Define USD Concentration (what you are measuring)
USD Concentration is a way to summarize how much of an exposure is denominated in U.S. dollars at a given time. To measure volatility, you first need a consistent definition of the numerator and denominator.
A common structure is:
- USD share = USD-denominated exposure / total exposure across all relevant currencies.
“Exposure” should be defined the same way each time point (for example: spot holdings, open position notional, or mark-to-market value). If you change the exposure measure, the “concentration” series may change for reasons unrelated to currency mix.
Assumptions you must state:
- Which balances/positions are included in total exposure.
- Whether values are converted to a base currency for aggregation (and which exchange rates are used).
- The time step (daily, weekly, monthly) and the observation time within each step.
Mechanics: turning a concentration series into a volatility measure
Once you have a time series of USD share values, you can quantify how much it fluctuates.
1) Volatility as standard deviation of concentration
If you compute USD share for each period t, denoted C_t, then one simple volatility measure is the standard deviation:
- Volatility ≈ stdev(C_t) over the chosen window.
This treats upward and downward changes symmetrically around the mean concentration.
2) Volatility as average absolute change (robust to outliers)
Another choice is to look at how much the concentration changes from one period to the next:
- Average absolute change ≈ average(|C_t − C_{t-1}|).
This is often easier to interpret as “typical shift per period” and can be less sensitive to a few extreme jumps.
3) Growth-rate style measures (when shares behave nonlinearly)
If concentration can move sharply because the denominator changes (for example, total exposure changes as positions are added/closed), raw share volatility may mix “true mix change” with “size change.” In that case, you can separate effects by also tracking:
- USD exposure level (not just share)
- Total exposure level Then analyze whether the share is moving primarily due to USD exposure changing, total exposure changing, or both.
To do this consistently, state whether you are using:
- Value-based exposure (mark-to-market) or notional-based exposure.
- Fixed observation rules for when positions are counted.
Evidence or example: a self-check using simple assumptions
Consider a hypothetical setup with five monthly observations. Suppose you compute USD shares C_1…C_5 from the same exposure definition each month.
Example assumptions (you must adapt to your context):
- USD share values are between 0 and 1.
- You sample once per month at the same time.
- Total exposure includes all currencies you track, with no missing legs.
Then:
- If C_t is stable (e.g., stays near 0.40), standard deviation and average absolute change will be small.
- If C_t drifts upward or oscillates (e.g., 0.20 → 0.55 → 0.30), both volatility measures will be larger.
This approach does not predict movement. It only describes how concentration behaved during the window you selected.
Limitations and risks (what can break the measurement)
Limitation 1: concentration is not “risk” by itself
A volatile concentration measure describes fluctuation in currency mix, not the eventual outcome. The impact depends on your broader exposures, offsets, and how values translate into your base measurement.
Limitation 2: changing hedges and position coverage can mimic volatility
If the USD portion changes because hedges are added/removed, or because you start/stop tracking certain exposures, your concentration series can become more volatile even if underlying economic exposure is stable.
Limitation 3: denominator effects can dominate
If total exposure changes materially over time, the USD share can swing due to denominator movement. Without separately analyzing USD exposure level and total exposure level, you may misinterpret what drove the “concentration volatility.”
Limitation 4: the data source may define “exposure” differently
Different providers, feeds, or reporting systems may compute exposure using different conventions (timing, valuation method, netting rules). Even with the same formula, the inputs can differ, changing C_t and therefore the volatility.
A practical failure mode to look for:
- sudden step-changes that align with operational events (rebalancing, reporting cycles, transfers), not with gradual currency-mix change.