FX Volatility Tracker: Are USD/JPY Implieds Cheap?

Summary

We recently extended our measures of realised volatility to include versions that measure core (or base) realised volatilities that 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

• We find the conventional measure of realised volatility (close-to-close) could be understating volatility for EUR/USD and GBP/USD but could be overstating volatility in USD/JPY and AUD/USD (Tables 1 to 4). Our measures of realised vol incorporate intraday data.

• In terms of whether 3-month implieds are cheap or expensive versus our range of realised volatility measures, we find the following:
  • EUR/USD 3m implieds are fairly valued (Chart 1).
  • USD/JPY, GBP/USD and AUD/USD 3m implieds are cheap as implieds are below our measures of realised.
    
• Looking at term structure, we find that 1- and 3- month EUR/USD implieds (8.3% and 8.2%, respectively) are fairly valued. Meanwhile, implieds are lower than our realised measures for the rest of the curve (Chart 2). Elsewhere, we find 1-month USD/JPY implieds are fairly valued, while they remain cheap for the rest of the curve (Chart 3).
  
• Although USD/JPY implieds are cheap to realised (i.e., negative vol risk premium, Table 6), our statistical tests (Augmented Dickey-Fuller (ADF)) suggest that USD/JPY is not displaying clear mean reversion behavior yet relative to the past year (Chart 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 3-month EUR/USD implieds remain fairly valued, while 1- and 3-month USD/JPY implieds are no longer cheap as they are sitting within the realised range (Charts 4 and 5). This suggests that one needs to take a view on BoJ events to determine whether USD/JPY implieds are cheap.

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 over all periods (1 month to 1 year, Table 1).
• USD/JPY is most volatile over 1- and 3-month horizons (Table 2)
• AUD/USD is most volatile over 6 month and 1-year horizons (Table 4).

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 Trend

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.)