FX Volatility Tracker: EUR/USD Volatility Remains Lowest

Summary

We recently extended our measures of realised volatility to include versions that measure core (or base) realised volatilities and exclude major event days from their calculation. This complements our five different measures of realised volatility that include all days.

Our latest data finds the following:

Using all data including event days

• As was the case last week, the conventional measure of realised volatility (close-to-close) could be slightly understating 3-month volatility for EUR/USD and AUD/USD but slightly overstating it in GBP/USD and USD/JPY (Tables 1 to 4).

• In terms of whether 3-month implieds are cheap or expensive versus our range of realised volatility measures, we find the following:

• EUR/USD 3-month implieds are now expensive relative to our realised range (Chart 1). Previously, they were fairly valued.
  • GBP/USD 3-month implieds inch closer to the lower end of our realised volatility estimators (Chart 1).
  • USD/JPY and AUD/USD 3-month implieds remain below their measures of realised volatility (Chart 1).
    
• Looking at term structure, we find that 1- and 3-month EUR/USD implieds (8.9% and 8.5%, respectively) are expensive relative to our realised estimators while 6-month and 1-year implieds (8.2% and 8.1%, respectively) are cheaper (Chart 2). Elsewhere, we find that 1-month USD/JPY implieds (12.7%) are expensive relative to our realised estimators while they remain cheaper relative to realised for the rest of the curve (Chart 3).
  
• Our statistical tests (Augmented Dickey-Fuller (ADF)) are showing that neither EUR/USD and USD/JPY volatility risk premium are exhibiting mean-reversion behavior relative to the last year (Charts 8 and 9).

Using realised data that excludes event days

• Stripping out big events from our calculations can give a better sense of base volatility. We remove big events (FOMC, ECB, BoJ, US CPI and NFP days) from our calculations of realised volatility from the last three to fourth months.
• Using these base measures, we find that 1- and 3-month EUR/USD implieds become fairly valued (Chart 4). Meanwhile 3-month USD/JPY implieds are also now fairly valued (Chart 5).

Comparing Different Measures of Volatility Across FX

Tables 1-4 show the realised volatility for each currency we track using each of the different estimators along with the corresponding implied volatilities.

Looking at the average volatility across each estimator (excluding implied volatility), over each time period, for each currency, we find that:

• EUR/USD is the least volatile pair over all horizons (1 month to 1 year, Table 1).
• AUD/USD is the most volatile pair over 1-month, 6-month, and 1-year horizons (Table 4).
• USD/JPY is the most volatile pair over a 3-month horizon (Table 2).

Comparing Volatility Risk Premium Across FX

Tables 5-8 show the volatility risk premium (implied volatility minus realised volatility) for each currency across each of the different estimators.

Volatility Risk Premium Percentiles

Tables 9-12 show the 5-year percent rank of the volatility risk premium across each currency, estimator, and time period (1 month to 1 year).

Appendix

We recently published a note comparing different measures of realised volatility. This includes the most common measure of realised volatility – close-to-close volatility (also referred to as historical volatility), which is just the standard deviation of (logarithmic) returns over a historical period.

We also consider range-based volatility estimators that make use of intraday price data (e.g., open, high, low, and close) to derive an estimator of realised volatility that may be more precise and efficient than the standard close-to-close measure.

Notably, we implemented five different measures of realised volatility – Table 13 provides a summary of each.

In this report, we update the estimators with the latest data and look at the volatility risk premium across several currencies we track.

Dalvir Mandara is a Quantitative Researcher at Macro Hive. Dalvir has a BSc Mathematics and Computer Science and an MSc Mathematical Finance both from the University of Birmingham. His areas of interest are in the applications of machine learning, deep learning and alternative data for predictive modelling of financial markets.

Photo Credit: depositphotos.com

(The commentary contained in the above article does not constitute an offer or a solicitation, or a recommendation to implement or liquidate an investment or to carry out any other transaction. It should not be used as a basis for any investment decision or other decision. Any investment decision should be based on appropriate professional advice specific to your needs.)