Global Bond Strategies and Currency Hedging
A Historical Analysis, 1973 - 1997

Wesley Phoa, Ph.D.
President of Research

We analyze returns both country-by-country and on a portfolio basis, from the standpoint of a US investor, and reach some interesting conclusions. In particular, the evidence in favor of global diversification and currency hedging is overwhelming. Investors who fail to diversify globally, or who fail to hedge currency exposure, incur a massive risk penalty.

The data for this study was provided by Richard Mason and Sean Carmody at Deutsche Bank. Future articles will explore deeper aspects of global bond asset allocation, based partly on important unpublished work due to Mason.

Methodology and first results

Our analysis is based on monthly international bond and bill yields and exchange rates from 1973 - 1997. Data from the US, Australia, Canada, Great Britain, Germany, France, Holland, Switzerland and Japan is used. We compare historical returns and risk for unhedged and hedged global bond investments. Hedged returns assume zero transaction costs, and we also assume that currency hedges are implemented using forward rate agreements, rolled quarterly. The "cost of hedging" is discussed further at the end of this article.

In order to ensure that the analysis is robust, we follow two suggestions due to Mason: we factor out the impact of different debt maturity profiles in different countries by considering only investments in 10-year; and we factor out the impact of asset price bubbles and varying fiscal policies in different countries, we use a "World" bond index weighted by GDP rather than outstanding debt.

Tables 1 to 6 summarize the results for each country considered individually. Tables 1 and 2 compare the inherent risks of local bond markets and currency markets; for every country except Canada, currency risk is significantly greater than bond market risk. Tables 2, 3 and 4 show what this means for a US investor: they compare the risk of unhedged and hedged bond investments. Again, for every country except Canada, hedging currency exposure reduces risk tremendously - in most cases, hedging results in a 50% reduction in risk.

Thus, one would only be justified in taking currency risk if it greatly increased expected returns. There is no reason to believe that this is the case, and in fact table 6 shows that the historical returns from systematic open currency exposure have been inconsistent and not statistically different from zero. Therefore, one should never have open currency exposure unless one has a very strong view on a specific currency.

The preceding analysis was carried out country-by-country. It is possible that the case for currency hedging is weaker when one considers the effects of diversification. In other words, while individual currencies may be very risky, perhaps these risks tend to offset each other in a globally diversified portfolio, so that the risk reduction from currency hedging is lower.

The three graphs show the historical risk/return profiles, of the following assets:

• US Treasuries.
• Unhedged GDP-weighted global bond index.
• Currency hedged GDP-weighted global bond index.
• Post hoc efficient frontier, unhedged global bond portfolios.
• Post hoc efficient frontier, currency hedged global bond portfolios.

The 1970s, 1980s and 1990s have been analyzed separately. In each case, the post hoc efficient portfolios are those which are constructed with the benefit of hindsight, i.e. knowing the realized risks, returns and correlations of each asset.

To begin with, there are two strong conclusions about passive bond investment strategies, which are robust across all three historical periods. First, global diversification adds value - US Treasuries are consistently more risky than the unhedged global bond index, without offering higher returns. Second, currency hedging adds value - the unhedged index is consistently more risky than the hedged index, without offering higher returns. The hedged global index has comparable returns to US Treasuries, with around 50% of the risk.

When looking at the efficient frontiers, one should remember that these portfolios often correspond to very active investment strategies, i.e. holdings are highly concentrated on a small number of countries. The most robust conclusion is that the hedged global index is always close to the efficient frontier, but this is not true for either the unhedged index or US Treasuries. In particular, the hedged global index is an optimal passive strategy, and one should only depart from this strategy if one has a strong view on specific bond markets or currencies.

The remaining conclusions apply mostly to active investors, and vary depending on the historical period. In the 1970s, hedged portfolios clearly dominated unhedged portfolios for comparable levels of risk, though high risk currency bets offered very high returns. In the 1980s, hedged and unhedged portfolios behaved almost like two separate asset classes: unhedged portfolios were two to three times as risky while offering only a few percentage points added return.

In the 1990s, the menu of risk/return options is more limited than it was in the 1970s or 1980s; the very high levels of risk available from developed country investments in those earlier periods would now require significant exposure to emerging markets. Furthermore, hedged and unhedged strategies now appear to form more of a continuous spectrum, with hedged portfolios dominating at lower risk levels and currency bets gradually becoming more attractive as risk tolerance rises.

It may be of interest to look at the composition of portfolios on the efficient frontier. The first thing to note is that optimal low risk hedged portfolios tend to be reasonably diversified. This is consistent with the observation that the hedged global index is always close to efficient, and generally close to minimum risk.

The exception was in the 1980s, where one could have achieved lower risk by heavily overweighting Switzerland. Interestingly, low risk hedged portfolios have a consistently high allocation to Swiss bonds in all three periods. However, this would have to be regarded as a very active strategy, since Swiss bonds form only about 1.5% of the index.

Optimal low risk unhedged portfolios correspond to a plausible passive unhedged strategy: they tend to have around 50% - 60% allocated to US Treasuries with the remaining holdings diversified. This is unsurprising. However, it is important to note that the minimum risk unhedged portfolio is always dominated by a suitable hedged portfolio. We can conclude that the passive unhedged strategy is never optimal.

Optimal high risk hedged portfolios correspond to large bond market bets: on Germany and Switzerland in the 1970s, on France, Japan and the US in the 1980s, and on Australia and Japan in the 1990s. Prior to the 1990s, optimal high risk unhedged portfolios corresponded to large currency bets. In the 1990s, however, the optimal high risk unhedged strategy was a more sophisticated combination of a bond market bet (on Australia) and currency bets (on France and Germany). Optimal active strategies seem to be growing more complex.

There is a common misconception about currency hedging. If an investor hedges a foreign bond holding with (say) a 3-month FRA, and the foreign currency appreciates, then rolling the hedge at the end of 3 months requires a net cash payment. Thus, it appears that there is a "cost" to rolling the hedge, quite separate from the FX dealer's bid/offer spread.

This is an illusion. If the currency has appreciated, the value of the foreign bond has risen in USD terms. Rolling the hedge on the entire bond holding is equivalent to increasing one's USD allocation to the foreign market, which corresponds to a change of strategy - in fact, it is a form of trend-following. In order to maintain a constant percentage USD allocation, one should instead sell an appropriate quantity of the foreign asset.

The conclusion is that the only true costs involved in hedging currency exposure are transaction costs; and in today's currency markets, these are minimal. Thus, passive investors have absolutely no excuse for maintaining open currency exposure. More detailed calculations will appear in a future article.