It is widely accepted that investment portfolios should have an international allocation. An investor wishing to buy a foreign asset is required to purchase the currency of that country to settle the transaction. This second decision, or hidden purchase of currency in international investment, has a significant risk associated with it. Irrespective of the asset return, currency values can fluctuate to have an impact on the market value of the investment in base currency terms.
Historically, 30% of the volatility (as measured by standard deviation) of an international equity portfolio has been associated with its inherent currency exposure. The equivalent statistic for international fixed income is 60%. This risk has been persistent and fairly stable since the most recent period of floating exchange rates began in 1973 (see Figures 1 and 2).
Unlike financial assets, the risk that this currency exposure brings is not rewarded with a long run return. Economic theory suggests that the long-term currency return (in excess of the interest rate differential) should be zero. Empirical evidence on our international portfolio supports this theory.
Moreover, when measured on a monthly basis from 1973–2000 the correlation between local equity and currency is –0.01. A low number is expected. Conceptually, currency is a medium of exchange between two national money supplies, the demand of which depends on the relative demand for goods and assets across economies. Therefore exchange rates are driven by differences in price levels, differences in quality of traded goods and differences in expected returns on assets. These fundamental drivers are clearly different from those pertaining to the local asset markets.
Individual exchange rates, like many price series, can exhibit very short-term price moves that are too large to be explained by a normal distribution. However, when viewed over periods of a month or more, or when viewed as a portfolio of exchange rates, or both, evidence suggests that currency returns are indeed normally distributed (see Figure 3).
In summary, currency exposure in international investment represents a significant risk, offers no strategic return and is uncorrelated with the underlying asset return.
The implication is that the currency exposure inherent in international investment ought to be managed and not simply assumed as a hidden investment. Because of the separate nature of currency from assets one should and can prudently unbundle currency from assets and explicitly resolve for currency the main policy issues that need to be addressed for any separate asset class. These are:
q What is the appropriate strategic or long-run exposure to currency?
q Should currency be managed actively around this strategic position?
q Who should manage currency exposure and how?
It is important to emphasise that these currency policy issues, while they may appear new and alien, are in fact issues that most pension plans have addressed already and have implicitly made decisions about. In most cases, by default the investor has chosen to have a currency exposure equal in size to the international allocation; the investor has chosen to engage in active currency management by allowing currency exposures to change through time as the asset managers change their country weightings; and the investor has chosen to take equity managers’ (and/or fixed income) country weighting decision process to be the appropriate approach for actively managing the currency portfolio.
Until investors clearly understands the separate and independent nature of each of these decisions, it is unlikely they will approach the strategic currency decision in an appropriate way. We provide below the appropriate methodology for addressing the first issue: How much currency exposure should I be hedging as a strategic policy decision?
It is our view, and now industry standard practice, that a total portfolio asset allocation study is the appropriate way of deciding the strategic hedging policy. In making the decision it is useful to recall that our international investment consists of two components: hedged assets and currency. By viewing the entire portfolio as a portfolio of assets, both domestic and foreign but containing no currency exposure, plus the currency exposure, we can separately consider asset allocation and currency hedging.
With expectations of return, risk, and correlations plus the risk preference, we simply use mean variance optimisation to derive an optimal currency exposure. Moreover, we can use asset information to derive an optimal currency exposure that is consistent with the investor’s risk preference as indicated by the asset mix.
Clearly the optimal hedge ratio (the amount of currency exposure removed from the portfolio relative to the initial amount) will be unique to each portfolio and will be a function of risk preference and expectations of returns, volatilities, and correlations. It is then inappropriate to recommend a single hedge ratio for all investors, but instead a study should be undertaken to generate a hedge ratio consistent with the overall objectives of the plan.
Each investor should undertake such analysis to identify the hedge ratio that is consistent with its overall plan characteristics.
A criticism of the mean variance approach is that it is sensitive to input values. With an expected currency return so close to zero, only a small change in this expectation can change the optimal hedge ratio. However, with a completely neutral currency outlook we can make some general statements on currency hedging:
q Partial hedging is optimal for a typical 60/40 portfolio plan with significant international exposure.
q As the international exposure increases, the hedge ratio also increases as currency has a greater impact on total portfolio risk.
q As the asset allocation moves more into fixed income than equity the investor is illustrating a lower appetite for risk and more hedging becomes optimal. Additionally, there has been a small positive correlation between domestic fixed income and currency (whereas between domestic equity and currency the correlation appears close to zero) and so currency becomes less of a diversifying investment. Therefore more hedging becomes optimal.
q Expectations of equity returns and fixed income returns are also required inputs in the strategic hedging study. As the difference between these two expected returns increases, so equity becomes more attractive than fixed income, and the asset allocation comes to represent an increasingly risk averse investor. The hedge ratio therefore increases. The same argument holds for a decrease in the difference in volatility between equity and fixed income.
q As the correlation between currency and any of the assets increases, so the diversifying property of currency diminishes and the hedge ratio increases. The sensitivity of the increase depends on the size of the allocation to the asset.
If liabilities are included in the investment portfolio construction process then we can incorporate them into the strategic currency study in an identical way to the incorporation of them into an asset/liability allocation study. If the liabilities have a high correlation with equity assets, then a high equity proportioned asset portfolio now reflects a more risk adverse investor and more hedging is appropriate. If the liabilities have a high correlation with fixed income assets then the risk aversion of the investor is little changed, but since currency has a small correlation with fixed income, holding currency will diversify some of the liability risk and less currency hedging is optimal. With the crude approximation that liabilities for retirees and active members are like fixed income and equity respectively, the level of hedging in an asset liability framework thus depends on the maturity of the fund.
The fact that the long run currency returns appear normally distributed means that mean variance analysis is entirely appropriate for the strategic hedging study. However, the asymmetry of option protection has excluded option hedging from such a framework. It is useful then to note that the return distribution of an option-based hedging strategy tends to a symmetric normal distribution in the long run and the mean variance framework is still valid. This is an important point for the investor to understand: short term (eg annual) protection programmes do not provide long term (eg five to 10 years) return protection. In fact, assuming a neutral currency outlook, there is a one-to-one equivalence between each hedge ratio and an option protection programme with varying levels of protection. As a rule of thumb, a fully hedged, half hedged and unhedged portfolios are equivalent to in-the-money, at-the-money, and out-of-the-money option protection respectively.
The investor should use the mean variance study to select the appropriate hedge ratio, then use another technique (such as Monte Carlo integration) to decide on the instrument to implement the hedge: forwards, options, or a combination of them both.
In summary, investors should unbundle currency from the assets that gave rise to this exposure – address this exposure strategically as one would any separate exposure, using portfolio optimisation techniques. The strategic exposure decision is the most important decision an investor makes about assets and it is only after that decision that he must address the other two decisions:
q Active versus passive currency management, and
q Who should manage such exposure – asset manager or overlay specialist?
Adrian Lee is president and CIO and David Buckle is head of research at Lee Overlay Partners in Dublin