The notion that a currency hedging strategy be determined with respect to a portfolio’s underlying asset allocation is a popular one amongst institutional investors. Taking a total portfolio perspective the following analysis sets out to challenge this rather narrow approach. In doing so our broader analysis sheds light on the importance of the foreign currency in which a portfolio’s assets are denominated as well as to the choice of hedging strategies ultimately employed. In particular we show how treating each currency separately and evaluating how combining the different hedging approaches available (passive, dynamic & no hedge) can simultaneously increase returns and reduce risk.
In order to explore to what extent the choice of a hedging policy affects the risk and return characteristics of an investor’s portfolio, we build an asset allocation that is representative of a typical Swiss pension fund.*
The following two graphs illustrate this allocation at the end of 2018 in terms of asset classes and currencies.
* Data on asset allocations were obtained through surveys conducted by Swisscanto and UBS. Alternative assets were discarded due to the lack of publicly available and representative return data. The indexes used to compute performances are listed in the appendix.
For the purpose of this study, we simplify the currency allocation of the foreign assets to the USD and EUR. Turning our attention to the currency hedging policy of this typical Swiss pension fund, we assume that currency risk is hedged to a level corresponding to the bonds allocation in that currency. In other words, if the exposure to the EUR is split 40% into bonds and 60% into equities, then 40% of the EUR exposure will be hedged (passively). Similarly, for the USD. This assumption is supported by data from the various Swiss pension fund allocation surveys as well as recommendations made by consultants* and anecdotal evidence gathered from our own clients.
The rationale behind this approach is twofold. First, academic literature fails to show that currency risk is associated with a premium, in other words it doesn’t pay on average to bear currency risk. Secondly, the volatility of the exchange rate is disproportionately large when compared to the volatility of the returns of foreign bonds in their local currency. This stands in contrast with foreign equities where the volatility of the returns in local and domestic currency is roughly the same hence why currency risk is left unhedged.
In this study, we want to test this well-established view by evaluating its historical performance in the context of a portfolio of foreign assets. To do so, we estimate the value of the foreign assets allocation of our representative portfolio where the bonds allocation is hedged for currency risk and the stocks allocation is not, for every month between January 2011 and September 2019. We plot the cumulative performance in blue in the graph below.
* See for example : Quelle stratégie pour les risques de change ?, Banco (juillet-août 2019), p.23
While it is true that a 94% gain over nearly 9 years looks very attractive, one must evaluate this performance in relative terms. In order to do this we simulate three additional currency hedging policies over the same portfolio of foreign assets to yield a total of four distinct trajectories:
Several observations can be made from the previous graph. First, the performance of the Full Passive hedge (red) is the lowest but also the least volatile. This makes sense intuitively since it hedges the most currency risk at the cost of the interest rate differentials between the foreign currencies and the CHF. In recent years in particular, the USD/CHF interest rate differential has been a huge performance drag for a Swiss-based investor.
In contrast, the Full Dynamic hedge yielded the highest performance yet with more volatility. This reflects the manager’s trading style of adjusting the hedge according to temporary opportunities and risks reflected on the market, thereby yielding additional returns with some added volatility. Finally, the Partial and Zero hedge come out very close in terms of cumulative performance, although the former is less volatile as expected.
In conclusion, the graphical illustration clearly and intuitively highlights that hedging more currency risk reduces both the volatility and the returns of the portfolio of foreign assets. As such, this illustration does not allow us to rank the four hedging policies appropriately. In order to do so, we must compare them on the basis of risk-adjusted returns. The following table reports the annualized return, volatility and risk-return ratio of each hedging policy:
|Annualized return||Annualized volatility||Risk-return ratio|
|Full Passive Hedge||7.13%||6.89%||1.03|
|Full Dynamic Hedge||9.07%||7.91%||1.15|
The return and volatility numbers echo our previous remarks. The risk-return ratios lead to an interesting finding: the Full Dynamic hedge yields the largest risk-adjusted returns while the opposite is true for the Zero Hedge. The Full Passive and Partial Passive hedge fall somewhere in between, with a small edge for the latter. These results are consistent with the widespread idea that bearing currency risk yields no premium, but with an important nuance: disciplined currency risk management in the form of dynamic hedging does indeed provide a risk premium alongside a risk reduction. Indeed, deploying this hedging approach in the currency markets allows investors to harvest attractive risk-adjusted returns.
If one adopts the view that currency risk bears no premium and thus decides to passively hedge the currency exposure, one faces the harsh truth that currency risk hedging bears a cost. This is highlighted here by the low returns (in absolute and risk-adjusted terms) of the Full and Partial Passive hedging policies.
Finally, while we do not dispute the notion that currency risk represents a disproportionate share of the risk of foreign bonds returns translated in domestic currency, we argue that the choice of the hedging policy will have a considerable impact in terms of risk and returns at the level of the portfolio (whether the whole one or the portfolio of foreign assets). Because these effects are a function of the hedging policy applied to a particular currency, the analysis should be undertaken at the currency level rather than at the underlying asset class level.*
* In our example, we said that 40% of the EUR allocation was hedged because 40% of the EUR allocation was made of bonds. But when we compute the returns of the aggregate EUR portfolio in CHF, only the size of the hedge matters and the idea that the hedge was applied solely to the bonds allocation (as opposed to the stocks allocation) becomes meaningless.
Although we have highlighted the ability of dynamic hedging to deliver superior risk-adjusted returns for the portfolio of foreign assets, we must consider the fact that a fully dynamic policy entails a level of risk and net currency exposure that investors such as pension fund may potentially find too high. Moreover, leaving the currency exposure unhedged at times can result in a pension fund finding itself beyond the OPP 2 guidelines in terms of maximum authorized currency risk exposure (30%).
To adjust for these potential challenges, we conducted an optimization exercise over our sample in which we look for the best hedging policy for the portfolio of foreign assets given a tolerable level of risk. Assuming that Swiss pension funds are comfortable with the risk of their actual portfolio of foreign assets, we set the risk level parameter to be 7.30% (i.e. the annualized volatility of the foreign portfolio with the Partial Passive hedge policy).* Our findings are summarized in the table below. The portfolio with the optimized hedging policy yielded an additional 1.34% annualized return (12.35% cumulated) which increased the risk-return ratio by 18% with respect to the original case.
* As an ex-post optimization, this process has a look-forward bias which may inflate its performance. On the other hand, the mix of hedging approaches is fixed throughout the period, irrespective of market conditions, which may tilt the performance to the downside.
|Annualized return||Annualized volatility||Risk-return ratio|
Recognizing that both portfolios are made up of the same unhedged portfolio of foreign assets plus a hedging policy layered on top (“overlay”), it is useful to isolate the two hedging policies and compare them in terms of their risk and return contributions. The graph below reveals how the optimized overlay actually reduces the risk of the unhedged portfolio while simultaneously adding returns, whereas the original overlay sacrifices returns for risk reduction.
What does the optimized overlay program look like? We represent below the original and optimized overlay programs in stacked bars for each currency.
As the original overlay hedged the currency exposure to a level equivalent to the bond allocation (and because we assumed that the relative share of bonds versus stocks was the same for each currency), around 40% of the EUR and USD are hedged at all times, leaving the remaining 60% unhedged. On the other hand, the optimized overlay program treats each currency separately and evaluates how combining the different hedging approaches available (passive, dynamic & no hedge) can yield the best returns possible at the portfolio level for the set level of risk. As a result, 100% of the USD exposure is hedged dynamically (meaning the hedge ratio is allowed to fluctuate between 0 and 100%) whilst 53% of the EUR exposure is hedged passively and 47% dynamically (i.e. the total EUR hedge can vary between 53% and 100%, virtually ensuring that the OPP 2 constraints at the portfolio level are always fulfilled).
The results of this optimization exercise can be explained intuitively: the USD/CHF interest rate differential has largely outweighed the EUR/CHF one over the period under consideration, as such it makes sense to hedge a good share of the EUR passively in order to reduce the volatility of the portfolio towards the target level (7.30% annualized). Beyond that, applying a dynamic hedge to the remaining share of both currencies provides added returns and risk reduction.
In this research paper, we have argued that when considering the implementation of a currency hedging policy on a portfolio of foreign assets, the decision must be evaluated at the currency level. Indeed, currencies and the associated hedging approaches (no hedge, passive hedge and dynamic hedge) have very different risk and return characteristics that should be combined in an efficient way in order to provide the best return-boosting and risk-reducing enhancement possible to the portfolio.