FRM Part II · FRM Exam Part II · The Art of Term Structure Models: Volatility and Distribution
In a normal model with dr = λ dt + σ dw, the short rate is 4.00%, and annual basis-point volatility is 120 bps. A trader wants to rescale to a lognormal model that gives the same volatility at the current rate level. What percentage (lognormal) volatility should be used?
The equivalent lognormal volatility is 30%. Normal volatility of 1.20% divided by the current rate of 4.00% gives 0.30, because lognormal volatility times the rate level must equal the absolute basis-point volatility at that point.
- A3.0%
- B30.0%Correct
- C4.8%
- D12.0%
Explanation
Lognormal volatility times the rate equals the basis-point volatility: σ_LN × 4% = 1.20%. So σ_LN = 1.20/4.00 = 30%. Using 120/4 without unit conversion gives 30 expressed wrongly, and multiplying 1.2% by 4 gives 4.8%, which is the wrong operation.
Did you get it right without looking?
One question tells you little. A timed set on The Art of Term Structure Models: Volatility and Distribution shows your real accuracy, how long you take and where you lose marks.
More The Art of Term Structure Models: Volatility and Distribution questions
- In the Vasicek model dr = k(θ − r)dt + σdw, a risk analyst notes that the short rate is currently well above θ. Which statement best describ…
- A risk analyst compares a normal (Ho-Lee type) short-rate model with a lognormal short-rate model in a low-rate environment. Which statement…
- In the Vasicek model dr = k(θ − r)dt + σdW, a risk analyst notes that the short rate is currently well above θ. Which statement best describ…
- A CIR model has k = 0.30, θ = 5%, and σ = 0.20 (decimal rates). A risk manager checks the Feller condition, 2kθ ≥ σ². What is the result and…
- The Salomon Brothers (lognormal) model in the reading is one in which the short rate's volatility is proportional to the rate, and a practit…
- A desk calibrates a Hull-White model by fitting σ(t) to cap volatilities. Compared with a constant-σ Model 3 calibrated to the same long-dat…