Investment Research
FX Effects: Currency Considerations for Multi-Asset Portfolios
Juan Mier, CFA, Vice President, Portfolio Analyst
The impact of currency hedging for global portfolios has been debated extensively. Interest on this topic would appear to loosely coincide with extended periods of strength in a given currency that can tempt investors to evaluate hedging with hindsight. The data studied show performance enhancement through hedging is not consistent. From the viewpoint of developed markets currencies—equity, fixed income, and simple multi-asset combinations— performance leadership from being hedged or unhedged alternates and can persist for long periods. In this paper we take an approach from a risk viewpoint (i.e., can hedging lead to lower volatility or be some kind of risk control?) as this is central for outcome-oriented asset allocators. 2
“The cognitive bias of hindsight is The Debate on FX Hedging in Global followed by the emotion of regret. Portfolios Is Not New A study from the 1990s2 summarizes theoretical and empirical Some portfolio managers hedge papers up to that point. The solutions reviewed spanned those 50% of the currency exposure of advocating hedging all FX exposures—due to the belief of zero expected returns from currencies—to those advocating no their portfolio to ward off the pain of hedging—due to mean reversion in the medium-to-long term— regret, since a 50% hedge is sure to and lastly those that proposed something in between—a range of values for a “universal” hedge ratio. Later on, in the mid-2000s make them 50% right.” the aptly titled Hedging Currencies with Hindsight and Regret 3 —Hedging Currencies with Hindsight and Regret, took a behavioral approach to describe the difficulty and behav- Statman (2005) ioral biases many investors face when incorporating currency hedges into their asset allocation. Why Hedge FX? In addition to academics, many industry practitioners have Foreign currencies (FX) are a means of exchange for global trans- tackled the FX hedging dilemma.4 Our study explores this actions. However, FX plays a key function in global investing both question drawing from similar methodologies and benefits from as an asset class in its own right and as a component of financial the data availability of hedged indices for both equity and fixed asset returns. Very simply, global investors need to convert foreign income, data, which to our knowledge, was not extensively avail- currency-denominated investments into their portfolio’s base able in the past. As such, we take a data-driven approach to glean currency. This activity can lead to gains or losses that may be insights from the historical risk-return profiles in equities, fixed totally unrelated to a security’s fundamentals. A hedged position income, and simple 50/50 multi-asset portfolios. can remove the losses when the currency moves adversely, but also mute gains if the currency moves favorably. Data and Methodology The analysis and results center on global developed markets equity While we recognize currency effects can be considered within and fixed income indices. Results are calculated from the point fundamental security analyses—by adjusting valuation model of view of the following developed markets currencies: US dollar parameters—it is also an important concern at the portfolio (USD), Australian dollar (AUD), British pound (GBP), Canadian construction level or for institutional asset owners. Hedging has dollar (CAD), euro (EUR), Japanese yen (JPY), and Swiss franc been a widely discussed subject for many years and interest in the (CHF). Importantly, EUR returns for both equity and fixed topic would appear to fluctuate along with major currencies’ moves. income are only available since its inception in 1999. The main We are approaching this paper by focusing on currency hedging reasons for excluding emerging markets is insufficient availability as a potential tool for portfolio risk control. While we believe that of data. Perhaps more importantly, hedging emerging markets value can be added from active currency management1 we need to currencies in practice can be more expensive and results using separate the return generation goals from the risk control objec- cost-free index data may underestimate actual results by a greater tives—as these two are competing goals. In other words, we will margin than in the developed world. look at asset returns where hedging is “passive” and there is no The time period analyzed spans almost three decades of monthly tactical timing element to put in the hedges or to take active views returns and is restricted by our availability of hedged index on certain currency pairs. data. In equities, we cover the period from December 1987 to December 2017 based on the MSCI World Index. We utilized price returns only, given that total return hedged indices have a shorter history. In cases where hedged indices were not avail- able, we used the “local” return calculated by MSCI. This “local” return represents the theoretical performance by removing all currency effects. This differs from the hedged returns calculation where a specific hedge impact (calculated by MSCI as a hypo- thetical 1-month forward contract) is used to adjust the returns in a given currency. In terms of performance, hedged and “local” returns will be different, but in terms of risk their profiles are virtually the same. With this in mind, using local for risk-based analysis is appropriate wherever hedged indices are missing (we 3
ended up using local returns for the CAD and CHF equity Exhibit 1 cases). In fixed income, we relied on unhedged and hedged Timing Performance from Currency-Hedged Equities Can Be currency versions of the Bloomberg Barclays Global Aggregate Very Challenging Bond Index5 which goes back to January 1990. For multi-asset Rolling 1Y, Annualized results—combining equity and fixed income—we used 50/50 m n s combinations of the respective equity or fixed income indices. n e e All of our data are gross of fees and exclude transaction costs. In t e ms addition, we have focused entirely on index data so the impact of fully hedged returns relies on each currency’s weights in the
benchmarks. In practice, an investor’s portfolio could be more home biased or have different active views on regions that would affect the sizing of hedges and the outcome. However, e e t e ms we believe using standardized, transparent benchmarks is an adequate baseline approach. Rolling 3Y, Annualized Within each currency, we calculated a monthly series of relative returns of unhedged minus hedged indices for different rolling m n s periods to then obtain performance and volatility. While this is n e e t e ms a straightforward approach, very often we see results presented for the entire analysis period or one to two subperiods, which suffer from bias to the start date chosen and mask entry point risk. Given our risk focus, we think looking at rolling volatility can give investors an idea of times when risk may be higher/lower than what is masked by a single, full period summary number. e e t e ms Selected results are presented in the exhibits and expanded in the Appendix. We recognize there are other interpretations of port- folio risk6, but we believe using the standard deviation of returns Rolling 5Y, Annualized is a suitable approach. m n s For the remainder of the paper, references in exhibits will be n e e t e ms labeled with a currency code only, but it represents the return of the underlying index. Hence, if we are talking about equities and label a data point or series “USD,” it means the US dollar return of the MSCI World Index or “USD H” for the hedged USD return of the MSCI World Index. Similarly, in fixed income, e e USD would be the Bloomberg Barclays Global Aggregate Index t e ms and USD H would be the hedged version of that index.
As of December 2017 Equity: Hedging May Not Be Justified Source: Bloomberg, FactSet from a Risk Standpoint The contribution of currency effects to equity performance is useful in practice (i.e., maintaining a 30-year hedging program). an important consideration. As a result, it is tempting to view Therefore, looking at rolling periods of relative returns currency hedging as a possible return enhancer in equity portfo- (unhedged minus hedged) gives us a better sense of different lios. However, a hedging decision can have a positive or negative regimes when hedging has been effective (Exhibit 1). Not only impact depending on the direction of the currency fluctuation. do relative returns change direction, but also the spread can be This leaves portfolio managers with the added complexity of significant—exacerbating the behavioral “regret” risk of being determining hedging entry points. For example, while hedged on the wrong side of the FX hedge. USD equities slightly outperformed unhedged equities (+5.9% versus +5.6%) over the 1987–2017 period, this is not very 4
If past results are a rough guide to future behavior one could Exhibit 2 argue for hedging global equities for Swiss investors and going Summary Results for Equity: Unhedged vs. Hedged unhedged for Japanese investors based on the 3-year and 5-year For the period 1987–2017, monthly observations. Blue = unhedged return results (Exhibit 2). However, overall results are closer outperforms; yellow = hedged outperforms to “50/50” in most cases, highlighting the difficulty of getting % Unhedged outperforms the hedging decision right. This is especially relevant in 1-year e 1Y Rolling "win-loss" ("win") rolling periods, which would be a more typical strategic hedging AUD - AUD H 38 time horizon in practice. CAD - CAD H 50 With this in mind, we think it is more constructive to approach CHF - CHF H 40 currencies from a risk viewpoint for equity portfolio construction EUR - EUR H 47 decisions. In other words, does removing currency risk lower GBP - GBP H 45 a portfolio’s volatility? At first glance, removing the effect of JPY - JPY H 53 currencies should lead to lower volatility. However, this overlooks USD - USD H 50 that the interaction between equity and currency returns will also
drive the volatility. % Unhedged outperforms Looking at 3-year rolling volatility we can see periods where e 3Y Rolling "win-loss" ("win") hedging actually increases risk (Exhibit 3) for AUD, CAD, and AUD - AUD H 45 EUR-based equity investors. While this happens especially in CAD - CAD H 55 times of stress (i.e., 2008–2009) it also appears elsewhere in CHF - CHF H 33 the time series. By hedging, we are removing assets (in this case EUR - EUR H 51 currency) that can provide diversification benefits. The Global GBP - GBP H 42 Financial Crisis helps illustrate this further. Taking the AUD as JPY - JPY H 63 an example, both global equities and the Australian dollar fell at the same time “safe haven” currencies such as the JPY, CHF, and USD - USD H 53 the USD strengthened (to a certain extent) in this period of stress. e % Unhedged As a result, with hedging in place, the AUD-based global investor outperforms e 5Y Rolling "win-loss" ("win") is removing the “safe haven” diversification effects and therefore obtaining a more volatile asset than by remaining unhedged. AUD - AUD H 41 This can be seen in the right panel charts of Exhibit 3, where CAD - CAD H 50 the correlation of hedged equities to the local currency spikes CHF - CHF H 21 higher relative to unhedged equities. As a final point on equities, EUR - EUR H 57 multinational companies also undertake FX hedging programs, GBP - GBP H 38 which further complicates measuring the effect of a hedging JPY - JPY H 69 overlay program. USD - USD H 54
Disentangling this effect will require deep knowledge of the As of December 2017 company fundamentals and analyzing individual historical Source: Bloomberg, FactSet return patterns to gauge if a given stock is already acting as an FX-hedged return series due to the hedging activities of the company itself. This is not an issue in fixed income, where a given bond is representing a cash flow stream in a determined currency. 5
Exhibit 3 Currency-Hedged Equities Can Have Higher Volatility in Certain Cases 3Y Rolling Volatility 3Y Rolling Correlation n e t n n e te t n e te t