Industry: Banking and Financial Services
Product: ModelRisk
Application: Currency Risk Management

A 64-Basis-Point Drag: Sizing FX Hedges on a $4.2B Multi-Currency Operating Book

EUR/USD moved 15.4% from peak to trough in 2022. A US-headquartered industrial group with $4.2B of recurring revenue split across EUR (34%), GBP (12%), JPY (8%), CAD (6%) and a long tail of EM currencies booked a translated-revenue swing of $310M over the year — none of which had anything to do with operating performance. The CFO's question was simple: what is the right hedge ratio per currency, and what does the residual P&L volatility look like once we have hedged optimally? The deterministic answer ("hedge 50% of forecast exposure") gave one number; it didn't say what the residual was, how often the residual would be larger than the unhedged risk on a single-currency basis, or what the optimum was.

The treasury team replaced the deterministic policy with a Vose Software ModelRisk simulation that priced stochastic FX, stochastic exposures and the cost of carry on the hedge leg jointly.

Stochastic FX, stochastic exposures, real correlations

• FX returns — log returns for each of the seven major currencies modeled as a multivariate Normal for the body with a t-copula (df = 5) in the tail, calibrated to 15 years of monthly data. Annualized vols: EUR/USD 9.2%, GBP/USD 10.4%, JPY/USD 11.8%, CAD/USD 7.6%, AUD/USD 12.9%, BRL/USD 18.5%, MXN/USD 13.2%. EUR/GBP correlation 0.66, EUR/CHF 0.91, EM-currency correlations between 0.35 and 0.55 — the diversification benefit shrinks fast under stress.
• Forecast exposures — quarterly exposure per currency as a LogNormal with a forecast-coefficient of variation of 12% over a one-year horizon. This is the variable most FX risk models ignore: the quantity of foreign currency is itself uncertain.
• Hedge cost — forward points modeled from a stochastic interest-rate differential (USD 10y at 4.6% mean reverting; EUR 2.8%; JPY 0.9%; BRL 11.4%) feeding into covered interest parity. The carry cost of being long USD vs JPY is negative 380 bps annualized; vs BRL it is positive 680 bps. A hedge program that ignores carry is hedging at random.

From an exposure ladder to a P&L distribution

The deterministic dashboard showed a single number: $4.2B of FX exposure translated at spot. The Monte Carlo simulation turned that stock figure into a flow: a one-year translated-USD P&L distribution under the firm's existing 30% blanket hedge.

Mean P&L impact: -$3M (carry cost net of expected FX drift). One-year 95% VaR: $259M. 99% VaR: $383M. Probability of an FX-driven P&L impact worse than -$100M: 26%. The deterministic measure - "exposure of $4.2B" - is a stock; the P&L impact is a flow with a heavy left tail driven mostly by EUR and GBP joint moves.

What actually drives the residual

The largest driver of the 95% VaR is the EUR-GBP joint move (the two correlated DM exposures move together; together they account for $48M of the VaR). Stochastic exposure uncertainty is second ($31M); the EUR-USD volatility level is third ($26M); the t-copula tail-dependence parameter is fourth ($21M). The currency the treasury team had been most worried about — BRL — comes seventh, because the exposure is only $84M even though the volatility is the highest of all seven. Variance contribution is exposure × vol, not vol alone.

Per-currency hedge optimization

Holding total hedge spend constant, the team optimized hedge ratios per currency to minimize 95% VaR.

The optimum is not 100% on every leg. The carry-corrected variance contribution says:

• EUR: hedge 78% (was 30%) — high exposure × high correlation with GBP, low carry cost
• GBP: hedge 72% (was 30%) — clusters with EUR
• JPY: hedge 25% (was 30%) — negative carry (-380 bps) eats most of the variance-reduction benefit
• BRL: hedge 12% (was 30%) — carry cost of 680 bps annualized is more expensive than the volatility it protects against
• MXN, AUD, CAD: 40-55%

The 95% VaR drops from $259M to $137M (-47%). Concentrating the hedge on the high-variance DM legs (EUR, GBP) while pulling JPY and BRL back costs roughly $5M more in annualized carry than the blanket policy — a deliberate trade of a small carry premium for a large reduction in tail risk.

What the model changed

• Per-currency hedge ratios replaced the 30% blanket policy, cutting 95% VaR by 47% ($259M → $137M) for roughly $5M of additional annualized carry — tail risk bought down far faster than carry rose.
• The BRL hedge program was cut to 12%; the 680 bps carry cost exceeded the residual variance-reduction benefit by a wide margin.
• A natural-hedge revenue-mix policy was introduced for the EM tail: invoice in USD where the customer can absorb it; this cut tail FX exposure by an additional $22M at zero hedge cost.
• Earnings guidance moved from a single FX assumption to a $90M/$140M one-year FX uncertainty band quoted alongside the operating forecast.

ModelRisk Functionality Used

• t-copula (df = 5) on multi-currency log returns that captured the joint-tail correlation EUR-GBP-CHF observed in 2022, raising the 95% VaR estimate by 19% versus an independence-assumption baseline.
• Carry-aware hedge optimization that priced JPY and BRL forwards correctly under covered interest parity and showed the BRL hedge was paying a 680 bps carry premium for negligible variance benefit — so its ratio was cut to 12%.
• Stochastic exposure modeling (LogNormal CV = 12%) that surfaced exposure-quantity uncertainty as the second-largest VaR driver — bigger than EUR/USD volatility itself.
• Hedge-ratio sweep with VaR objective that produced the per-currency optimum (EUR 78%, GBP 72%, JPY 25%, BRL 12%) instead of a single blanket policy ratio.
• Output distributions for earnings guidance translating "we are exposed to FX" into a $90M-$140M one-year P&L band, suitable for an investor-relations conversation.

A blanket hedge ratio averages over differences that matter: Monte Carlo on multi-currency exposures shows the optimum is per-currency, carry-aware, and never the same number on every leg.