Excess Returns In FX Volatility Markets

The academic paper we review this week shows that a carry trade in volatility, a long-short strategy that buys implied volatility at a discount and sells implied volatility at a premium, generates a significant Sharpe ratio. The associated volatility returns are virtually uncorrelated with the traditional carry trade and other popular currency strategies.

The Study in a Nutshell

There is a growing body of literature documenting the existence of pronounced volatility risk premia, particularly in the short run. A new paper, published in the Journal of Financial Economics, investigates the properties of these premia in FX markets by examining the cross-sectional differences in the returns of forward volatility agreements (FVAs).

Their analysis yields two key results:

1. A zero-cost strategy that buys FVAs with downward sloping implied volatility curves and sells FVAs with upward sloping curves (a vol:carry strategy) generates significant excess returns.
2. The volatility carry premium remains both statistically and economically large for the implied volatility curve out to two years.

The proposed vol:carry (VCA) strategy outperforms the traditional FX carry trade strategy whereby an investor is long currencies with high yields and short currencies with low yields.

Methodology and Data

The authors use forward volatility agreements (FVAs) to examine volatility risk premia in the FX market. These are over-the-counter contracts that allow an investor to take positions on the future level of implied volatility. For example, an investor believes that EUR/USD three-month implied volatility will rise from the current (say) forward level of 7.8% to 10% over the next month; they would buy an FVA in EUR/USD three-month implied vol. If their prediction was correct, after a month they would roughly earn 10% minus 7.8% (i.e. 2.2%) multiplied by some payout amount. However, they would suffer a loss if the level of implied volatility was lower than the forward volatility of 7.8% at the contract’s start. This is much like taking a position in FX forwards to take a view on spot, but instead implied volatility is the target instrument rather than spot.

For their data sample, the authors use 20 developed and emerging market currencies: those of Australia, Brazil, Canada, the Czech Republic, Denmark, the Euro area, Hungary, Japan, Mexico, New Zealand, Norway, Poland, Singapore, South Africa, South Korea, Sweden, Switzerland, Taiwan, Turkey, and the United Kingdom.

The data starts with nine currencies at the beginning of the sample in 1996 and ends with 20 currencies at the end of the sample in 2015. The data is on European calls and puts for currency pairs vis-à-vis the US dollar for the following maturities: one month, three months, six months, 12 months, and 24 months.

The Existence of Volatility Risk Premia

The authors then use the data to examine average excess returns for different maturities and currency pairs. They also examine how the returns vary over time.

On excess returns, they find there does exist a FX volatility premium across a basket of currencies. However, it only exists for shorter maturity FVAs, i.e. a one-month agreement for two-month implied volatility. For such FVAs, one could earn a Sharpe ratio of 0.85 by systematically selling such FVAs. The returns disappear for longer maturity options. However, they do find that the volatility risk premium varies over time. Across all 20 currencies they find that forward implied volatility is a biased predictor of future spot implied volatility even at longer maturities. This means that whenever the implied volatility curve is steep (or forward implied vol. is above current implied vol.), there is scope to earn positive returns.

They test an FVA investment strategy where they use the slopes of the implied volatility curves as the sorting variable. They allocate the FVAs into five baskets using the volatility slope of each currency pair. They then rank these portfolios from high to low slope such that Portfolio 1 contains the 20% of all FVAs with the highest slope and Portfolio 5 comprises the 20% of all FVAs with the lowest slope.

Two additional strategies are created: a level strategy, denoted LEV, which corresponds to a zero-cost strategy that equally invests in all implied volatility portfolios; and, importantly, a volatility carry strategy, denoted VCA, which is equivalent to a long-short strategy that buys Portfolio 5 and sells Portfolio 1.

They find the following results:

• Average excess returns increase monotonically from the highest slope FVAs to the lowest slope FVAs.
• Average excess returns of the LEV portfolio are statistically significant for shorter maturities with a Sharpe ratio of 0.65 (if one sells the positive slopes).
• Average excess returns of the VCA strategy are always positive and highly significant, with Sharpe ratios of 1.46 for shorter-dated options and 1.25 for longer-dated months.

Chart 1 presents the results from the two zero-cost portfolio strategies. Average excess returns are clearly much higher (and statistically significant) for the vol:carry strategy.

Source: Page 15 in Della Corte, Kozhan and Neuberger (2020)

Chart 2 presents the one-year rolling Sharpe ratio for the VCA strategies. The Sharpe ratios are monotonically decreasing across longer maturities. The chart shows that the strategies exhibit a clear counter-cyclical pattern producing higher risk-adjusted excess returns during the financial crisis and lower risk-adjusted excess returns otherwise.

Source: Page 8 in Della Corte, Kozhan and Neuberger (2020)

The Bottom Line

A zero-cost portfolio strategy that buys forward volatility agreements in currencies with the lowest slopes (or forward volatility premia) and sells those with the highest slopes produces significant excess returns. A volatility carry strategy fully explains variations in these volatility excess returns. The lower the slope of the implied volatility curve, the more the FVA return is exposed to the volatility carry premium. Unlike the unconditional volatility risk premium, which vanishes beyond two to three months, the volatility carry premium is manifest at maturities up to 24 months (longer-term volatility shocks are priced conditionally).

To view the full paper, please click here