The Vasicek model has one structural flaw for a floor-trading desk: it assigns a positive probability to negative rates. For most interest-rate derivatives, this barely matters. But for options struck near zero — floors on overnight rates, corridor trades, barrier swaptions — the ability to rule out negative rates analytically changes the pricing by a meaningful amount. The Cox-Ingersoll-Ross (CIR) model (1985) addresses this by replacing Vasicek's constant diffusion coefficient with . The square-root diffusion naturally collapses toward zero as , preventing from going negative under the Feller condition. The cost is a Riccati ODE for the bond price coefficient rather than the linear ODE in Vasicek, and bond option pricing requires non-central chi-squared quantiles rather than Gaussian ones — but both are still computable in closed form.
Learning objectives. By the end of this note you should be able to:
- write down the CIR SDE and explain why makes the model non-negative;
- state the Feller condition and explain what happens when it fails;
- give the conditional distribution of given (non-central chi-squared, rescaled);
- state the CIR bond price formula with explicit and ;
- compare Vasicek and CIR along the key dimensions: rate non-negativity, volatility structure, analytical complexity;
- identify the CIR model as a special case of the affine term-structure framework.
In this lesson
-
- The square-root diffusion SDE
-
- The Feller condition: when does stay strictly positive?
-
- Distribution of the short rate
-
- Affine bond price formula
-
- Vasicek versus CIR: comparison
-
- Bond options under CIR
Before you start
- The Vasicek Model: Mean Reversion, Affine Bond Prices, and the Gaussian Distribution
- Derivation — Vasicek Bond Price via the Term-Structure PDE and Affine Riccati ODEs
The full lesson, with worked examples and exercise solutions, is for members.
Read the full lessonNew to SelfMadeQuant.com? See membership options or browse the Fixed Income and Interest-Rate Models syllabus.