk reduction benefit from holding them. In fact, a recent study ("Credit Risk Diversification: Evidence From the Eurobond Market" by Simone Varotto) found that international diversification caused a bigger reduction in portfolio credit risk than domestic diversification across industries, maturities, or credit ratings.

Second, foreign currency bonds are attractive because of the way their correlation with other asset classes tends to vary over time. Specifically, the correlation between domestic equities and foreign currency bonds tends to be lowest during periods of equity market distress (due to the flight to quality phenomenon). This is exactly the opposite of what happens to the correlation between domestic and foreign equities during periods of distress. In other words, foreign currency bonds tend to provide diversification benefits when they are most needed, while foreign equities tend to do the opposite.

Third, foreign currency bonds provide a natural inflation hedge, because an increase in domestic inflation relative to the rest of the world will result in a depreciation of one's home currency and an appreciation in the value of foreign currency denominated assets. Of course, other asset classes also provide a similar benefit, including real return bonds, and, to a lesser extent, commodities and property.

Unfortunately, the currency hedging issue seems to overshadow these arguments. The key issues involved include (a) whether or not an investor should hedge the exchange risk on his or her foreign currency bonds, and, if so, (b) how extensive that hedging needs to be. On the one hand, there is an argument that completely hedging foreign exchange risk (at a cost of between .25% and .50% per year) reduces the volatility of the foreign bonds asset class much more than its return, and, in the findings from the Berger and Warnock study, results in "the optimally allocated bond portfolio [assuming a one year holding period] being comprised almost entirely of non-U.S. bonds."

However, there are also strong counter-arguments to this view. In "Currency Hedging Over Long Horizons", Kenneth Froot finds that "the ability of hedging to reduce portfolio risk holds at short horizons, but not at long horizons." As he explains it, "at short horizons, exchange rate changes in excess of the forward rate [that is, the rate at which the hedge is entered into when the foreign currency bond is purchased] average about zero, and have virtually no correlation with any variable, including local currency asset returns. Given this, taking currency risk offers no additional reward, and therefore should be hedged away…However, currency hedges have very different properties at long horizons compared to short horizons. The data show that while over short horizons hedging reduces risk substantially, over long horizons hedging often does not reduce risk at all. In fact, at long horizons, many fully hedged international investments actually have greater return variance [i.e., volatility or risk] than their unhedged counterparts."

"Hedge returns at different horizons are driven by very different factors. At relatively short horizons, hedge returns are dominated by changes in real exchange rates (i.e., the purchasing power of one currency compared with another). However, mean reversion in real exchange rates implies that these purchasing powers tend towards parity, so that real exchange rates over time remain roughly constant. At long horizons, hedge returns are instead dominated by fluctuations in cross country dfferences in expected inflation and real interest rate differentials." Given this, "the optimal portfolio hedging strategy will depend on the investment horizon…[and]… investors with longer horizons may want to hedge much less than 100% of their exposure."

Similarly, Campbell, Viceira and White (in their paper "Foreign Currency and Long Term Investors") note that the "conventional wisdom holds that investors should avoid exposure to foreign currency risk." They go on to argue that "the conventional wisdom may be wrong for long term investors." They note that any domestic bond with a maturity (duration) that is shorter than the investor’s horizon is risky, because it must be rolled over (that is, reinvested) at an uncertain future rate. This risk can be hedged by holding foreign currency bonds if the domestic currency tends to depreciate when the domestic real interest rate falls, as implied by the theory of Uncovered Interest Rate Parity. They go on to note that while uncovered interest rate parity doesn’t work well in the short run (as a forecaster of future exchange rates), recent studies find support for it over longer periods. As a result, the authors find much higher foreign bond holdings for long term as opposed to short term investors.

As our model portfolios have a twenty-year horizon, we have included unhedged foreign currency bonds as an asset class, rather than hedged foreign currency bonds.

A more difficult challenge is developing an estimate of the future average annual real returns they are likely to deliver, and how volatile these are likely to be.

We assume that current inflation differentials between countries (estimated from their nominal government bond yield curves) will drive future exchange rate changes, and then use these to estimate future foreign bond returns (using the current relative weights of different bond markets in the Salomon Brothers World Government Bond Index). This yields the following expected returns (in local currency) on foreign currency bonds (as of October, 2003):

Our arguments in favor and against the use of the foreign currency bonds asset class can be summarized as follows: