This action might not be possible to undo. Are you sure you want to continue?
May 26, 2010
Managing FX hedge ratios
A framework for strategic and tactical decisions • Unprecedented volatility over the past two years has heightened investor and corporate attention towards managing FX hedge ratios. • For investors, four issues predominate: (1) how to determine the longterm, optimal hedge ratio; (2) how to time entry into a hedging program; (3) how to deviate from the strategic hedge ratio to generate profits or manage cash flows; and (4) how to choose between forwards and options in implementing a hedging program. • For corporates, the most frequent concerns are variants of the first, second and fourth points: how to set a baseline range for hedge ratios; which currencies are most likely to post large moves higher or lower; and which instruments are best for executing a hedging program. • Previous J.P. Morgan research has discussed optimal hedge ratios and proposed currency models based on long-term valuation and short-term momentum. This paper updates those studies, adapts the models for dynamic hedging over various horizons, and extends the original analysis to four base currencies (USD, EUR, GBP, AUD). • The conventional wisdom that hedging currency risk minimizes portfolio volatility was proven false at some points during the credit crisis, when hedged portfolios in some currencies became more volatile than unhedged ones. A more dynamic strategy is therefore required to avoid onerous cash flow requirements. • Dynamic hedging should begin with signals from long-term fair value models, to indicate which currencies merit the most attention over a 6 to 18-month horizon. Investors can use these signals to time entry into hedging programs, while corporates can use them to focus on currencies most vulnerable to large moves in either direction. • Short-term trading models commonly used by currency overlay managers and global macro funds can be adapted to drive deviations around the benchmark hedge ratio over a one to three-month horizon. A price momentum model which adjusts dynamically around a 50/50 hedge ratio outperforms the benchmark by about 100bp annually, depending on the base currency. Information ratios on the strategy range from 0.2 to 0.5. A rate momentum (forward carry) model generates comparable outperformance but with more consistency across sample periods. • Data limitations constrain long-term analysis of the efficiency of forwards versus options in a hedging program. But since 2003, options have outperformed forwards for AUD, EUR and GBP-based investors with USD exposure. A signaling model that switches between forwards and options is inconclusive.
www.morganmarkets.com/GlobalFXStrategy J.P. Morgan Securities Ltd.
(44-20) 7325-5222 email@example.com
Gabriel de Kock
(1-212) 834 4254 firstname.lastname@example.org
(1-212) 834 4565 email@example.com
(1-212) 834 2304 firstname.lastname@example.org
I. Overview: Four common questions on hedging policy II. The merits of long-term FX exposure III. Using fair value models to focus strategic hedge ratios IV. Using alpha models to adjust hedge ratios tactically V. Hedging with forwards versus options 22 34 16 2 4
The certifying analyst is indicated by an AC. See page 38 for analyst certification and important legal and regulatory disclosures.
Global FX Strategy Managing FX hedge ratios May 26, 2010 John Normand (44-20) 7325-5222 email@example.com J.P. Morgan Securities Ltd.
I. Overview: Four common questions on hedging policy
Unprecedented volatility over the past two years has heightened investor and corporate attention towards managing FX hedge ratios. For investors, four issues predominate: • how to determine the long-term, optimal hedge ratio for global stock and bond portfolios; • how to time entry into a hedging program, to focus on the most expensive currencies and avoid covering exposure in undervalued ones; • how to deviate from the strategic hedge ratio to generate profits or manage cash flows; • how to choose between forwards and options in implementing a hedging program. For corporates, the most frequent concerns are variants of the first, second and fourth points: how to set a baseline range for hedge ratios; which currencies are most likely to post large moves (higher where they have expenses, lower where they have earnings); and which instruments are best for executing a hedging program. Academic studies and previous J.P. Morgan research over the past two decades have discussed optimal hedge ratios in detail.1 More recent J.P. Morgan studies have proposed long-term valuation and short-term momentum models which can be adapted to answer the second, third and fourth questions (see blue box on next page). This paper updates those studies, adapts the models for dynamic hedging over various horizons, and extends the original analysis to four base currencies (USD, EUR, GBP, AUD). Section II reviews and critiques the conventional wisdom on optimal hedge ratios for those unfamiliar with the framework. Section III applies J.P. Morgan’s long-term fair value model to answer the investor question of how to time entry into a hedging program, and the corporate question of which currencies merit hedging focus over a 6 to 18-month horizon. Section IV modifies short-term FX trading models (alpha models) based on price momentum and interest rate
Early J.P. Morgan publications include Currency-hedged international fixed income investment, Peter Rappoport (1990); Managing currency risk in global portfolios, Jan Loeys (1999); and Introduction to Portfolio Management, John Normand (2002 and subsequent revisions).
momentum to drive tactical deviations around a 50/50 hedged benchmark over one to three-month horizons. Section V concludes with a discussion of how these signals can be used to inform the choice of forwards versus options in a hedging program. We find that: • The conventional wisdom that hedging currency risk minimises portfolio volatility was proven false at some points during the credit crisis, when hedged portfolios in some currencies became more volatile than unhedged ones. The cash flow implications for some funds was disastrous, hence the necessity to consider more dynamic frameworks for managing hedge ratios over time and across currencies. • Long-term fair value models can be the starting point for dynamic hedges, since they can indicate the most vulnerable currencies over the medium term. J.P. Morgan’s original fair value model (2008) has provided reliable signals for identifying and trading misaligned currencies versus the dollar over 6 to 18-month horizons. That analysis is extended to base currencies such as EUR, GBP and AUD. Investors can use the model’s signals – reported quarterly in flagship publication World Financial Markets and FX Markets Weekly – to time the entry into currency overlay programs by raising hedge ratios on the most overvalued currencies, lowering them on the cheapest currencies and remaining at benchmark on others. For corporates, the signals highlight currencies on which hedge ratios should be raised, because the company has expenses in undervalued currencies or earnings in overvalued ones. • Over shorter horizons (one to three month rebalancing), tactical trading models commonly used by overlay managers and global macro funds can inform deviations from the benchmark hedge ratio. A price momentum model which adjusts dynamically around a 50/50 hedge ratio outperforms the benchmark by about 100bp annually, depending on the base currency. Information ratios on the strategy range from 0.2 to 0.5. A rate momentum (forward carry) model generates comparable outperformance but with more consistency across sample periods. • Data limitations constrain long-term backtesting to compare the efficiency of forwards versus options. Still, since 2003 hedging with options has outperformed forwards for AUD, EUR and GBP based investors with USD exposure. A signalling model which switches between forward and options is inconclusive and will be addressed in more detail in forthcoming research.
Global FX Strategy Managing FX hedge ratios May 26, 2010 John Normand (44-20) 7325-5222 firstname.lastname@example.org J.P. Morgan Securities Ltd.
Previous J.P. Morgan publications on currency overlay, indices and models available on www.morganmarkets.com/GlobalFXStrategy, or by clicking the hyperlinks below.
Introduction to Portfolio Management, John Normand, 2002 and subsequent revisions. J.P. Morgan effective exchange rates: revised and modernized, Derek Hargreaves and Carl Strong, May 30, 2003. JPMorgan’s FX Barometer, Normand, Mustafa Caglayan, Dan Ko, Nikolaos Panigirtzoglou and Lei Shen, September 22, 2004. Introducing the JPMorgan VXY & EM-VXY, Normand and Arindam Sandilya, December 11, 2006. JPMorgan Tradeable Currency Indices, Normand, July 2, 2007. Rotating Between G-10 and Emerging Markets Carry, Normand, Jul 9, 2007. A new fair-value model for G10 currencies, Gabriel de Kock, September 6, 2008. Trading and Hedging Long-Term FX Fundamentals with J.P. Morgan’s Fair-Value Model, de Kock, April 24, 2009. Alternatives to standard carry and momentum in FX, Normand and Ghia, August 8, 2008. The month-end effect in FX: small but predictable, Normand, October 23, 2009.
Global FX Strategy Managing FX hedge ratios May 26, 2010 John Normand (44-20) 7325-5222 email@example.com J.P. Morgan Securities Ltd.
II. The merits of long-term FX exposure2
• The conventional wisdom is that unhedged FX exposure raises volatility more than returns over the long run. This conclusion tends to hold regardless of base currency. There are three exceptions to this conclusion: emerging markets exposure for long-term FX appreciation; G-10 forex exposure as catastrophe insurance; and G-10 forex exposure to diversify portfolio risk. The more-risk-than-return view usually motivates high hedge ratios on fixed income investments. FX hedge ratios on equities are more variable, however. Some investors will hedge to minimise volatility. Others will not due to FX’s small marginal contribution to an asset class which is already quite volatile. Cash flow implications can also deter hedging. The credit crisis has forced a rethink of the conventional wisdom and standard practices. USDbased investors suffered massive losses on unhedged portfolios as the dollar strengthened/foreign currencies weakened. Non-US investors who hedged USD exposure incurred significant cash flow obligations from being short the dollar as it appreciated. These developments now motivate many to consider a more dynamic approach to FX risk management.
The notion that passive long-only currency exposure offers no long-term return stems from two conditions in international finance known as covered and uncovered interest parity. Covered interest parity (CIP) states that the interest rate differential between two countries should equal the forward premium (discount), expressed algebraically as rforeign - rdomestic = (FXforward - FXspot)/ FXspot [Equation 1] This relationship is a no-arbitrage condition rather than a theory. Consider the following example. Assume 12-mo Japanese rates are 1%, US rates are 4%, and USD/JPY spot is 110. The no-arbitrage condition in equation 1 implies that the 12-mo forward rate must equal 106.70. If it were 110, an investor could lock in a guaranteed profit by borrowing JPY at 1%, buying a USD deposit at 4% and selling USD forward at 110 for no-risk returns. Uncovered interest rate parity (UIP) is an extension of CIP. If markets are efficient, the forward rate will be an unbiased estimator of the future spot rate. rforeign - rdomestic = (FXexpected - FXspot)/ FXspot [Equation 2] If this condition holds, then there should be no return to taking currency risk, since high (low) interest rate currencies should depreciate (appreciate) to the level of the forward rate. CIP tends to hold in practice, as free capital mobility among the major markets enforces the no-arbitrage condition. The evidence on UIP is mixed, however: the passive return from G-10 currency exposure tends to be negligible over the long-term, but sometimes significant over the short term. One way to illustrate this no-return concept is to consider asset market returns measured from three perspectives: local currency, unhedged into a given base currency and hedged into a base currency. The difference between local currency and unhedged returns is the foreign currency’s contribution to total returns. A zero difference would support the view that currency exposure does not augment returns, so perhaps should be hedged if volatility differentials are high. USD-based investors For USD-based investors, the return differential between local currency and unhedged stocks and bonds is modest over the long-term (1988 - 2009).3 Currency moves have augmented asset returns by roughly 1% for investments in the Euro area, Canada, Australia and Japan over the past twenty years, due to those currencies’ appreciation versus the dollar (charts 1 and 4). Currency has subtracted roughly
Strategic FX exposure: high risk and low return, with three exceptions
The conventional wisdom on FX makes two claims: that FX exposure delivers more risk than return over the longrun; and that FX offers abundant short-term profit opportunities due to structural inefficiencies. If these points are correct, then investors should fully hedge FX exposure but run active overlay programs to capture short-term profit opportunities. Likewise, corporates should hedge as a matter of policy but alter target ratios over shorter horizons when they hold strong directional views.
This section updates and extends analysis on currency management originally published in Introduction to Portfolio Management, John Normand (2002 and subsequent revisions).
Sample period chosen by the availability of hedged and unhedged return indices across G-10 markets.
unhedged World ex US. % Source: J. 1988 .P. Morgan Source: J. 2010 John Normand (44-20) 7325-5222 john. 1988–2009 Return.com J. 1988-2009 12% 9% local ccy unhedged hedged Chart 4: Bond returns hedged vs unhedged into USD. 1988-2009 30% local ccy unhedged hedged Chart 5: Bond volatility hedged vs unhedged into USD. Morgan 5 .Global FX Strategy Managing FX hedge ratios May 26. Morgan Securities Ltd.2009 Return. hedged 5 10 15 20 25 Risk. 1988-2009 15% local ccy 12% unhedged hedged 20% 9% 10% 6% 3% 0% USD JPY EUR GBP AUD CAD MSCI exUS 0% USD Source: J.normand@jpmorgan. 1988-2009 12% local ccy unhedged hedged 9% 6% 3% 0% -3% USD Source: J. Morgan Chart 3: Equities for USD-based investors: return vs risk. Morgan Source: J.P.P.P.P.P. Morgan 6% 3% JPY EUR GBP AUD CAD MSCI ex-US 0% USD JPY EUR GBP AUD CAD GBI ex US Chart 2: Equity volatility hedged vs unhedged into USD. Stock and bond returns and volatility with USD as base currency Chart 1: Equity returns hedged vs unhedged into USD. hedged 10 US equities 5 World ex US. % - World ex US. unhedged 4 2 - 5 10 15 Risk.P. % 15 Chart 6: Bonds for USD-based investors: return vs risk. Morgan JPY EUR GBP AUD CAD GBI ex US Source: J. % 8 6 US bonds World ex US.
As a group. Though FX’s return impact may be modest. Unhedged equity exposure in the US. Others would consider a hedging program cumbersome and expensive since FX’s marginal contribution to equity volatility over the long run is small. For Euro area investors. Exception 1: emerging markets as a source of long-term returns Unlike G-10 currencies which tend to mean-revert. Unhedged foreign stocks are 3 to 10 percentage points more volatile (appendix 1. consider three exceptions to the conventional wisdom that passive FX exposure represents uncompensated risks. unhedged bond market returns are two to three times more volatile than local currency returns (appendix 3.3% per annum on Euro area investments to -1. Currency moves have had almost no long-term impact on Japanese stocks holdings given that EUR/JPY was close to unchanged between 1989 and 2009 (appendix 1. 1. Chart 8 shows the Sharpe ratio on the index and its regional sub-components . We track these excess returns through JPMorgan’s Emerging Local Markets Index (ELMI). unhedged non-UK equities (MSCI ex-UK) add 1. UK and Canada are roughly 2. T-bills.P. chart 2). and Euro area bonds are 6 points more volatile. the underlying asset class is already highly volatile.8% higher. and those in Japan 1. chart 2). chart 5). This combination of real exchange rate appreciation and interest rate differentials can persist for years (chart 7). interest rates also tend to be above those in G-10 markets. Since unhedged bond market returns are two to three times more volatile than local currency returns.4% more volatile. As in other countries. its volatility impact can be significant. US stock volatility is only 0. however. emerging market currencies can offer trend positive returns. The volatility of unhedged returns in UK. higher return volatility offsets this return advantage. Unhedged Canadian and Australian stock returns are 5% more volatile. Canadian and Australian bond markets are roughly 4 points higher than local currency returns (chart 5). reflecting a higher marginal productivity of capital. but Euro area equity exposure is 0. For Australian investors. the difference in annualised volatility between local currency returns and unhedged returns can be massive: 7% . GBP and AUD-based investors Similar conclusions hold for investors with other base currencies. charts 1 and 4). Emerging markets currencies often experience long-run real appreciation due to a faster rate of productivity growth (the convergence process). and Japanese exposure has delivered small currency gains (roughly 0. This volatility differential is much more meaningful in bonds than in stocks.4% for US and less than 1% for Australian and Canadian stocks.9% per annum to stock and bond returns over the past twenty years due to sterling’s trend decline (appendix 2.com J. Some investors primarily concerned with risk minimization would hedge. chart 1). chart 2). In equities.1. however.normand@jpmorgan. unhedged overseas exposure has reduced returns by roughly 2% per annum for stocks in the UK. the hedging decision is less clear-cut.5% more volatile than local currency returns. Exposure to Canadian stocks and bonds has resulted in 6 almost no currency loss given the steadiness of the AUD/CAD cross.1% per annum to returns but are 2. unhedged foreign currency exposure has added 1% . chart 5). Volatility differences are substantial. the Australian dollar’s trend appreciation over the past two decades has resulted in a currency loss from unhedged foreign exposure ranging from -0. 2010 John Normand (44-20) 7325-5222 john.Global FX Strategy Managing FX hedge ratios May 26. FX as catastrophe insurance and FX as a portfolio diversifier. catastrophe insurance and portfolio diversifier Within the G-10. These exceptions concern emerging markets exposure. But before delving further into equity investors’ hedging decision. whether measured in local currency or unhedged terms (chart 2). it is generally accepted that FX exposure brings uncompensated volatility over the long run. unhedged exposure is two or three times more volatile than local currency returns (appendix 2.8% for Australian and Canadian exposure. In many cases. and thus generates a meaningful return differential over time to justify their higher risk. This excess return stems from two sources — real appreciation and carry. For UK investors. 1% from UK investments due to sterling’s depreciation since the late 1980s. Volatility differentials on bonds are also substantial. which measures the return in dollars of cash instruments (Libor. FX forwards) in 24 emerging markets. a strategic – or benchmark – policy of fully hedging currency risk is sensible. but Euro area and Japanese exposure is 2%-3% less volatile (appendix 3. Morgan Securities Ltd. Only Euro area stocks exhibit comparable return volatility. however. Unhedged JGB returns in USD terms are 7 percentage points more volatile than local currency returns. 3% for UK exposure and 1% for Japanese exposure. while foreign bonds are about two times more volatile (appendix 1. For stocks.7% more.2%). Volatility differences have been significant. At the same time. a risk premium for convertibility and a policy tool to promote disinflation. Three exceptions: emerging markets. EUR. since government bonds are already a relatively low-volatility asset class.8% less volatile (appendix 2.7% per annum on UK investments (appendix 3 charts 1 and 4). charts 2 and 4). For most bond markets.
42 0.12 0. Morgan ELMI+ index 20% 15% 10% 5% 0% -5% -10% -15% -20% -25% -30% 94 96 98 00 02 04 06 08 -10% -9% -24% return from FX appreciation/depreciation return from carry uncertainty or financial crises.Global FX Strategy Managing FX hedge ratios May 26.10 ELMI+ -0. many insurance companies often have significant. Morgan Exception 2: catastrophe insurance and asset/liability matching require unhedged FX exposure Strategic FX exposure is also useful as a hedge against local event risks. foreign [Equation 4] Hedging out the currency risk leaves the variance of the portfolio as a function of the individual asset variances and the covariances between the asset returns. (4) covariance between the domestic and foreign assets. (2) foreign asset’s variance.P. For example. In the event of political 7 . unhedged USD holdings. Under certain conditions.50 0. 1994-2009 Based on returns from J. Similarly. Equation 3 represents the variance of the unhedged (or partially hedged) portfolio.5%. Since the variance of a portfolio is the sum of individual variances plus a measure of their comovement. fx ) < 0 [Equation 5] 13% 13% 10% 17% 15% 4% -5% -7% 6% 7% 14% 7% -6% 8% 6% 6% 4% -3% 2% 9% 1% 2% -12% 5% 3% -10% -11% Source: J. the hit to their balance sheets would be partially offset by gains on the local currency value of foreign holdings.P.5% annually. unhedged exposure can reduce overall portfolio volatility.P. Exception 3: FX exposure may reduce portfolio risk The third exception which justifies holding unhedged benchmark exposure is for risk diversification: currency and asset risk may be sufficiently negatively correlated so as to lower overall portfolio volatility. The last two terms of Equation 3 drop out and the expression reduces to σ fully hedged = w2domestic σ2domestic + w2foreign σ2foreign + 2 wdomestic wforeign σdomestic. 2010 John Normand (44-20) 7325-5222 john. This relationship is expressed algebraically as σportfolio = w2domestic σ2domestic + w2foreign σ2foreign + 2 wdomestic wforeign σdomestic. Risk-adjusted returns on the index are decent (0. providing some offset to the losses on domestic assets.14 -0. EM FX has outperformed 6-mo dollar cash by 4% annually since the early 1990s. the variance of a portfolio combining a domestic and foreign asset will be a function of the (1) domestic asset’s variance. For the fully hedged portfolio. fx + wforeign σforeign. 1994-2009 Based on returns from J. which is comparable to that on US equities and highgrade credit. The last two terms capture the impact of currency exposure on total portfolio risk. foreign + H2 σ2fx + 2H( wdomestic σdomestic. Morgan Securities Ltd.P. Finally. assuming the dollar appreciated in response to those developments.42). Rearranging terms (subtracting equation 4 from equation 3) illustrates that currency exposure will reduce total portfolio risk if the following condition holds: w2foreign σ2fx + 2 wforeign (wdomesticσdomestic. while Latin currencies have outperformed by 1. fx + wforeign σforeign. (3) currency’s variance. multinational corporations or central banks — likewise match that exposure with unhedged foreign currency assets.61 0.02 Asia ME/Africa Europe Latam 0. such as political uncertainty and natural disasters. and (5) covariance between the assets and the currency. Chart 7: ELMI+ carry and spot returns. In the event of a natural disaster. the variance of the currency exposure will be zero (since H = 0). fx) [Equation 3] where H is the proportion of the portfolio with foreign currency exposure. Europe or Japan) unhedged. Morgan ELMI+ index 0. investors who have significant foreign currency liabilities — such as global pension funds. Morgan Chart 8: Emerging markets FX Sharpe ratios.normand@jpmorgan. local currency would likely weaken versus the G-3. many investors in emerging markets hold their foreign exposure (in the US.10 Source: J.com J.30 0.P. since 1994.70 0. and equation 4 that of the fully hedged portfolio. the same as if this were a domestic two-asset portfolio. Outperformance has been much higher for convergence countries: Central European currencies have outperformed USD libor by over 7.
For USD-based investors over the past two decades. the negative correlation can be somewhat closer to zero and still reduce portfolio volatility. fx < -wforeign σfx / 2σforeign [Equation 6] Plugging in various volatilities for the assets and currency and different allocations to foreign assets generates breakeven correlations which reduce total portfolio risk.05 in order to lower overall risk.0 -0. Morgan Chart 10: Correlation of foreign stocks and FX. implying that their currencies depreciate versus the dollar when equities rise (chart 10 and table 1). In this case. if the currency and asset returns are equally volatile (vol ratio = 1. as in chart 9) and 10% of the portfolio is allocated to foreign assets. Morgan Securities Ltd.2 -0. then the breakeven correlations would need to be even more negative for a given foreign allocation in order to reduce portfolio volatility. If 50% of exposure were foreign.10 -0. the covariance of currency returns with asset returns must be sufficiently negative to overcome the volatility of the currency itself.0 vol ratio = 0.normand@jpmorgan. • A positive correlation between the assets and the currency increases total portfolio risk. Morgan 8 .com J. the currency’s correlation with the domestic asset must be more negative in order to reduce total portfolio volatility.2 vol ratio = 1.60 -0.2 -0. • If sufficiently large.25.6 0. How negative must the correlation be for currency exposure to lower portfolio variance? Consider an example in which the volatility of the domestic and foreign assets are equal.P.00 -0.4 -0. only European stocks have correlated negatively with their currency performance versus the dollar. Chart 9: Threshold correlation for portfolio risk reduction 0.20 -0. 1990-2009 for USDbased investor Based on monthly returns over 3 year rolling window 0. the correlation would need to be more negative than -0.P.6 90 93 96 99 02 05 08 CAD AUD JPY Source: J. If the currency were more volatile than the asset market (vol ratio = 1.40 -0. and the domestic and foreign asset correlations with the currency are equal.2). a negative correlation between assets and the currency can offset currency volatility.5 percent allocation to foreign equities Source: J. Morgan Chart 11: Correlation of foreign stocks and FX.2 0. 2010 John Normand (44-20) 7325-5222 john.6 -0. For example. • If there is zero correlation between the currency and the foreign asset. • If currency returns are less volatile than asset returns.30 -0. More specifically.P. then the correlation between currency and asset returns would need to be more negative than -0.70 100% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% vol ratio = 1.P. In practice asset/FX correlations are neither sufficiently nor consistently negative to reduce portfolio volatility by leaving currency exposure unhedged.50 -0.8 90 93 96 99 02 05 08 EUR GBP Source: J. 1990-2009 for USDbased investor Based on monthly returns over 3 year rolling window 0.2 0.4 0. Equation 3 reduces to ρ foreign.0 -0.Global FX Strategy Managing FX hedge ratios May 26.4 -0.4 0.8 0. breakeven FX/asset correlation In general.
30 -0.14 0.42 -0.29 -0.17 9 . a positive correlation of 0.09 -0.04 -0.29 0.12 -0.27 0.49 -0.04 -0.19 0.06 0.17 0. lower dollar).47 -0.17 NA NA NA NA NA NA CAD -0.02 0. Correlations between currencies.08 -0.37 -0.2009 (20 years) 2005 . For bonds.48 -0.23 -0.25 -0.14 -0.09 (20 years) 2005 .2009 (20 years) 2005 .02 0.34 0.10 0.08 0.09 (10 years) 1990 .16 -0. The correlation between the Nikkei and the yen’s performance versus the dollar has been negative.25 -0. Foreign currency Base currency USD Foreign asset bonds stocks bonds stocks bonds stocks EUR bonds stocks bonds stocks bonds stocks GBP bonds stocks bonds stocks bonds stocks AUD bonds stocks bonds stocks bonds stocks Source: J. Correlation is only half the picture.25 -0.normand@jpmorgan. For example. GBP and AUD) Correlations between asset market performance and foreign currency performance versus base currency.07 0.44 0.09 (10 years) 1990 .50 0.33 -0. but the strength of this relationship varies over time.P.04 -0.13 -0.33 -0.40 -0. But that apparent diversification benefit disappears over a ten-year sample period (correlations close to zero) and over a five-year sample (the long-term correlation has flipped to positive).12 NA NA NA NA NA NA 0.14 -0.29 -0.46 0.04 -0. even high and consistently-negative correlations are insufficient to reduce portfolio volatility.19 0.19 -0.19 -0.18 -0.27 0. which makes unhedge exposure riskier.34 0.02 -0.29 -0.05 0. unhedged exposure also subjects local investors to a translation loss.12 -0.36 0.2009 (20 years) USD NA NA NA NA NA NA 0. Morgan Securities Ltd.02 -0.03 0.44 0.52 0.32 0.24 -0.26 -0.48 0.com J.04 0. Correlations based on monthly return data.09 -0.15 -0.22 GBP -0.12 0.05 0. the dollar’s performance versus the euro has correlated negatively with US stocks over the past five years (higher stocks. but given the AUD’s long-term appreciation trend.24 0.25 EUR -0.12 -0.05 -0.12 0.33 0.Global FX Strategy Managing FX hedge ratios May 26.25 0.09 (5 years) 2000 .20 -0.28 -0.09 (5 years) 2000 .12 -0.09 (5 years) 2000 . stocks and bonds from the perspective of four based currencies (USD.19 -0. however.00 0.13 -0. which thus makes unhedged equities more volatile for US investors.09 0.03 -0. 2010 John Normand (44-20) 7325-5222 john.24 -0. EUR.49 -0. Commodity currency moves versus the euro and sterling correlate positively with their equities.04 -0.58 0.58 -0.13 JPY 0.09 (10 years) 1990 .20 0. For euro and sterling-based investors. Yen correlations are negative but sometimes not strong enough to diversify risk.00 NA NA NA NA NA NA -0. The combination of yield and commodity exposure leads AUD to rise (and foreign currencies to fall) when equities rise.16 0. which suggests diversification benefits from leaving exposure unhedged.01 -0.P.35 -0. Viewed in isolation this feature lends diversification benefit to unhedged foreign exposure. This point is illustrated by the appendix charts at the end of Section II.04 -0.14 -0.18 -0.21 -0.56 0. Correlations between Australian/Canadian equities and their currencies versus the dollar have always been positive (chart 11 and table 1). Since currency volatility is so much higher than bond market volatility. Morgan Sample period 2005 .07 -0.39 -0. but not over the past twenty years.09 -0.11 -0.04 0.09 (5 years) 2000 .36 0.19 -0.23 0.39 -0.49 0.21 -0. Table 1. many currencies correlate negatively with fixed income performance.11 -0.29 -0.18 for Euro area stocks with USD as base currency implies that the euro tends to rise versus the dollar when European stocks rise.26 -0.01 -0.39 -0.04 -0.11 0. The atypical market is Australia.09 (10 years) 1990.28 -0.26 AUD -0.
unhedged and semi-hedged benchmarks. Morgan Securities Ltd The optimal hedge ratio: no size fits all In the final balance. The opposite holds for AUD-based investors: unhedged portfolios are less volatile over the long term. FX exposure should be mostly hedged given how currency volatility often drives overall portfolio risk. • FX hedge ratios for equity portfolios are more debatable. (2) the currency allocation of foreign assets. (3) the consistency of historical volatilities and correlations in the future. 2010 John Normand (44-20) 7325-5222 john. 1988-2010 Based on monthly returns over rolling 12 month window 35% 30% 25% 20% 15% 10% 5% 0% 88 90 92 94 96 98 00 02 04 06 08 Source: J.com J. what hedge ratio makes sense strategically? There is no uniform.P. 1988-2010 Based on monthly returns over rolling 12 month window 35% 30% 25% 20% 15% 10% 5% 0% 88 90 92 94 96 98 00 02 04 06 08 Source: J. 100% hedged benchmarks entail important limitations. Even for those investors seeking to minimise volatility. The optimal hedge ratio therefore will vary by investor and over time. If the benchmark is 100% hedged against the base currency. Thus the manager can only outperform the benchmark in environments when the foreign currency is rising. Morgan unhedged hedged Chart 14: Volatility of hedged and unhedged Nikkei returns for USDbased investors.Global FX Strategy Managing FX hedge ratios May 26. The appropriate policy is particular to the investor given four variables: (1) the allocation between domestic and international assets.P. Even a 100% hedge ratio with the flexibility to deviate from the policy is quite constrained. EUR and GBP-based investors given the generally positive correlation between foreign currency performance and foreign equity markets.P. • For investors in G-10 bonds. Chart 12: Volatility of hedged and unhedged S&P500 returns for GBP-based investors. While there are no absolutes. the manager may hedge less than the benchmark but not more. The exceptions are for catastrophe insurance and asset/liability matching discussed on page 7.P. 100% hedged benchmarks are lower variance than unhedged ones for USD. Morgan 10 . several guidelines apply to the merits of fully hedged. For investors concerned mainly with minimising volatility over the long term. strategic hedge ratio. • Full hedging eliminates potential short-term gains from tactical trading. Given the profitability of active currency managers and rule-based trading strategies.normand@jpmorgan. 1988-2010 Based on monthly returns over rolling 12 month window 50% 40% 30% 20% 10% 0% 89 92 95 98 01 04 07 10 unhedged hedged Source: J. and (4) the investor’s risk preference. 100% hedging precludes a potential alpha opportunity. Morgan unhedged hedged Chart 13: Volatility of hedged and unhedged S&P500 returns for AUD-based investors. Section III discusses this issue in more detail.
Chart 15: Performance with and without overlay 12% pass-through to investor 9% 6% 3% 0% -3% -6% -9% -12% -12% -9% -6% -3% 0% 3% 6% 9% 12% currency return Source: J. particularly for those investors concerned about entry levels. Symmetric benchmarks appeal because they address the risk inherent in overseas investing. And an unhedged benchmark limits flexibility as much as a 100% one does. the manager may hedge more than the benchmark but never less. unhedged equity returns were more volatile than hedged ones. some investors found their fully-hedged benchmarks to be even more volatile than unhedged ones because they were short a foreign currency which was appreciating as the foreign equity market fell. many situations require a more dynamic approach. If they expect the base to depreciate. Thus the manager can only outperform the benchmark in an environment when the foreign currency is depreciating. Semi-hedged benchmarks (50/50) faced a similar dilemma. With symmetry. thus reinforcing the decline in equity and currency markets. they can buy additional foreign currency exposure over the benchmark. 50/50 hedge ratio as policy of least regret Given the particulars of investor or corporate exposure and the tradeoffs inherent in various policy options. risk managers have two options for selecting the optimal hedge ratio. and minimize the risk of overfitting inherent in mean-variance optimization. or US hedgers who were short the yen as an overlay to their Nikkei investments. although to a lesser degree Such would have been the predicament of UK and Australian investors who hedged the currency risk on S&P500 exposure (so were short USD vs GBP and AUD as both currencies collapsed). Ironically in 2008. Morgan Securities Ltd • Full hedging can also impose substantial cash flow requirements during period of extreme market stress.normand@jpmorgan.P. Alternatively they can adopt a symmetric benchmark hedge ratio of 50%. In some cases investors were forced to liquidate underlying assets to fund the cash flow obligations of a currency hedging program.P. If the benchmark is unhedged against the base currency. It is thus the policy of least regret for many investors. Even if the 50/50 policy appeals intuitively as a benchmark. Charts 12 – 14 highlight how. allow flexibility to manage cash flow. such as the 2008-09 credit crisis. then overlay some discretion in the final policy decision. Sections III and IV of this paper address these issues in more detail by applying fair value and momentum models. They also allow investors to capture profit opportunities.Global FX Strategy Managing FX hedge ratios May 26. In those case where a zero hedge ratio seems sensible long-term due to a negative correlation between foreign currency and equity returns – the case in foreign equities from an AUD perspective – such a policy could expose investors to substantial short-term volatility. The payoff structure resembles a zero cost call option on the currency 11 . Morgan symmetric hedge + overlay unhedged which allows the investor to participate in currency gains while avoiding currency losses (chart 15). if managers believe the base currency will appreciate. 2010 John Normand (44-20) 7325-5222 john.com J. for the first time in decades. cash flows and profit opportunities. They can run a mean-variance optimization to determine the minimum variance hedge ratio. they can hedge the foreign currency.
2010 John Normand (44-20) 7325-5222 john. % 20 15 10 5 10 15 20 25 30 Risk. unhedged 10 15 Risk. Morgan Chart 3: Bond returns hedged vs unhedged into EUR. 1988 – 2009 Return.Global FX Strategy Managing FX hedge ratios May 26. Morgan JPY GBP USD AUD CAD MSCI ex-EU 0% EUR JPY GBP USD AUD CAD GBI Source: J. unhedged Chart 6: Bonds for EUR-based investors: return vs risk. % 20 15 Euro area equities 10 Euro area bonds 5 World ex Euro area. % World ex Euro area. 1993-2009 15% local ccy unhedged hedged 20% 10% 10% 5% 0% EUR Source: J.P.P.normand@jpmorgan.P. 1989-2009 12% 9% 6% 3% 0% -3% -6% EUR Source: J. 1989-2009 40% local ccy 30% unhedged hedged Chart 4: Bond volatility hedged vs unhedged into EUR. Morgan 12 . Morgan Chart 3:Equities for EUR-based investors: return vs risk.com J.2009 Return.P. Morgan Securities Ltd Appendix 1: Stock and bond returns and volatility with EUR as base currency Chart 1: Equity returns hedged vs unhedged into EUR. 1993-2009 9% local ccy unhedged hedged local ccy unhedged hedged 6% 3% JPY GBP USD AUD CAD MSCI ex-EU 0% EUR Source: J.P.P. hedged 5 World ex Euro area. hedged World ex Euro area. % Source: J. 1993 . Morgan JPY GBP USD AUD CAD GBI Chart 2: Equity volatility hedged vs unhedged into EUR.
1993-2009 18% local ccy 15% 12% 9% 6% 3% unhedged hedged unhedged hedged JPY EUR USD AUD CAD MSCI ex-UK 0% GBP Source: J. unhedged World ex UK. hedged 5 5 World ex UK.com J.Global FX Strategy Managing FX hedge ratios May 26. 1988-2009 12% local ccy 9% 6% 3% 0% -3% GBP Source: J. Morgan Source: J. 1988-2009 30% 25% 20% local ccy unhedged hedged Chart 5: Bond volatility hedged vs unhedged into GBP. hedged 10 15 Risk Source: J. 2010 John Normand (44-20) 7325-5222 john. Morgan 13 . 1993 .P.P.P.normand@jpmorgan.P. Morgan Chart 4: Bonds hedged vs unhedged into GBP.P. Morgan JPY EUR USD AUD CAD MSCI exUK JPY EUR USD AUD CAD GBI ex UK Chart 3: Equities for GBP-based investors: return vs risk.P. Morgan 10% 5% 0% GBP Source: J. unhedged UK equities World ex UK. Morgan Securities Ltd Appendix 2: Stock and bond returns and volatility with GBP as base currency Chart 1: Equity returns hedged vs unhedged into GBP. 1988 . 1993-2009 25% 20% 15% local ccy unhedged hedged 15% 10% 5% 0% GBP Source: J.2009 Return 25 20 Chart 6: Bonds for GBP-based investors: return vs risk.2009 Return 20 15 15 10 5 15 20 25 Risk 10 UK bonds World ex UK. Morgan JPY EUR USD AUD CAD GBI ex UK Chart 2: Equity volatility hedged vs unhedged into GBP.P.
1993-2009 25% local ccy 20% unhedged hedged 20% 15% 10% 5% 10% 0% AUD Source: J.2009 Return 10 8 6 World. Morgan 14 . 1988 .Global FX Strategy Managing FX hedge ratios May 26. Morgan Securities Ltd Appendix 3: Stock and bond returns and volatility with AUD as base currency Chart 1: Equity returns hedged vs unhedged into AUD. unhedged Australia bonds Australia equities World.P. hedged World. 2010 John Normand (44-20) 7325-5222 john. Morgan Chart 4: Bonds hedged vs unhedged into AUD.P.com J. Morgan JPY EUR USD GBP CAD GBI ex AUD Chart 3: Equities for AUD-based investors: return vs risk. 1988 .P.P. Morgan 0% JPY EUR USD GBP CAD MSCI world AUD Source: J. unhedged - 4 2 - 5 10 15 20 Risk Source: J.P. hedged World.normand@jpmorgan.P. 1988-2009 12% 9% 6% 3% 0% -3% AUD Source: J. Morgan JPY EUR USD GBP CAD GBI ex AUD Chart 2: Equity volatility hedged vs unhedged into AUD.P.2009 Return 20 15 10 5 15 20 25 Risk Chart 6: Bonds for AUD-based investors: return vs risk. 1993-2009 9% local ccy unhedged hedged local ccy unhedged hedged 6% 3% JPY EUR USD GBP CAD MSCI world 0% AUD Source: J. 1988-2009 30% local ccy unhedged hedged Chart 5: Bond volatility hedged vs unhedged into AUD. Morgan Source: J.
Morgan Chart 2: Emerging market bond volatility. based on J.P. Morgan 15 . 2010 John Normand (44-20) 7325-5222 john.P. based on J.normand@jpmorgan. Morgan EM-GBI 20% local ccy 15% 10% 5% 0% EM-GBI GBI-EM Europe GBI-EM Asia GBI-EM Latam GBI-EM Mideast & Africa unhedged hedged Source: J. USD terms.com J. Morgan Securities Ltd Appendix 4: Emerging market bond returns and volatility with USD as base currency Chart 1: Emerging market bond returns.P. 2002-2009 Annual %.P. USD terms.P. Morgan EM-GBI 30% 25% 20% 15% 10% 5% 0% EM-GBI GBI-EM Europe GBI-EM Asia GBI-EM Latam GBI-EM Mideast & Africa local ccy unhedged hedged Source: J.Global FX Strategy Managing FX hedge ratios May 26. 2002-2009 Annual %.
driving up real interest rates and thereby attracting foreign capital. Using fair value models to focus strategic hedge ratios4 • 100% hedging can create onerous cash flow obligations. J. In short. Morgan’s fair value model (April 24. Rising demand tightens capacity constraints. Current account balance and external debt: We interpret a current account deficit as a capital account surplus. and zero hedging leaves investors and corporates too exposed to exchange rate swings. FX misalignments are statistically and economically significant in predicting currency moves out of sample over 6-. together with trade weights and historical bilateral nominal exchange rates are used to calculate mutually consistent fair-value bilateral nominal exchange rates. not cheap ones. Thus. reflecting the attractiveness of domestic assets in the global marketplace and the fact that capital can cross borders instantaneously. Starting from the presumption that real trade-weighted exchange rates vary systematically with long-run economic fundamentals. An alternative strategy would focus hedging on the most misaligned currencies: hedge expensive pairs. the model defines the relationship between real trade-weighted exchange rates and their fundamental drivers as the long-run equilibrium (or sustainable) real exchange rates (ERER). 2010 Gabriel de Kock (1-212) 834-4254 gabriel. Determining which currencies qualify as cheap or expensive is a valuation exercise.Global FX Strategy Managing FX hedge ratios May 26. and a 50% hedge ratio strikes some as arbitrary. what are the alternatives? From a strategic perspective. By implication. Morgan’s fair long-term fair value model can inform this process. exchange rates. because • • If 100% hedging can create cash flow problems in highvolatility environments. This section summarizes J. the ERER impact of the terms of trade is proportional to the openness of the economy and the duration of the terms of trade change. defining fair values in terms of trade-weighted exchange rates. a rising current account deficit is associated with exchange rate appreciation. so should be fully or mostly hedged over a medium-term horizon such as one year.P. a country cannot run a current account deficit – spending more than it earns – forever. 16 .P.P. Government debt: High levels of government debt depreciate the currency over the long run. Estimated ERER. 0% hedging leaves investors and corporates too exposed to exchange rate swings. while trade and investment flows change slowly over time. • • • Long-term FX fundamentals The fundamental drivers of long-term equilibrium real exchange rates in JP Morgan's fair-value model are those that economic theory suggests should affect a wide range of currencies over a wide range of time periods and circumstances. However. and the model’s performance in guiding currency investment and hedging decisions. The model derives fair value estimates for G-10 currency cross rates from the long-term relationship between real trade-weighted exchange rates and fundamental economic drivers.kock@jpmorgan. so should be left unhedged or partially hedged over the medium term. 12. for which we employ a fair value framework developed in 2008 (see footnote).de. 2009). It takes a multicurrency approach to fair value exchange rates.s. Persistent external deficits imply an accumulation of external debt and rising debt service payments to foreign investors. investors and corporates could focus hedging activity on those currencies which are most misaligned. As a result. The fair-value exchange rate model is designed as an objective method for estimating exchange rate benchmarks that would be sustainable over longer time horizons between one and a half and five years. attracting capital inflows and boosting capex and consumer demand.com JPMorgan Chase Bank NA III. Very cheap currencies would be biased to appreciate over time. A back test over the 2004-08 period shows that trading rules exploiting misalignments consistently outperform carry. both by Gabriel de Kock. Very expensive currencies would seem most vulnerable to a fall. 2008) and Hedging and trading long-term FX with J. It explains our approach to modelling fair values.and 18-month horizons. These variables and their expected impacts on the ERER are: Terms of Trade: A rise in the terms of trade (export prices relative to import prices) increases profits and incomes and hence national wealth. Morgan’s fair-value model for the G-10 4 This section updates and extends to other base currencies the fair value model outlined in two previous publications: A new fair value model for G10 currencies (September 6. the drivers of fair value exchange rates and their empirical importance. adjusting medium-term hedge ratios in response to misalignments should improve portfolio performance or reduce hedging costs. higher foreign liabilities require a weaker currency to generate the increase in export earning to service the debt.
s.9869 -0. Next. led by NOK. SEK. By contrast.6427 (-1.2331 (-12. The key difference between the two currencies is that as a commodity price-driven swings in Australia’s terms of trade account for much of the swings in the real AUD and hence for much of the swings in its fair value. 1980-1Q08 Driver Estimate Permanent Log(RER) Log(Terms of Trade) x Openness Current Account Balance (% GDP) Net Inv estment Income (% GDP) Gov ernment Debt (% GDP) Inflation (%) 1.” The elasticities measuring the equilibrium real exchange rate impact of shift R = 0.Global FX Strategy Managing FX hedge ratios May 26.92) 0. we use a panel cointegration analysis to estimate the stable long-term relationship between the real exchange rate and its fundamental drivers. since price levels at each point in history are given. simultaneously.3429 (26. as of March. This step combines data from all of the G-10 economies and by assuming that the fundamental drivers affect real exchange rates similarly across countries. while most of the swings in the AUD real exchange rate are attributed to swings in its fair value. However. Table 2 highlights that. to the extent that real observed real exchange rate changes predict future shifts in the permanent component of the macroeconomic drivers. given that the US is a relatively closed economy have only a small impact on the real exchange rate.86) -1. we use the trade weights to distribute nominal misalignments across the G-10 to match.09) 0.3501 -0. Our latest fair value estimates for G-10 real trade-weighted exchange rates and bilateral exchange rates vs. deviations from fair value had declined notably since 2009 Q3. while GBP was about 15% undervalued and the JPY 11% overvalued. Intuitively. Table 1: Estimated long-run equilibrium real exchange rate model. To compensate. the USD are shown in Tables 2 and 3. And. Source: J.9550 0. 2010 Gabriel de Kock (1-212) 834-4254 gabriel. we use the our estimated long-run equilibrium real exchange rates obtained from this decomposition.4755 (2. to calculate fair-value bilateral nominal exchange rates. CAD. Finally. Charts 1 and 2 illustrate historical real equilibrium exchange rate estimates for the USD and the AUD.P.4732 Estimating fair values We estimate the fair-value exchange rates in three steps. reflecting the forward-looking nature of financial markets. Table 1 shows the equation used to estimate the G10 equilibrium real exchange rates. period-to-period swings of the drivers contain transitory noise that distorts the estimates when plugged into a relationship designed to capture long-term shifts. the estimated real fair value of the USD appears quite stable over time compared to the swings in the market real exchange rate. and EUR. the US with its well-diversified import and export baskets experiences mild terms of trade swings over time and those swings. Inflation: High inflation typically reflects bad economic policies and is associated with high inflation volatility. allows us to estimate a long-term relationship with relatively short data spans. the real misalignments of all of the G10 countries.88 Note: Standard errors in parentheses. the USD and EUR were close to fairly valued in real-trade-weighted terms. lowering potential economic growth and longterm asset returns and raises expected future tax rates and inflation rates. we use the parameters of the estimated long-term relationship and observations on the fundamental drivers to find the equilibrium or fair-value real exchange rates. in the economic drivers are of the expected sign and magnitude. we use time-series techniques to decompose the data into permanent and temporary components.2637 0.1716 -0. To determine the appropriate scale factor. First. all of the real misalignment can be attributed to nominal misalignments. Strikingly. the analysis attributes about onequarter of the swings in observed real exchange rate to shift in the underlying fair value.com JPMorgan Chase Bank NA government borrowing crowds out private borrowing and investment. both of which distort resource allocation lowering the return on capita and the ERER.27) -0. So.2971 (-10. Morgan 2 17 .02) -0. we incorporate a fraction of period-to period changes in the observed real exchange in the estimated long-run equilibrium real exchange rate.kock@jpmorgan. with the scaled-down coefficients denoted as “permanent.de. both of which tend to undercut long-term asset market performance. we scale down the elasticities on the fundamental drivers in proportion to the relative importance of transitory noise in their period to period fluctuations. along with trade weights and historical nominal exchange rates. However. on average. Moreover.
9 108. while GBP was more than 20% undervalued against AUD. CAD and NZD.5 109.Global FX Strategy Managing FX hedge ratios May 26.2 111.3 102.2 103.4 102.4 119.2 110.2 118. 85 90 95 00 05 10 Chart 2: Equilibrium and actual real trade-weighted AUD.4 112.com JPMorgan Chase Bank NA The EUR.and 18-month horizons.6 100. (b) go long FX.3 112.2%.1 118.7 102.6 106. CAD and AUD were less than 3% mispriced against the USD in 1Q10.9 96.2 105. typically performing best over 6-month investment horizons. Morgan Fair Value 85 90 95 00 05 10 Table 2: G-10 actual and fair-value real trade-weighted exchange rate indexes (2Q09=100). buying 1 unit of the base currency forward if the percentage overvaluation of the foreign currency exceeds its forward discount (and hence the cost of carry) by more than a predetermined critical or threshold value. particularly for GBP and JPY. Chart 1: Equilibrium and actual real trade-weighted USD.P.7 FV 96.9 109.P.6 4Q09 Actual 93. Misalignments relative to fair value predict exchange-rate changes out of sample.2 107. particularly over 12-18-month horizons.3 92.1 111. the misalignments are much more extreme for some G-10 crosses. CHF.s.7 FV 96. 12.2 107. as Table 4 shows.3 3Q09 Actual 96. 2Q09-1Q108 2Q09 Actual USD EUR JPY GBP CHF CAD AUD NZD NOK SEK 100 100 100 100 100 100 100 100 100 100 FV 99.1 102.1 101.P.2 102. Of course the usual caveat applies to these results: they reflect FX developments in a particular time period.5 113. while GBP.5 104.8 104.8 92.9 103. For example. But.6 92.1 106. 18 . 2010 Gabriel de Kock (1-212) 834-4254 gabriel.9 114.3 108. we compare its misalignment with its forward premium or discount as follows: (a) go short foreign currency.4 97. suggesting that using deviations from fair value into account in hedging decisions could improve portfolio performance.2 92.9 100. Reverse the position at the spot rate when the forward contract matures. A back test of fair value signals To assess whether the fair-value model provides useful trading/hedging signals.0 118. the JP Morgan fair value model identifies significant misalignments among the G-10.3 102. Below we demonstrate that a simple buy-andhold trading rule – acting when misalignments exceeded the cost of carry – generally outperformed the forwards over 6-. Similarly.3 Note: Exchange rates are quarterly averages.5 104.9 106.3 101.4 1Q10 Actual 94.4 108. Source: J.6 96.5 120. Morgan Actual Fair Value Currency misalignments as signals for hedging/investment decisions In most quarters.3 109. Morgan.0 109.0 111. we back-test a simple buy-and-hold trading rule.6 103. At the most recent update of the fair value estimates in mid-March 2010 the average misalignment against the USD stood at 5.2 116.9 101.2 101.0 109. selling $1 forward if the percentage undervaluation of foreign currency exceeds its forward premium by more than the predetermined threshold and reverse the position at the spot rate when the forward contract matures.2 119. 15% against SEK and 19% against NOK. 1980-1Q10 105 Actual 95 85 75 65 55 80 Source: J. For a signal to go long or short FX against a base currency.7 111.4 98.4 106.3 120.0 FV 98. NOK and SEK remained cheap and JPY the most persistently overvalued G10 currency.de. and cannot prove that using the fairvalue model in this manner would necessarily produce profits in the future.7 110.3 97.1 102.3 101.8 95.kock@jpmorgan. 1980-1Q10 160 150 140 130 120 110 100 80 Source: J. JPY was 13% overvalued against CHF.
firstname.lastname@example.org 0.64 1.50 Note: Exchange rates are quarterly averages. A 10% misalignment threshold might seem large.5 GBP -18.35 6.de.79 0.7 -5. EUR-.60 5.06 0. 19 .03 0.9 AUD 4.85 1.4 -17. 2Q09-1Q10 2Q09 Actual EUR/USD USD/JPY USD/CHF 1. To accurately reflect the time lags in the availability of the data used in estimating fair values. for higher threshold values.41 100 1.8 SEK -5. that is. in our sample.0 CHF -2. acting on initial carry-adjusted misalignments greater than 10% outperforms carry by about 3. the strategy is more profitable for greater initial misalignments.0 -1.47 6. for ease of interpretation and to explore the robustness of the results we report back-tests for USD-.19 3Q09 Actual 1. the profitability of the buy-and-hold strategy deteriorated over longer holding periods.84 7.s.0 4.6 -15.3 13. Table 3: G-10 actual and fair-value bilateral exchange rates vd USD. In Table 4.86 1.67 6.57 1.17 AUD/USD 0.4 27.Global FX Strategy Managing FX hedge ratios May 26.P.03 1.9 0.01 FV 1.7 -19.05 0.6 12.0 12.04 0.73 5.28 FV 1.and 18-month horizons (two.91 USD/NOK 6.9 -10.69 7.8 -5. Indeed. But for the G10 currencies 19 out of 45 crosses were misaligned by more than 10%. we measure misalignments relative to the quarterly fair value estimate lagged one quarter.65 5.6 4.83 0.0 -6.55 USD/CAD 1. on average.P.39 90 1.36 97 1.86 0.06 1. A negative number denotes undervaluation of the column currency vs the row currency. And to mimic a realistic trading environment we assume that transactions occur in the first trading week of every quarter and at the averages of the forward and spot rates over the first five trading days of the quarter. As expected.com JPMorgan Chase Bank NA The objective of the back test is to determine whether there are values of the threshold for which this trading strategy would have yielded profits in excess of carry. about one-and-a-half times the average return for a 0% threshold. 12-18 Mar 10 EUR USD EUR JPY GBP CHF CAD AUD NZD NOK Misalignments measured as average market rates for 12-18 Mar vs 1Q10 fair values.41 101 1.10 0.76 NZD/USD USD/SEK 0. Finally. only the JPY and GBP were misaligned by more than 10% against the USD in mid-March.0 7.75 GBP/USD 1. JPY 10.8 -5.60 7.2 7.10 0.0 3. Source: J.7 -2.8 23.90 0.91 0.3 15.6% over a 6-month horizon. The strategy outperformed carry.07 1.2 . 2010 Gabriel de Kock (1-212) 834-4254 gabriel.8 -15.4 20. Morgan.48 90 1.05 0.71 5.5 2. And the signal is derived by comparing the average forward rates over the first five trading days of the quarter to the quarterly fair-value estimate. irrespective of the decision threshold.9 -3.5 7. 5% and 10%.47 7. GBP-. 12.39 103 1.03 1. and AUD-based investors and corporates.5 NZD 9.0 5.19 FV 1.36 105 1.7 CAD 2.8 -13. USD-based investors and corporates Table 5 shows the results of the strategy for a USD-based investor or corporate over 6-.65 5.61 1Q10 Actual 1. Table 4: G-10 bilateral exchange rate deviations from 1Q10 fair values (%). Finally.9 -9.5 -14.83 1.5 NOK -8. Source: J.61 5.02 1. The back-test covers the period from 1Q04 to 4Q08 and is based on 21 sequential estimates of the fair value model on data samples from 1Q80 to 4Q03 through 1Q80 to 4Q08.4 13.05 1.06 1.84 1.81 4Q09 Actual 1. four and six quarters) and decision threshold values at 0%.2 -8.2 -1.31 6.2 9.43 94 1.2 -13. performing best over 6-month investment horizons irrespective of the size of the decision threshold.81 0.63 1.5 -29.11 7.11 FV 1. Morgan.
5% 3.9% 0.3% SEK 6. 2.3% -0.5% CHF 0.P.2% -0.4% NZD 7. 2010 Gabriel de Kock (1-212) 834-4254 gabriel.8% NOK -0.0% 1.s.1% GBP -3.1% 0.9% EUR -0.5% 0.4% 3.3% 0.5% 0.1% NA NZD -2.8% CAD -0. but the excess returns are modest – about one-fifth on average the excess returns enjoyed by USD-based decision-makers.6% 0.1% NZD 1.1% JPY 2.5% -0.5% 2.profitability of fair value-based trading rules (% annual. 2.1% 0. 0. As for USD-based investors.3% 2.4% -1.0% 0.6% 0. 0. positions taken on the basis of fair value model signals generate excess returns over carry.4% 0.9% 1.5% 0.5% -0.4% -2.1% -1.4% AUD 4.4% 1.1% 0.7% 0.5% 0.1% SEK -2.8% CAD NA NA NA EUR -0. but would have done so if we had extended the sample to allow more positions to mature.com JPMorgan Chase Bank NA Table 5: USD as base currency .3% CAD NA 0.profitability of fair value-based trading rules (% annual.8% Threshold = 10% JPY -0.P.2% NA CHF 1.9% CHF 1.2% Avg. net of carry). the G-10 fair value model is the least successful in generating excess returns for EUR-based investors or corporate hedgers.4% -2.1% 0. 2004-08 Threshold = 0% Horizon 2 4 6 Horizon 2 4 6 Horizon 2 4 6 Avg.3% 0.6% 1.5% 0.1% NZD 5.4% -2.4% -2.3% 0. Table 6: EUR as base currency .5% -0.3% NOK 2.3% GBP 2.1% 3.5% -0.4% -2.1% -1.1% NOK 3.3% 1.6% CHF 0.2% 0.8% 2.2% 0.1% JPY 2.7% SEK 2.6% 2.0% -0.9% 0. Consistent with the patterns observed for USD-based investors.8% -0.6% 1.0% CHF 3.9% -0.1% 0. Morgan EUR-based investors and corporates Among the four base currencies considered here.8% SEK 4.7% 0.0% -2.1% AUD -0.2% 2.7% 2.9% 0.2% 0.1% -1.2% 1. 3.1% 1.3% NZD 4.3% 0.6% 0.7% Threshold = 5% JPY 2.2% AUD 0.4% -1.4% -1.5% -1.9% 3.5% 2.4% 0.4% -0.5% AUD NA 0. irrespective of the decision horizon or the reaction threshold.0% 0.1% Note: NA reflects situations when there were no trades for a particular currency and time horizon.0% 2.6% -0.2% 0.8% AUD 2.7% CAD 2.5% 0.9% 1.6% NZD 3. Source: J.9% 2.2% Avg. EUR/USD positions always underperformed interest rate spreads. And more clearly than elsewhere the signals are more valuable at higher decision thresholds.9% 0.8% GBP -0.7% 0.0% Threshold = 5% JPY 1. The euro trended above and away from its fair value for most of our back-test sample – the powerful trend reversal in the second half of 2008 simply came too late to affect the overall results.3% GBP 6.0% -0.1% 0.6% -0.0% 0.9% 1.1% -1. albeit by a relatively small return margin.7% NA NA NOK 0. net of carry).email@example.com% Threshold = 10% JPY 5.6% 0.5% NA USD -0.1% -1. Source: J.5% -0.9% 1.1% -0. This outcome appears to reflect the specifics of the sample period used.Global FX Strategy Managing FX hedge ratios May 26. 0. Morgan 20 .4% 1.7% NA CAD NA 1.3% GBP 6.de.7% SEK NA NA NA CHF 0.5% -0.1% -1.0% -0.3% Avg.4% 1.5% NA GBP -1.0% 0.5% AUD 2.7% NA CAD NA 0.0% SEK -2.6% NOK 3.9% USD 2.4% 0.4% -1.9% 0.6% Note: NA reflects situations when there were no trades for a particular currency and time horizon.4% 1.4% 1.9% USD -0.7% NOK -1.9% EUR 2.2% 2.5% 2. 2004-08 Threshold = 0% Horizon 2 4 6 Horizon 2 4 6 Horizon 2 4 6 Avg.3% Avg.3% 1.
0% 2. Morgan NOK -0.1% Threshold = 5% NOK -0.8% NZD 6.0% Threshold = 10% NOK 0.7% 2.8% GBP 4.9% 2.3% 9.1% -0.4% -0.7% CHF 2.5% NA NA USD 4.1% US 6.3% USD 2.1% 0.0% NOK -0.7% AUD 0.5% -1.0% 7.5% 0.4% 0.3% NZD 0.9% CHF 6.6% EUR -0. the decisions based on a 10% action threshold resulted in a very small number of positions and. before late 2008. therefore.0% 1.2% 4.4% 3.4% CAD 3.9% 1.6% 6.6% -0.0% SEK 2.9% 3.3% 1.2% 5.com JPMorgan Chase Bank NA GBP-based investors and corporates The results of the misalignment-driven buy-and hold strategy for a GBP-based decision-maker are shown in Table 7.4% 1.2% CAD 6.3% 1. the performance of the model is quite consistent over decision thresholds and investment horizons.0% 3. Morgan NZD 2.1% 5.5% 7.7% 1. on average. the deterioration of trading strategy’s performance at longer horizons is relatively small.5% EUR NA 0.2% 2.de.5% annualized.4% 4.5% 1. 2010 Gabriel de Kock (1-212) 834-4254 gabriel.4% 0.5% NA NA CAD 4.3% 0.5% Avg.1% 1.P. GBP misalignments tended to be modest.9% 0.1% Source: J.7% 0.9% Threshold = 5% JPY 2.1% 1.3% 0.9% -0.1% CAD 1.1% 0.2% 2.8% NZD 1.4% 0.kock@jpmorgan.P.1% 1.8% 3.1% 1.1% 0.4% Note: NA reflects situations when there were no trades for a particular currency and time horizon.4% 0.2% 0.8% JPY 6.9% 1.7% EUR -0.5% 3. 21 .7% EUR -1.9% 0.7% Avg.5% 1.1% 7.3% 2.7% GBP 0.6% 1.1% CHF 2.4% -1. over a 6-month investment horizon. 2. Table 7: GBP as base currency . 2004-08 Threshold = 0% Horizon 2 4 6 Horizon 2 4 6 Horizon 2 4 6 Avg.2% NA NA NZD 0.5% 2.9% NA NOK 0. 3. Perhaps counter-intuitively and unlike the case of the USDbased decisions.4% -0.9% 0.9% CAD 5.9% NOK -3.6% 2.2% NA NA NZD NA NA NA SEK NA NA NA CHF 1. 1.9% 0.3% -0. higher action thresholds did not improve portfolio performance. But.5% 3.profitability of fair value-based trading rules (% annual.8% JPY 6.5% NA 1.s.6% 0. net of carry).9% Avg.3% 4. the result are potentially less reliable. Also.5% 3.9% SEK 3.9% 0.5% NA 1.5% USD 2.4% SEK 5. As a result.3% 1.1% SEK 2. In general.2% EUR 0.6% 3. In comparison with the results using the USD as the base currency.4% 3.5% 11.5% 3. 2. 2.2% 2. Table 8 shows that the signals are most reliable for a 6-month investment horizon.9% -0.2% 1.0% Threshold = 10% JPY 2.1% 3. an investment strategy based on the fair value signals outperformed carry by more than 2% annually on average and the deterioration in performance over longer decision horizons appears quite modest.1% -1.4% 6.2% 2.5% 1. waiting for large initial misalignments does not consistently improve performance.4% 5.0% 1.5% EUR -3.3% AUD 0.4% 2. by the same token. as in the cases of decisions where the USD and GBP are the base currencies.profitability of fair value-based trading rules (% annual.3% 0.1% CHF 2.4% US 6.6% CHF 8. Table 8: AUD as base currency .4% 3.7% -2.4% -0. 2004-08 Threshold = 0% Horizon 2 4 6 Horizon 2 4 6 Horizon 2 4 6 Avg.7% Source: J. And similar to the results for GBP-based investors and corporate hedgers.2% 1.3% JPY 8.9% NA CAD 1. AUD-based investors and corporates The fair-value model also proved a reliable guide for AUDbased decision-makers.1% 0. 2.4% 10. net of carry).7% US 2.5% 2.7% NA NA JPY 2.7% AUD 4.1% 0.0% NA NA SEK 0.7% GBP 0.Global FX Strategy Managing FX hedge ratios May 26.4% -2.4% 2.7% 2.3% 0.5% Avg.2% -1.0% -0.3% Note: NA reflects situations when there were no trades for a particular currency and time horizon. The strategy outperformed carry by over 2.4% -1.
thus leading prices to overshoot fundamental value. . 6 5 This section amplifies models originally detailed in Alternatives to Standard Carry and Momentum in FX. are employed extensively in currency overlay and global macro hedge fund communities to replicate the decision making process of investors. We use the term alpha strategy since it conveys the underlying objective of outperforming a benchmark. but given its high volatility (12%). A price momentum model which adjusts dynamically around a 50/50 hedge ratio outperforms the benchmark by about 100bp annually. is forward carry.P. This strategy has returned 4. at roughly 3. its risk adjusted returns are the lowest of all alpha strategies we track (table 1). Accordingly. which are more appropriate for hedge rebalancing. it simply buys (sells) the best (worst) performing pairs.6 In FX.7% over the past ten years (table 1). such as systematic. Over the past decade J.com JPMorgan Chase Bank NA IV.d. would buy currencies in whose favor interest rate expectations are moving. the profits from such a strategy have been decent.2 to 0. either because they await fuller information to make a decision. in the process producing short-term trends. tactical. Morgan • Momentum only (9 USD pairs) • • 02 04 06 08 10 Alpha strategy basics Short-term trading models. Standard industry models generate signals for two to four week horizons. and perform poorly in range bound environments or those exhibiting frequent reversals. Despite the tendency to dismiss these frameworks as overly simplistic.morganmarkets. Price momentum is the most basic trading strategy. Normand and Ghia. usually due to behavioural biases such as underreaction and overreaction. Hence the poor performance of a basic momentum rule over the past year in which the dollar has experienced several sharp reversals (chart 1). Trading rule uses 1-mo lookback for price momentum and daily rebalancing 180 170 160 150 140 130 120 110 100 90 00 Source: J.franklin-lyons@jpmorgan. or forward carry. the standard momentum model trades in the direction of previous spot movements. (2008).P. 22 Underreaction reflects investors’ inability or unwillingness to adjust views and positions quickly. intuitive trading rules to replicate common investment strategies across all major asset classes. depending on the base currency. The approaches go by several names. some extrapolate this positive news into the future. Strategies are robust to various sample periods and model specifications. A rate momentum (forward carry) model generates comparable outperformance but with more consistency across sample periods. building on two momentum models commonly used by currency overlay managers and global macro funds. or lookback rule. Overreaction is also based on cognitive biases. The very low frequency of signal changes – a carry model would almost always be long AUD and short JPY – also renders it in appropriate for dynamic hedging. An alternative to standard carry. as determined by a filter.Global FX Strategy Managing FX hedge ratios May 26.com. Although most investors adjust their expectations fully. Morgan has developed a suite of simple. Chart 1: Returns on a simple price momentum rule since 2000 Indexed returns based on trading G-10 currencies versus USD.5. Information ratios on the strategy range from 0. prices adjust slowly towards a market’s fundamental value. • • Carry and momentum are the most commonly followed trading strategies across asset markets and in currencies. The other standard alpha model is carry. or because they are reluctant to appear non-consensus. moving average. The series of methodology notes – entitled Investment Strategies – is archived on www.2% per annum over the past decade. or alpha strategies. Momentum strategies perform better in high-volatility environments but incur sizable drawdowns. however. A rate momentum strategy. which buys high yield currencies versus low-yield ones. Momentum strategies are based on the empirical tendency of strong-performing assets to outperform again in the future. can generate signals of one to three months. Using alpha models to adjust hedge ratios tactically5 • Dynamic hedging around a medium-term benchmark requires discretion or trading rules. short-term or alpha. which derives signals from the level of interest rate differentials. The latter is tested in this section. 2010 Matthew Franklin-Lyons (1-212) 834 4565 matthew. Minor modifications.
which is more consistent with the FX hedge rebalancing frequency of long-term investors and Chart 2: EUR/USD versus Euro – US rate expectations rate expectations based on 1-mo rates 3-mos forward 300 200 100 0 -100 -200 -300 99 Source: J. Over the past decade this strategy has returns 6% per annum on a volatility of 5% (table 1).40 1. regardless of whether it is a low or high-yielder.50 1. In the case of EUR/USD.P.10 1. A simple strategy to exploit this phenomenon would buy the currency in whose favour interest rates are moving. rate differentials as measured by the difference in Euro and US libor rates reliably signaled currency moves in the late 1990s and early 2000s. In many other periods. either because rates are falling more quickly in the high-yield country or rising more quickly in the lowyield country.60 1. Price momentum is tested since it is the simplest to construct.20 1.70 EU -US libor differential.06).30 1. the more significant driver has been the change in rate spreads: the euro rallied as the rate gap between the US and Europe narrowed. the mere existence of a US rate advantage over the yen has brought only limited spot appreciation (2000 – 02 and 2005 .90 0. In USD/JPY as well. so delivers performance with much less drawdown than most other alpha models (chart 4). 2010 Matthew Franklin-Lyons (1-212) 834 4565 matthew. bp USD/JPY 140 130 120 110 100 90 80 01 03 05 07 09 Source: J. Morgan 1. Morgan 23 . 1-mo lookback for rates and daily rebalancing 200 180 160 140 120 100 00 01 02 03 04 05 06 07 08 09 10 Forw ard Carry (9 USD pairs) Source: J. Low-yielding currencies can appreciate versus high-yielding currencies when rate spreads narrow in their favor. bp EUR/USD 1. The euro depreciated when it yielded less than the dollar (1998 – 2000) and appreciated when it yielded more (2002 – 2004).P.d. This strategy is based on the observation that currencies respond as much to spread changes as they do to spread levels. the forward carry framework proposes to position on changes in expected cash rate spreads rather than current cash rate differentials.00 0. This relationship between carry and EUR/USD has broken down since 20052008 when the euro rallied almost continuously despite a rate disadvantage to the US. even though the dollar was the higher-yielding currency (chart 2). consistent with the investment horizon of many tactical trading accounts. For this reason. In this case.80 01 03 05 07 09 Chart 3: USD/JPY versus US – Japan rate expectations rate expectations based on 1-mo rates 3-mos forward 700 600 500 400 300 200 100 0 99 US-JA libor differential. the change in rates has been more significant than the level of rates. The original models were constructed to issue signals ever two to four weeks. Price momentum and forward carry (rate momentum) are the two we modify in this section for the purpose of a dynamic hedging model. this strategy uses rate momentum as a signal for trading currencies. as was the case in 2002 – 04 and in 2007 (chart 3).com JPMorgan Chase Bank NA which derives signals from changes in interest rate differentials.P. Minor modifications to the parameters generate signals for one to three month horizons. Trading rule used 1mo rates 3mos forward. Morgan Chart 4: Forward carry performance since 2000 Based on G-10 currencies vs USD. This pattern is evident in both EUR/USD and USD/JPY in recent years (charts 2 and 3). In essence.franklin-lyons@jpmorgan. and forward carry because it has offered the best combination of performance and low volatility/drawdown. Table 1 displays the historical performance of various alpha strategies and compares to those of currency managers and global macro hedge funds.Global FX Strategy Managing FX hedge ratios May 26.
Modifying alpha strategies for dynamic hedging The alpha signals have proven to be effective tools for positioning among currencies.2% 3. and positive returns. meaning that even default decisions to ascribe to a fixed coverage level implicitly results in foreign currency exposure.35 10.d.0% 12.6% 1.0% 5.2% 12. J. The original strategy backtests revealed that shorter lookback windows tend to generate the best results for forward carry signals. the following backtests use a 50% benchmark hedging index.4% -0.P.1% 1.67 -1.59 G-10 carry (unlevered) Emerging Markets carry (IncomeEM) Price momentum Forward Carry Forward Carry (9 USD Pairs) (22 Major pairs) Barclay HFR global Currency Barclay Group Parker macro hedge Traders Index BTOP FX Blacktree CMI funds -7. currency managers and global macro funds Alpha strategies Manager performance G-10 carry (IncomeFX) 2010 YTD YTD return Std dev IR 2009 Avg annual return Std dev IR 2005 .com JPMorgan Chase Bank NA corporates. all models are backtested against four base currencies: USD.4% 0.67 -1.6% 7. Symmetric mandates allow the manager to be over or underhedged relative to the benchmark.6% 0.1% 4. but longer holding periods can also be imposed by restricting the rebalancing frequency. and (2) rebalancing frequency (daily.7% 8.0% 1.5% 0.1% 3. and in what direction to adjust coverage ratios.3% 14.7% -0. and this section incorporates the strategies within dynamic hedging systems. If systematic signals generated positive returns over time.1% 1.0% 1. As discussed in Section II. price momentum and forward carry.1% 1.01 7.4% 3.7% 5.6% 0.9% 1.7% 2.6% 2.2% 2. and anecdotal evidence confirms that this describes a fairly large cross-section of typical hedging behavior.13 20.1% 0.2009 (10 years) Avg annual return Std dev IR 9.1% 1.9% 3.7% 7. Since hedgers are unlikely to adjust their coverage at high frequencies.26 Source: J.61 8.8% 5.56 4.2% 16.0% 18.7% -0.4% 0.2% 5.69 -0.2% 1. monthly).6% 0.1% NA NA 4.9% 3.3% -0. Morgan’s original models are rebalanced daily.9% 1.2% -0.41 6. have two parameters in common which can be used to calibrate the signals to hedging behaviour: (1) the lookback period for measuring changes in price or rates.1% 0.09 -11. Various lookback windows impact returns and holding periods. then they should provide a useful guide to timing hedging decisions by signalling when. Morgan 24 .33 1.51 -1.09 7.2009 (5 years) Avg annual return Std dev IR 2000 .15 57.4% -0.6% 6. GBP and AUD.92 -7.77 7.6% 13. EUR.8% -0. Signals are evaluated relative to a 50% hedged benchmark index.0% -0.9% 0.07 1.2% 3.3% 19. Both strategies.92 3.0% -0. several adjustments to the basic strategies are necessary so as to make them relevant for medium-term hedgers.63 3. A wide range of lookback windows are tested to identify strategies with appropriate holding periods.1% 16.9% 0.8% -0.18 5.3% 19.40 3.34 -11.07 7.5% 8.3% 6. but a semi-hedged benchmark brings some advantages.41 4.P. Most select a 50% hedged benchmark for practical considerations. it is first necessary to extend the trading rules in such a way as to lengthen the holding period of the trades.4% 0.1% 0.92 7.0% 10.5% 0.51 10.2% -0. Hedging decisions result in increased or decreased long exposure to one or more currencies.0% 12.61 -0.48 1.3% 0.76 -1.11 2.61 -2. 2010 Matthew Franklin-Lyons (1-212) 834 4565 matthew.7% firstname.lastname@example.org 9.41 4.07 -5.7% -0.0% 10.0% 2.1% 6. As in Section III.2% 7.7% 2. Table 1: Long-term performance of FX alpha strategies.9% 8.Global FX Strategy Managing FX hedge ratios May 26.53 0.58 5.12 3.9% 27. Before testing the efficacy of the signals.2% 0. For this reason. no uniform hedge ratio applies to all investors or corporates.8% 31.0% 22.
The appendix tables replicate the analysis across three additional base currencies (GBP. and was profitable across 100% of the 9 major G10 pairs (table 2). GBP based backtests are more successful. Backtests are conducted using J. and longer (200 to 252-day) lookback windows generate holding periods of three to four months. These +/-25% fluctuation bands are chosen for simplicity but are consistent with anecdotally-acceptable hedge ratios. and two sample periods (past five and ten years). 200 and 252 business days. 40. 8 These base currencies were chosen to highlight performance across three economically diverse regions.4 to 0. in the case of a 100 or 40-day lookback windows. Results are generated across G-10 pairs. Table 3 displays the results for only the 40-day lookback window. The EUR strategies are the worst performing. or as indicated by a given signal. 40.com JPMorgan Chase Bank NA From the perspective of a USD-based investor. If a sell USD signal is issued. the strategy is profitable across all G-10 pairs. 75% unhedged If the model issues a buy signal on the dollar. 20 to 40-day lookback windows generate holding periods of around one month. from the standpoint of USD based investors. EUR. These results are broadly consistent with that of the USD based strategies as most lookback windows are profitable across most G10 pairs. Finally.P.franklin-lyons@jpmorgan. These cross currency averages mask fairly significant intracurrency details. across G-10 currencies. GBP based investors. Price momentum is tested on lookback windows of 20. over both time periods. 50% unhedged • Signal based strategy: Long USD signal: 75% hedged. 2010 Matthew Franklin-Lyons (1-212) 834 4565 matthew.d. the volatility of that outperformance (tracking error) and their information ratio (excess return divided by volatility of excess return). NZD. All strategies are assessed for the excess return (outperformance) versus the 50/50 benchmark. This provides an overview which helps to gauge the general performance of a given strategy across all of the currencies. The backtests focus on the two most simple strategy building blocks.5% above that of the benchmark strategy in most cases (appendix table 1). 8 The full set of results include backtests across a wide range of strategy specifications. 100 and 120 business days. sample periods and model specifications. As already indicated in table 2. • Benchmark: 50% hedged.Global FX Strategy Managing FX hedge ratios May 26. Both strategies are also tested with and without restrictions on a fixed rebalancing day at month-end. 60. it appears that price momentum tends to favor shorter lookback windows over the past decade. with average annual excess returns of around 1% for 40 to 100-day lookback windows using ten years of data (appendix table 3).8 over the past decade (appendix table 5). the hedge ratio is increased to 75% to capture the expected depreciation of the foreign currency. which are Bloomberg government bond indices. complete results are confined to the Appendix. Results: Price momentum The simple momentum results are displayed within tables 2 and 3. the AUD based results have positive excess returns. Performance tables also display averages The hedging simulation relies on the JPMorgan hedged and unhedged bond indices across G10 currencies with the exception of NOK and CHF. simple momentum and forward carry. and EUR. Please see JPMorgan Government Bond Indices. The percentage of the portfolio is chosen according to the benchmark specification (50%). Forward carry is tested using lookback windows of 20. 100. The best performances over a ten-year sample are AUD. with highly different exchange rate behavior. This strategy generates significant and consistently positive excess returns across currencies. Not surprisingly. 100-day lookbacks generate holding periods of about two months. Morgan Index Research’s hedged and unhedged bond indices7 which hedge monthly via forward contracts and rebalance at the close of the first business day of the month. Francis Diamond (2002). with the best performances for shorter lookback windows where average annual returns are 1-3% above that of the benchmark index. the strategy generated an average excess performance of around 2% over 10 years resulting in an average information ratio of around 0. with average annual performance of only 0. In the aggregate. For example. alpha models are used to adjust hedge ratios as follows. For example. CAD. resulting in information ratios of around 0. and AUD based investors hedging G-10 pairs back to their respective home currencies in each case. 7 across all of the individual currency results for a given strategy. Given these permutations. the hedge ratio is reduced to 25% to profit from expected foreign currency strength. 25 . 25% unhedged • Short USD signal: 25% hedged. and AUD) and the full range of lookback windows. the average holding periods for the strategies are dependent on the lookback window chosen.7.
2. and was profitably over all currencies over both periods (table 4). Table entries are averages of results across all G-10 USD pairs 252 Day 1.43 163 1 2.32 192 1 0.8% 3.43 196 1 1.9% 0. NZD. day lookback window USD/CAD 2000 .5% 0.2% 0.7% 1.12 44% 52 4 200 Day 1. sample periods and model specifications.7% 3. 40 bus. For example. This strategy performs well across most currencies.6% 0. the strategy generated an average excess performance of 1.87 100% 97 2 40 Day 2. The results are consistent with that of the USD-based strategies. As before. All of these results appear well suited for dynamic hedging where investors or corporates rebalance on a one to three month horizon.2009 (10 years) Avg annual return Std dev IR 2005 .0% 3. Morgan Table 3: Individual currency excess returns of price momentum rule-based index versus benchmark index: 9 USD pairs Excess returns over the 50/50 benchmark.1% 2.4% 2% 1. CAD and GBP. with average annual excess returns of 0.1% 2.7% 0.0% 0. Forward carry tends to favor 60 to 100-day lookback windows.P.3% 3.2% 2.2% 3.2009 (5 years) Avg annual return Std dev IR # of trades* average months per trade* *Using 10y or maximum available history 2.42 140 1 1.83 NA NA NA 1. As before.7% 3.49 2. delivering excess returns of 1-2% per annum.3% 4. The appendix also presents results for the other base currencies of EUR. The EUR results are good.25 170 1 1.8% 0.8% (appendix table 2). and has been best for AUD.2009 (5 years) Avg annual return Std dev IR %ccy with profitable strategies # of trades* average months per trade* *Using 10y or maximum available history Source: J.73 100% 40 Day 1.4 over 26 .7% 2.6% to 0.2% 0.28 189 1 1.d.96 USD/EUR USD/NOK USD/GBP USD/SEK USD/CHF USD/AUD USD/JPY USD/NZD Source: J. the average holding period is longer for longer lookback windows.7% 3. For medium-term lookbacks (40 to 100-day).10 140 1 2.57 86% 200 Day 0. Table 5 displays the individual currency results for a 60-day lookback.36 3.2% 2% 0.0% 3.9% 6.0% 0. 2010 Matthew Franklin-Lyons (1-212) 834 4565 matthew.9.2% 3. information ratios ranged from 0. the strategies generate decent excess returns across currencies.60 180 1 2.1% 2% 0. GBP and AUD.84 89% 256 1 2000 .2% 3% 0.51 100% 172 1 20 Day 2.9% 2.62 NA NA NA 3. and as long as three months for the longest lookback (120-day).2% over five years.Global FX Strategy Managing FX hedge ratios May 26.3 to 0.2009 (10 years) Avg annual return Std dev IR %ccy with profitable strategies 2005 .0% 0.2% 0.P.9% 4% 0.18 2.8% 0.3 to 0.3% 5.1% 3.1% 0.4% 2% 1.5% over ten years. Holding periods are in the one to two month range depending on the lookback. in the case of 60-day lookback. GBP-based strategies result in excess performance of around 1% and IRs of 0.74 181 1 2.5% 0.7% -0.9% 0.com JPMorgan Chase Bank NA Table 2: Dynamic hedging with a price momentum signal: performance of G-10 currencies with USD as the base Excess returns over the 50/50 benchmark.29 89% 63 3 100 Day 2.0% 3.44 100% 252 Day -0.66 1.0% email@example.com% 0. with average trades of one month for shorter windows.1% 0.0% 0.65 100% 100 Day 2. Morgan Results: Forward carry Forward carry results are displayed within Table 4 and Table 5.01 100% 20 Day 2.
The month-end restriction has a significant impact on performance in most cases. Chart 5: Cash flows on CAD hedged into USD based on 60-day forward carry signal Cumulative and monthly cash flows 10% 8% 6% 4% 2% 0% -2% -4% -6% 00 01 02 Monthly Signal CF. rhs Cumulativ e Signal CF. Chart 6: Cash flows on USD hedged into GBP based on 60-day forward carry signal Cumulative and monthly cash flows 10% Monthly Signal CF.5 (appendix Table 6). and from an AUD perspective against USD. This is of course not surprising. Interestingly. lhs Cumulativ e Benchmark CF. rhs Cumulativ e Signal CF. lhs Cumulativ e Benchmark CF. Medium (40-100d lookback windows) are able to generate 1% of excess returns versus the benchmark on average.com JPMorgan Chase Bank NA shorter lookbacks (20 to 60-day). Morgan Chart 7: Cash flows on USD hedged into AUD based on 60-day forward carry signal Cumulative and monthly cash flows 10% Monthly Signal CF. Morgan Robustness tests: month-end restrictions Tables 6 and 7 display the results of the simple momentum and forward carry strategies respectively. most holding periods are in the range of three or four months. rhs 40% 30% 20% 10% 0% -10% -20% 5% 0% -5% -10% 00 01 02 03 04 05 06 07 08 09 10 -30% Source: J. (table 6) it appears that profitability is preserved under the restricted strategy. 03 04 05 06 07 08 09 10 Source: J. In the case of price momentum. rhs 30% 20% 10% 0% -10% -20% -30% Cash flow analysis Excess returns from these signaling models can also be illustrated through cash flows.P.d. with the best performance for shorter lookback windows where average annual returns are around 1%. In each of these cases. resulting in information ratios of around 0.franklin-lyons@jpmorgan. rhs Cumulativ e Signal CF.P. both with rebalancing restricted to the month-end. from a GBP perspective against USD. the monthly rebalancing restriction actually increases returns. as shown in appendix table 4. 2010 Matthew Franklin-Lyons (1-212) 834 4565 matthew. Finally. Additionally.3 to 0. In cases where the alpha strategy signals generate excess returns over the benchmark. for certain lookback windows. and the cumulative monthly returns of the signal based index and the benchmark index from a USD perspective against CAD. Morgan 27 . A shorter lookback window provides consistently positive performance across the G10. the signal based strategy is able to outperform the benchmark index consistently.P.Global FX Strategy Managing FX hedge ratios May 26. lhs Cumulativ e Benchmark CF. The dramatic effect of the month-end restriction on holding periods is also apparent: even in the case of shorter lookback windows. but performance remains positive across G-10 pairs. the AUD based results have positive annual excess returns. rhs 30% 5% 20% 0% 10% -5% 0% -10% 00 01 02 03 04 05 06 07 08 09 10 -10% Source: J. The month-end restriction results in profitable average annual returns in the case of forward carry as well (table 7). Charts 5 through 7 display the monthly returns of the signal based index. adding month-end restrictions greatly increases the holding period of the trades across all strategies. the strategy’s cumulative cash flows will be greater than that of the benchmark strategy. since the monthly restriction is essentially a lower bound on the number of trades that can be taken. although this is likely an aberration.
4% 0.32 NA NA NA 1.2% 0.41 144 1 2.3% 0.7% 1.89 100% 136 1 20 Day 1.Global FX Strategy Managing FX hedge ratios May 26.0% 1.2009 (10 years) Avg annual return Std dev IR %ccy with profitable strategies 2005 .2% 1.3% 0.44 93 2 2.56 100% 55 3 100 Day 1.7% 3.6% 3.com JPMorgan Chase Bank NA Table 4: Dynamic hedging with forward carry signal: performance of G-10 pairs with USD as the base currency Excess returns over the 50/50 benchmark.55 86% 2000 .37 82 2 1.1% 0.P.7% 1.5% 2.2009 (10 years) Avg annual return Std dev IR 2005 .franklin-lyons@jpmorgan. Morgan Table 5: Individual currency performance with forward carry signal with USD as the base currency 60 bus.9% 2.64 100% 60 3 60 Day 1.1% 3.4% 1.2% 1.5% 0.5% 3.9% 0. 2010 Matthew Franklin-Lyons (1-212) 834 4565 matthew.3% 2.3% 4.60 86% 40 Day 1.6% 1.2009 (5 years) Avg annual return Std dev IR %ccy with profitable strategies # of trades* average months per trade* *Using 10y or maximum available history Source: J.9% 0.19 NA NA NA 1.94 101 2 2.9% 3.37 USD/EUR USD/NOK USD/GBP USD/SEK USD/CHF USD/AUD USD/JPY USD/NZD Source: J.4% 1. Table entries are averages of results across all G-10 USD pairs 120 Day 0.7% 2.4% 0.2009 (5 years) Avg annual return Std dev IR # of trades* average months per trade* *Uses 10y or maximum available data period 1. Morgan 28 . day lookback window USD/CAD 2000 .7% 0.9% 0.1% 0.3% 4.6% 0.27 0.29 71% 120 Day 1.7% 0.3% 0.9% 2.32 0.44 90 2 1.6% 3.25 130 1 2.17 4.2% 5.1% 5.6% 0.7% 1.d.32 100% 100 Day 1.86 0.62 88 2 1.P.82 100% 98 2 40 Day 1.8% 0.1% 2.6% 3.2% 3.3% 3.9% 2.19 80 2 1.6% 0.6% 2.3% 3.8% 0.5% 0.50 100% 60 Day 2.13 100% 221 1 20 Day 1.69 74 2 4.9% 0.
9% 3.P.8% 0.26 71% 120 Day 1.1% 2.9% 3.6% 3.9% 0.10 67% 14 14 100 Day 1.63 100% 20 9 60 Day 0.12 67% 41 4 20 Day 1.8% 0.2009 (10 years) Avg annual return Std dev IR %ccy with profitable strategies 2005 .40 100% 200 Day 0.3% 0. Morgan 29 .60 100% 12 15 100 Day 0.9% 2.64 100% 13 13 60 Day 1.0% 2.2% 0.1% 3.0% 0.26 71% 2000 .3% 3.36 89% 22 9 20 Day 0.2% 3.31 86% 100 Day 1.2009 (5 years) Avg annual return Std dev IR %ccy with profitable strategies # of trades* average months per trade* *Using 10y or maximum available history Source: J.3% 3.7% 3. Morgan Table 7: Performance of forward carry signal with month-end restriction and USD as base currency Excess returns over 50/50 benchmark 120 Day 0.9% 0.9% 3.74 89% 31 5 20 Day 0.Global FX Strategy Managing FX hedge ratios May 26. 2010 Matthew Franklin-Lyons (1-212) 834 4565 matthew.9% 2.0% -0.com JPMorgan Chase Bank NA Table 6: Performance of price momentum signal with month-end restriction and USD as base currency Excess returns over 50/50 benchmark 252 Day 1.6% 2.4% 2.2009 (10 years) Avg annual return Std dev IR %ccy with profitable strategies 2005 .6% 0.45 71% 40 Day 1.3% 2.1% 0.4% 3.6% 2.4% 0.9% 0.8% 2.82 100% 22 9 40 Day firstname.lastname@example.org% 0.9% 0.51 86% 100 Day 2.2009 (5 years) Avg annual return Std dev IR %ccy with profitable strategies # of trades* average months per trade* *Using 10y or maximum available history Source: J.38 100% 20 Day 0.35 100% 252 Day -0.1% 0.P.d.8% 0.2% 0.03 67% 57 3 2000 .7% 0.27 86% 40 Day 1.36 100% 40 Day 0.0% 0.21 44% 11 19 200 Day 1.1% 0.8% 3.
Global FX Strategy Managing FX hedge ratios May 26.3% 0.7% 2.1% 2.52 100% 236 1 20 Day 0.32 86% 2000 .4% 0.2009 (10 years) Avg annual return Std dev IR %ccy with profitable strategies 2005 .6% -0.6% 0.03 67% 93 2 100 Day 0.4% 0.2009 (5 years) Avg annual return Std dev IR %ccy with profitable strategies # of trades* average months per trade* *Using 10y or maximum available history Source: J.5% 0. 120 Day 0.5% 2.35 71% 2000 .46 100% 131 1 60 Day 0.06 56% 91 2 200 Day 0.42 100% 77 2 100 Day 0.21 86% 100 Day 0.28 67% 172 1 20 Day 1.d.3% 0.0% 2.4% 0.26 67% 128 2 40 Day 0.04 67% 275 1 20 Day 0.04 44% 204 1 20 Day 0.2% 2.2009 (5 years) Avg annual return Std dev IR %ccy with profitable strategies # of trades* average months per trade* *Using 10y or maximum available history Source: J.5% 0.24 86% 100 Day 1.7% -0.0% 2.20 86% 200 Day 0. 2010 Matthew Franklin-Lyons (1-212) 834 4565 matthew.com JPMorgan Chase Bank NA Appendix charts: Excess returns for price momentum and forward carry signal with EUR.5% 2.0% 2.4% 0.50 100% 87 2 60 Day 1.8% 2.3% 2.5% 2.6% 0.4% 2.28 57% 40 Day 0.2% 0. GBP and AUD as the base currency.P.13 71% 40 Day 0. Table 1: EUR as base currency – performance of price momentum signal Excess returns over the 50/50 benchmark.9% 2.4% 2.2009 (10 years) Avg annual return Std dev IR %ccy with profitable strategies 2005 .5% 0.31 71% 40 Day 0.17 71% 120 Day 1.2% 2.5% 0. Table entries are averages of results across all G-10 EUR crosses.3% email@example.com% 0.6% 2. Table entries are averages of results across all G-10 EUR crosses.08 71% 252 Day 0.P. Morgan 30 .7% 2.5% 0.6% 2.5% 2.6% 0. Morgan Table 2: EUR as base currency – performance of forward carry signal Excess returns over the 50/50 benchmark. 252 Day 0.5% 0.6% 2.
5% 3.23 78% 94 2 100 Day 1.9% 0.4% 0.2009 (10 years) Avg annual return Std dev IR %ccy with profitable strategies 2005 .2009 (5 years) Avg annual return Std dev IR %ccy with profitable strategies # of trades* average months per trade* *Using 10y or maximum available history Source: J.4% 2.1% 3. Table entries are averages of results across all G-10 GBP crosses. 2010 Matthew Franklin-Lyons (1-212) 834 4565 matthew.18 86% 252 Day 1.8% 4.3% 4.14 71% 40 Day 0.18 71% 200 Day 0.34 89% 94 2 60 Day 1.7% 3.0% 3.5% 3.1% 2.5% 0.2% 0.3% 3. Table entries are averages of results across all G-10 GBP crosses.6% 3.P.52 89% 130 1 40 Day 0.0% 3.33 78% 68 3 200 Day 0.47 89% 272 1 2000 .7% 0.com JPMorgan Chase Bank NA Table 3: GBP as base currency – performance of price momentum signal Excess returns over the 50/50 benchmark.27 100% 60 Day 1.P.38 86% 2000 .8% 3. 252 Day 0.1% 0.Global FX Strategy Managing FX hedge ratios May 26.1% 2.31 100% 122 1 40 Day 1.8% 0.1% 0.9% 0.52 100% 20 Day 1.1% 0. 120 Day 0.64 89% 244 1 20 Day firstname.lastname@example.org% 3.1% 0.3% 2.9% 0.38 100% 100 Day 1.8% 0. Morgan Table 4: GBP as base currency – performance of forward carry signal Excess returns over the 50/50 benchmark.d.2% 0. Morgan 31 .28 78% 156 1 20 Day 1.2009 (5 years) Avg annual return Std dev IR %ccy with profitable strategies # of trades* average months per trade* *Using 10y or maximum available history Source: J.29 89% 79 2 100 Day 0.32 100% 40 Day 1.6% 0.0% 3.6% 0.1% 0.2% 2.06 78% 204 1 20 Day 1.07 71% 120 Day 0.8% 3.2% 0.1% 0.3% 0.8% 3.25 86% 100 Day 1.2009 (10 years) Avg annual return Std dev IR %ccy with profitable strategies 2005 .
4% 0.2% 2.11 50% 70 2 200 Day 0.84 100% 249 1 2000 .7% 3. 2010 Matthew Franklin-Lyons (1-212) 834 4565 matthew.26 83% 60 3 100 Day 1.P.1% 0.8% 0. Morgan 32 . 252 Day 0. Table entries are averages of results across all G-10 AUD crosses.9% 3.9% 3.6% 0.P.5% 3.0% 0. Morgan Table 6: AUD as base currency – performance of forward carry signal Excess returns over the 50/50 benchmark.3% 0.65 83% 103 2 40 Day 1.60 100% 226 1 2000 .6% 0.36 100% 100 Day 2.30 83% 140 1 20 Day 1.com JPMorgan Chase Bank NA Table 5: AUD as base currency – performance of price momentum signal Excess returns over the 50/50 benchmark.7% 0.3% 2.d.2009 (5 years) Avg annual return Std dev IR %ccy with profitable strategies # of trades* average months per trade* *Using 10y or maximum available history Source: J. Table entries are averages of results across all G-10 AUD crosses.45 100% 20 Day 1.00 50% 200 Day 0.1% 0.3% 2. 120 Day 0.43 100% 113 1 60 Day email@example.com% 0.3% 2.8% 4.50 100% 40 Day 1.2009 (10 years) Avg annual return Std dev IR %ccy with profitable strategies 2005 .4% 0.5% 3.43 100% 174 1 20 Day 2.9% 0.37 83% 100 Day 1.4% 3.13 67% 75 2 100 Day 1.79 100% 20 Day 3.3% 0.2% 3.8% 4.7% 0.6% 0.2009 (10 years) Avg annual return Std dev IR %ccy with profitable strategies 2005 .4% -0.8% 3.2% 3.4% 4.1% 3.38 100% 40 Day 1.8% 0.27 67% 120 Day 1.67 100% 65 3 60 Day 1.2% 4.7% 0.30 83% 40 Day 1.5% 0.10 50% 252 Day -0.Global FX Strategy Managing FX hedge ratios May 26.0% 0.2% 3.2009 (5 years) Avg annual return Std dev IR %ccy with profitable strategies # of trades** average months per trade** *Using 10y or maximum available history Source: J.8% 2.
2010 Matthew Franklin-Lyons (1-212) 834 4565 firstname.lastname@example.org JPMorgan Chase Bank NA This page left intentionally blank 33 .Global FX Strategy Managing FX hedge ratios May 26.
however. hedging with options has outperformed forwards since 2003.50% on the total FX risk of 100 units. This section compares the returns of a forward versus option strategy in hedging a portfolio employing a 50/50 hedge ratio. A theoretically appropriate way to proceed would back out a notional number that leaves the option-hedged foreign currency risk at 50% of the original exposure. A signalling model which switches between forwards and options does not provide substantial advantage over an outright option based hedge. It is trickier to compute the equivalent notional of options required to perform an apples-to-apples study. distinctly asymmetric return distributions. be higher or lower than the subsequent delivered performance. the literature on derivatives-based hedging is relatively light. exactly 34 . From a corporate perspective.g. whose value fluctuates solely due to variations in the local currency/USD exchange rate. this means buying 100 units notional of foreign currency puts. The universe of option hedges examined is limited to 1-mo vanillas. Note that the results in the table hold only for US investments by foreign investors. • • FX hedging analysis is typically conducted based on forwards rather than options. unlike Sections II – IV of this paper which presented results across a broader range of base currencies and related crossrates. Of course. Although there exist many empirical studies on forward contracts to hedge currency risk.x. An alternative uses the same notional of options as the notional of the benchmark forward hedge ─ in this example. This approach renders a portfolio with a (foreign currency) put option overlay systematically under-hedged vis-à-vis one with a short (foreign currency) forward overlay. but the second option using equivalent notionals will be worth exploring in forthcoming research. thereby avoiding the influence of moves in core holdings on the results. We also test the efficacy of a switching rule which positions in either forwards or options depending upon the conviction of the signal in forecasting a spot trend. causing the hedge ratio to vary between the upper and lower delta thresholds of 0% and 100%. The 50% hedge ratio holds true only at inception. EUR and GBP exposure. This section backtests a series of option-based alternatives to forward hedges since 2003. The notional size of the forward hedges considered in Section IV is an estimate of the expected value of the core asset holdings (e. The analysis contains one important simplification relative to the previous section. interest rates. the company is long USD and short AUD. and to only the two structures most commonly used for hedging purposes: straight ATM puts on the foreign currency. EUR and GBP based investors with USD exposure. this is taken to be 50% of the total currency risk hedged using 1-mo forwards: the benchmark hedge for 100 units of foreign currency exposure is to sell 50 units of it forward. Option premia can.com JPMorgan Chase Bank NA V. For completeness.sandilya@jpmchase. 50 units of the foreign currency. This conclusion is sensitive to methodology. which may not be desirable. The study assumes that non-US investors hold $100 of US assets.or over-hedged relative to the forward hedge benchmark. 25D and 40D risk reversals (long foreign currency puts versus short foreign currency calls). Because the purpose of this section is to compare the efficacy of two hedging instruments. JPMorgan bond indices). Chart 1 depicts the hedged and unhedged returns from the perspective of AUD. or a hedge ratio range of 0% . Unlike forwards that offer a constant coverage ratio. the directional exposure from an ATM option sized according to this scheme can range between 0 to 50 foreign currency units. But given that option deltas vary between 0 and 1. however. mimicking the base case forward hedge.Global FX Strategy Managing FX hedge ratios May 26. options outperformed for long USD/short AUD. options should also be considered. Thus market moves alone can leave investors either under. The sample period for the study is 2003 to the present. the non-linearity of payoffs comes at a cost ─ the option premium ─ which can be a disincentive for some hedgers given the zero upfront cost of forwards. EUR and GBP. Base case: replicating a 50/50 hedge ratio This analysis assumes a base case forward hedge for benchmarking the performance of option hedges. Hedging with forwards versus options • • Compared to extensive studies on forwards-based hedging. For ATM options with delta of approximately 50. however. EUR and GBP. work on options approaches is limited. As before.based investors. an unavoidable consequence of this approach. and 10D. From a corporate perspective. option deltas fluctuate with moves in spot. reflecting the availability of options data. and Table 1 details the performance of the various hedging strategies. For consistency with Section IV of this paper we prefer the former scheme. 2010 Arindam Sandilya (1-212) 834-2304 arindam. The primary difference between the two instruments is options’ non-linear payoff which leads to complex. For AUD. volatility and the passage of time. it is not immediately clear a priori if the necessity of an upfront cash payment in itself impairs the effectiveness of these instruments. we hold the notional of foreign currency exposure constant here.
6% 4.43 -0.8% 6.30 -0.5% 0. more in AUD than in EUR or GBP.19 0.2% 6.6% 0.8% 13.33 0.13 0.com JPMorgan Chase Bank NA The narrower sample here reflects the microstructure of FX option markets – options on crosses are less liquid than those in the USD majors.5% 5.6% 5. This pattern is largely attributable to the direction of the underlying spot rates over the backtest period.7% 5. Table 1. 2010 Arindam Sandilya (1-212) 834-2304 arindam. This means that while notional size is not an issue for AUD-based investors hedging European or UK assets through forwards.06 0.30 -0.6% -1.2% 5.2% -2. In other words.8% -3.2% 5.9% 5.5% -1. Benchmark hedge is a 50% short foreign currency forward.0% 6.24 -0.31 -0. options performance is shown for USD-pairs only. The more out-of-the-money the two strikes. With dollar weakness prevailing for most of the last seven years.3% 7. Given this liquidity issue.4% 8. Assumes USD exposure of $100 for all local currencies. Option-based hedges (ATM puts or risk reversals) have outperformed forwards and a rule-based switching strategy between forwards and options.4% -1.6% 8. This phenomenon was especially pronounced in AUD. USD put/local currency call options spent the majority of their lives in the money. Morgan Results: a signalling model for switching between forwards and options Table 1 also lists the performance of a signal-based hedging strategy that switches between a forward and an equivalent delta option or a risk reversal depending on the conviction from the alpha models described in 35 .35 -0. the more obvious the delinkling from the performance of forwards.0% 1.6% 5. thus detracting from hedge performance relative to forwards. EUR and GBP have encountered steeper and more extended spot declines against USD than AUD over the past six months. Intuitively. ATM foreign currency puts possess favorable dHedgeRatio/dSpot properties: the effective notional of the hedge increases precisely when needed ─ during periods of a declining foreign currency.28 -0.P.8% 0. This convexity of course comes at a price in the form of paying away premium for put options that expire worthless in periods of strengthening foreign currency. deploying EUR/AUD or GBP/AUD puts as hedges runs into notional constraints.26 -0.4% -0.8% 5.Global FX Strategy Managing FX hedge ratios May 26.x.9% 5.1% 4.6% -1. and ATM USD puts/currency calls therefore have given back a large part of their outperformance relative to forwards over this period.28 0.7% -6.1% -1.07 0.6% -1.8% 10. Option hedges are sized to provide the same hedge ratio as the benchmark forward hedge at inception.29 -0.15 0.38 -0. Results: forwards versus options The charts and table highlight the following: • Given the dollar’s general downtrend against most currencies over the sample.28 -0. period.5% -1.5% 4.18 0. • Risk reversals perform better in EUR and GBP than in AUD.9% 10. (Thus symmetry becomes clearer if one thinks of a forward as the combination of a call and a put.sandilya@jpmchase. where the penalty was substantially higher than in most currencies. In option parlance.9% 1.48 -0.5% IR -0. which turned out to be quite profitable in a declining dollar environment. options generally overhedged investors relative to the benchmark.1% -1.18 0.6% 4.01 -0.08 -0.5% -1. given their similarity to a forward in being exposed to both appreciation and depreciation of in the underlying exchange rate. risk reversals should track the forward hedge.0% 0.1% 5.8% 1.0% 1. No transaction costs Local Currency Strategy Unhedged Hedged v ia Forw ard Hedged v ia Option Hedged v ia 40dRiskRev ersal Hedged v ia 25dRiskRev ersal Hedged v ia 10dRiskRev ersal Hedged v ia Signal (option and fw d) Hedged v ia Signal (25d rr and fw d) Unhedged Hedged v ia Forw ard Hedged v ia Option Hedged v ia 40dRiskRev ersal Hedged v ia 25dRiskRev ersal Hedged v ia 10dRiskRev ersal Hedged v ia Signal (option and fw d) Hedged v ia Signal (25d rr and fw d) Unhedged Hedged v ia Forw ard Hedged v ia Option Hedged v ia 40dRiskRev ersal Hedged v ia 25dRiskRev ersal Hedged v ia 10dRiskRev ersal Hedged v ia Signal (option and fw d) Hedged v ia Signal (25d rr and fw d) Average Return -6. hedging currency exposure on US assets has tended to be profitable.19 AUD EUR GBP Source: J. 10D risk reversals exhibit interesting behavior in that selling wing USD calls for a leveraged notional was a painful experience during the credit crisis as vols and skews blew out.1% Volatility 15. both struck at the forward.5% -1. a risk reversal can be created by simply pulling the call and strikes away from the forward strike in opposite directions). • ATM options sized to provide the same hedge ratio as the benchmark forward hedge at inception generally dominate forwards as hedge instruments. and effective hedge ratios offered by options were generally greater than the 50% provided by the benchmark forward hedge.7% 0.
1% with options alone). so would justify hedging with options instead. a simple directional signal which does not take into account the expected extent of spot’s upward or downward movement becomes difficult to apply.x. This pattern probably owes to the filter rule.6% with forwards plus options versus 0. An outright option-based hedging strategy has delivered better performance for all currencies. The strategies’ information ratios – defined as outperformance versus benchmark divided by the volatility of outperformance – indicate that while the switching rule applied either to an ATM put or a risk-reversal outperforms forward hedges in most cases. The delta-adjusted methodology probably also shapes the outcome. the rule does not incorporate rich/cheap considerations in option prices as an input. By allowing the option to have a notional twice as large as the forward contract.com JPMorgan Chase Bank NA Section IV. which is a critical consideration for most investors. 2010 Arindam Sandilya (1-212) 834-2304 arindam. the magnitude is small. although this conclusion could be sensitive to sample period and baseline hedge ratio chosen (50/50 benchmark versus a neutral of either zero or 100% hedging). so would justify hedging with forwards.8%). AUD-based investors actually increase the volatility of their hedged portfolio by switching between forwards and options (8. A buy or sell signal from the models represents a high-conviction view on spot’s direction.sandilya@jpmchase. 36 .2%) as opposed to hedging purely with options (7. A more neutral framework to assess the signal’s value would maintain the same notional of the two hedging instruments. For instance. even as the returns from doing so are lower (-0. This issue can be addressed in subsequent research.Global FX Strategy Managing FX hedge ratios May 26. While correct more than 50% of the time in identifying the appropriate spot environments for switching into options. A neutral reading from the models represents a low-conviction view.
Hedging USD exposure has been prudent AUD. Morgan EUR/USD 100 100 95 90 90 85 80 75 Unhedged Fw d Hedged ATM Option Hedged Jun-04 Dec-05 Jun-07 Nov -08 May -10 70 65 Jan-03 Jun-04 Dec-05 Jun-07 Nov -08 May -10 40D RR Hedged 25D RR Hedged 10D RR Hedged 80 70 60 Jan-03 Source: J. and GBP.Global FX Strategy Managing FX hedge ratios May 26.com JPMorgan Chase Bank NA Chart 1.P.x.sandilya@jpmchase.P. Morgan 37 .P. 2010 Arindam Sandilya (1-212) 834-2304 arindam. Morgan 75 Unhedged Fw d Hedged ATM Option Hedged 70 65 60 55 50 45 40D RR Hedged 25D RR Hedged 10D RR Hedged Jun-04 Dec-05 Jun-07 Nov -08 May -10 Jan-03 Jun-04 Dec-05 Jun-07 Nov -08 May -10 Source: J. Morgan 40D RR Hedged 25D RR Hedged 10D RR Hedged Jun-04 Dec-05 Jun-07 Nov -08 May -10 100 Jun-04 Dec-05 Jun-07 Nov -08 May -10 Jan-03 Source: J. Option hedges are sized to provide the same hedge ratio as the benchmark forward hedge at inception. Morgan Source: J. EUR. Morgan GBP/USD 80 75 70 65 60 55 50 45 Jan-03 Source: J. Benchmark hedge is a 50% short foreign currency forward. No transaction costs AUD/USD 250 200 Unhedged Fw d Hedged ATM Option Hedged 180 160 140 200 150 120 100 Jan-03 Source: J.P.P. Assumes core USD exposure of $100 for all local currencies.based investors since 2003.P. and ATM options have performed at par or better than forwards.
the disclosure may be based on the month end data from two months’ prior. No.P. Morgan Securities Australia Limited (ABN 61 003 245 234/AFS Licence No: 238066) is a Market Participant with the ASX and regulated by ASIC. Ltd. Legal Entities Disclosures: U. Morgan Equities Limited. JPMSAL does not issue or distribute this material to “retail clients. Morgan Equities Limited is a member of the Johannesburg Securities Exchange and is regulated by the FSB. Australia: This material is issued and distributed by JPMSAL in Australia to “wholesale clients” only. 2010 John Normand (44-20) 7325-5222 john. is regulated by the Korea Financial Supervisory Service.hk/prod/dw/Lp.P.P. Frankfurt Branch which are regulated by the Bundesanstalt für Finanzdienstleistungsaufsicht. arranging. N. Morgan Saudi Arabia Ltd. the research analyst denoted by an “AC” on the cover or within the document individually certifies. South Africa: J.P. 47 and 49 of the Financial Services and Markets Act 2000 (Financial Promotion) Order 2005 (all such persons being referred to as "relevant persons"). Morgan Securities Indonesia is a member of the Indonesia Stock Exchange and is regulated by the BAPEPAM LK. Korea: J.com.A.K. Investment research issued by JPMSL has been prepared in accordance with JPMSL's policies for managing conflicts of interest arising as a result of publication and distribution of investment research. Morgan Chase Bank. Japan: There is a risk that a loss may occur due to a change in the price of the shares in the case of share trading.: 199405335R] which is a member of the Singapore Exchange Securities Trading Limited and is regulated by the Monetary Authority of Singapore (MAS) and/or JPMorgan Chase Bank. J. or will be directly or indirectly related to the specific recommendations or views expressed by the research analyst(s) in this report. Morgan Broking (Hong Kong) Limited is the liquidity provider for derivative warrants issued by J.P.P..Building 3. Seoul Branch. which is regulated by the Japan Financial Services Agency (FSA). For the purposes of this paragraph the terms “wholesale client” and “retail client” have the meanings given to them in section 761G of the Corporations Act 2001.P. is. where multiple research analysts are primarily responsible for this report.com J.P. N. Morgan Securities Ltd. is a member of the NFA. Singapore: This material is issued and distributed in Singapore by J.P. Taiwan: J. Saudi Arabia: J.P. Morgan Pakistan Broking (Pvt.A.A. Hong Kong: The 1% ownership disclosure as of the previous month end satisfies the requirements under Paragraph 16. An updated list can be found on HKEx website: http://www. PO Box 506551.P. Thailand: JPMorgan Securities (Thailand) Limited is a member of the Stock Exchange of Thailand and is regulated by the Ministry of Finance and the Securities and Exchange Commission. 38. with respect to each security or issuer that the research analyst covers in this research) that: (1) all of the views expressed in this report accurately reflect his or her personal views about any and all of the subject securities or issuers.P. Germany: This material is distributed in Germany by J.normand@jpmorgan. Morgan Securities Ltd Analyst Certification: The research analyst(s) denoted by an “AC” on the cover of this report certifies (or. 2711006. London EC2Y 5AJ. advising and custody. is a member of the Philippine Stock Exchange and is regulated by the Securities and Exchange Commission. N. Dubai: JPMorgan Chase Bank. Country and Region Specific Disclosures: U.P. (For research published within the first ten days of the month. U. Morgan Structured Products B. Riyadh 11553. Morgan Securities Philippines Inc.A. is regulated by the Comissao de Valores Mobiliarios (CVM) and by the Central Bank of Brazil.P. Morgan India Private Limited is a member of the National Stock Exchange of India Limited and Bombay Stock Exchange Limited and is regulated by the Securities and Exchange Board of India. Australia: J.O.P. and (2) no part of any of the research analyst’s compensation was. Many European regulators require that a firm to establish. UAE. P..hkex. Morgan Securities (Asia Pacific) Limited (CE number AAJ321) is regulated by the Hong Kong Monetary Authority and the Securities and Futures Commission in Hong Kong. only to persons of a kind described in Article 19 (5). the report has been issued to persons regarded as professional investors (or equivalent) in their home jurisdiction. Morgan Cazenove is a brand name for equity research produced by J.. Morgan Securities Singapore Private Limited (JPMSS) [MICA (P) 020/01/2010 and Co..” The recipient of this material must not distribute it to any third party or outside Australia without the prior written consent of JPMSAL.K. issued and approved for distribution in the U. Singapore branch (JPMCB Singapore) which is regulated by the MAS.htm. and that a loss may occur due to the exchange rate in the case of foreign share trading. Reg.. India: J. Morgan Bank International LLC. Pakistan: J.) J. Morgan Securities Ltd. Morgan Securities (Far East) Ltd..P. N.P. Al-Faisaliyah Tower.P. Hong Kong: J. JPMorgan Chase Bank. N. Dubai. In other EEA countries. Frankfurt Branch and J.S. King Fahad Road. Kingdom of Saudi Arabia.. and J.A. Malaysia: This material is issued and distributed in Malaysia by JPMorgan Securities (Malaysia) Sdn Bhd (18146-X) which is a Participating Organization of Bursa Malaysia Berhad and a holder of Capital Markets Services License issued by the Securities Commission in Malaysia.K. This report has been issued in the U. Morgan Securities Ltd. Dubai Branch is regulated by the Dubai Financial Services Authority (DFSA) and its registered address is Dubai International Financial Centre . (JPMSL) is a member of the London Stock Exchange and is authorised and regulated by the Financial Services Authority. FINRA and SIPC. Morgan Futures Inc. is a member of FDIC and is authorized and regulated in the UK by the Financial Services Authority. Indonesia: PT J.P.P.: J.P.V. Morgan Securities (Taiwan) Limited is a participant of the Taiwan Stock Exchange (company-type) and regulated by the Taiwan Securities and Futures Bureau. and the EEA by JPMSL. Morgan Australia Limited (ABN 52 002 888 011/AFS Licence No: 238188) is regulated by ASIC and J. JPMorgan Chase Bank..P. Disclosures: J. (JPMSI) and its non-US affiliates worldwide. J. Morgan S.P. implement and maintain such a policy. Registered Office 125 London Wall.A.P. Any investment or investment activity to which this document relates is only available to relevant persons and will be engaged in only with relevant persons. Registered in England & Wales No.: JPMSI is a member of NYSE.Global FX Strategy Managing FX hedge ratios May 26. Philippines: J. Morgan is the global brand name for J. Box 51907. This document must not be acted on or relied on by persons who are not relevant persons.5(a) of the Hong Kong Code of Conduct for Persons Licensed by or Registered with the Securities and Futures Commission. Dubai Branch. P. and European Economic Area (EEA): Unless specified to the contrary.P. Japan: This material is distributed in Japan by JPMorgan Securities Japan Co. with respect to securities business under licence number 35-07079 and its registered address is at 8th Floor.) Ltd is a member of the Karachi Stock Exchange and regulated by the Securities and Exchange Commission of Pakistan. Level 7. In the case of share 38 . Morgan Securities Inc. is authorised by the Capital Market Authority of the Kingdom of Saudi Arabia (CMA) to carry out dealing as an agent. J. and listed on the Stock Exchange of Hong Kong Limited. Brazil: Banco J.K.
the specific holding is disclosed in the Important Disclosures section above. a prospectus. an advertisement. This material is not intended as an offer or solicitation for the purchase or sale of any financial instrument. in the course of and for the purposes of their business. formed or created under the laws of Canada or a province or territory of Canada. sold or redistributed without the written consent of J. Morgan) do not warrant its completeness or accuracy except with respect to any disclosures relative to JPMSI and/or its affiliates and the analyst’s involvement with the issuer that is the subject of the research. market conditions or any other publicly available information. Morgan Securities Ltd trading. All pricing is as of the close of market for the securities discussed. Singapore: JPMSS and/or its affiliates may have a holding in any of the securities discussed in this report. or needs and are not intended as recommendations of particular securities. a public offering. Dubai: This report has been issued to persons regarded as professional clients as defined under the DFSA rules. not for sale.S. The opinions and recommendations herein do not take into account individual client circumstances. Morgan. objectives.P. Pakistan: For private circulation only.. affiliates and accepts responsibility for its contents. the information contained herein or the merits of the securities described herein. Morgan Securities (Far East) Ltd.. and any representation to the contrary is an offence. not for sale. The Financial Futures Association of Japan. Ltd. 2010. New Zealand: This material is issued and distributed by JPMSAL in New Zealand only to persons whose principal business is the investment of money or who...P. alternatively.General: Additional information is available upon request. Kanto Local Finance Bureau (kinsho) No. Any offer or sale of the securities described herein in Canada will be made only under an exemption from the requirements to file a prospectus with the relevant Canadian securities regulators and only by a dealer properly registered under applicable securities laws or. Canada: The information contained herein is not. an offer to sell securities described herein. No securities commission or similar regulatory authority in Canada has reviewed or in any way passed judgment upon these materials. Seoul Branch. JPMorgan Securities Japan Co. and under no circumstances is to be construed as. Clients should contact analysts and execute transactions through a J.S.. habitually invest money. Financial Instruments Firms: JPMorgan Securities Japan Co. JPMSAL does not issue or distribute this material to members of "the public" as determined in accordance with section 3 of the Securities Act 1978. 82 Participating Association / Japan Securities Dealers Association. The recipient of this report must make its own independent decisions regarding any securities or financial instruments mentioned herein. JPMSI distributes in the U. or its affiliates and/or subsidiaries (collectively J. This report or any portion hereof may not be reprinted.normand@jpmorgan. India: For private circulation only.com J. Past performance is not indicative of future results. To the extent that the information contained herein references securities of an issuer incorporated. Opinions and estimates constitute our judgment as of the date of this material and are subject to change without notice. financial instruments or strategies to particular clients. 39 . research published by non-U. The information contained herein is under no circumstances to be construed as investment advice in any province or territory of Canada and is not tailored to the needs of the recipient. The recipient of this material must not distribute it to any third party or outside New Zealand without the prior written consent of JPMSAL.P.P. All rights reserved. Ltd. any trades in such securities must be conducted through a dealer registered in Canada. Korea: This report may have been edited or contributed to from time to time by affiliates of J. Copyright 2010 JPMorgan Chase & Co. 2010 John Normand (44-20) 7325-5222 john. Periodic updates may be provided on companies/industries based on company specific developments or announcements. “Other Disclosures” last revised March 1. for securities where the holding is 1% or greater. pursuant to an exemption from the dealer registration requirement in the relevant province or territory of Canada in which such offer or sale is made. will be receiving a brokerage fee and consumption tax (shouhizei) calculated by multiplying the executed price by the commission rate which was individually agreed between JPMorgan Securities Japan Co. unless otherwise stated. and the customer in advance. Information has been obtained from sources believed to be reliable but JPMorgan Chase & Co. Ltd.P. in Canada or any province or territory thereof. or solicitation of an offer to buy securities described herein.. Morgan subsidiary or affiliate in their home jurisdiction unless governing law permits otherwise.Global FX Strategy Managing FX hedge ratios May 26.
email@example.com@jpmorgan.Global FX Strategy Managing FX hedge ratios May 26.s.com paul. 2010 John Normand (44-20) 7325-5222 john.com matthew.com junya.com firstname.lastname@example.org@jpmorgan.s.com New York Ken Landon Gabriel de Kock Niall O’Connor Arindam Sandilya Matthew Franklin-Lyons MD ED ED ED Analyst G-10 FX Strategy G-10 FX Strategy Technical Strategy FX Derivatives Strategy G-10 FX Strategy (1-212) 834-2391 (1-212) 834-4254 (1-212) 834-5108 (1-212) 834-2304 (1-212) 834-4565 kenneth.P.com Asia Tohru Sasaki Junya Tanase Yen Ping Ho Yoonyi Kim MD ED VP (Emerging Markets Research) Analyst G-10 FX Strategy G-10 FX Strategy Asia FX Strategy G-10 FX Strategy (81-3) 6736-7717 (81-3) 6736-7718 (65) 6882-2216 (81-3) 67367729 tohru.de.com email@example.com@jpmorgan.com firstname.lastname@example.org. Morgan Global FX Strategy London John Normand Paul Meggyesi Thomas Anthonj Kamal Sharma Talis Bauer MD MD ED VP VP Head.com gabriel.com kamal.com J.email@example.com@firstname.lastname@example.org yenping.com 40 .x.sharma@jpmorgan.P.com yoonyi.com thomas.franklin-lyons@jpmorgan. Morgan Securities Ltd J.ho@jpmorgan. Global FX Strategy G-10 FX Strategy Technical Strategy G-10 FX Strategy FX Derivatives Strategy (44-20) 7325-5222 (44-20) 7859-6714 (44-20) 7742-7850 (44-20) 7777-1729 (44-20) 7777-5276 email@example.com@jpmorgan.x.
This action might not be possible to undo. Are you sure you want to continue?