Term Structure of Interest Rates Models
Interactive simulation of the stochastic differential equation (SDE) dynamics of the Vasicek model and the CIR model, providing an intuitive comparison of the characteristics of these two classic interest rate models.
The Vasicek model (1977) is one of the earliest term structure of interest rate models. It assumes that interest rates have a mean-reverting property—interest rates fluctuate around a long-term mean \(\theta\), and the larger the deviation, the stronger the restoring force. The diffusion term is the constant \(\sigma\), meaning that interest rate volatility is not affected by the current interest rate level.
- \(\kappa\) Mean reversion speed: the rate of reversion to \(\theta\)
- \(\theta\) Long-term mean: the long-term equilibrium level of the interest rate
- \(\sigma\) Volatility: the strength of random shocks
- \(r_0\) Initial interest rate: the starting interest rate value for the simulation
The CIR model (Cox-Ingersoll-Ross, 1985) introduces a square-root diffusion term on the basis of Vasicek \(\sigma\sqrt{r_t}\). When the interest rate approaches zero, the volatility automatically decreases, thereby ensuring that the interest rate is always non-negative (strictly positive when the Feller condition is satisfied), which is more consistent with actual interest rate behavior.
- \(\kappa\) Mean reversion speed: same as Vasicek
- \(\theta\) Long-term mean: same as Vasicek
- \(\sigma\) Volatility parameter: multiplied by \(\sqrt{r_t}\)
- \(r_0\) Initial interest rate: must be non-negative
Plain-Language Introduction
Interest rates will not rise forever, nor will they fall forever. They are like a ball tied to a rubber band—mean reversion is the most core feature of interest rates. The Vasicek model and the CIR model both use mathematical language to describe this "reversion to the mean" behavior: when the interest rate is above the normal level (the long-term mean θ), there is a force pulling it back; when the interest rate is below the normal level, it is likewise pulled upward. κ controls the strength of this force, and σ controls the magnitude of random fluctuations.
The key difference between the two lies in the structure of volatility: Vasicek assumes that volatility is equally large regardless of whether interest rates are high or low—this may cause interest rates to become negative (which is rarely seen in reality). CIR instead makes volatility proportional to the square root of the interest rate—the lower the interest rate, the smaller the volatility, thereby cleverly "locking" the interest rate in a non-negative range and making it closer to the real world.
What Problem Does It Solve?
Before interest rate models appeared, the financial field lacked a unified mathematical framework to describe the stochastic evolution of interest rates. These two models solved:
- Interest rate modeling: providing a continuous stochastic process description for interest rates
- Bond pricing: allowing analytical solutions for zero-coupon bond prices to be derived
- Risk measurement: quantifying interest rate risk (duration, convexity, etc.)
- Derivatives valuation: providing a foundation for pricing interest rate derivatives (options, swaps, etc.)
Why Is It Important?
Interest rates are the "universal gravitation" of the financial system—they affect almost all financial instruments, including bonds, loans, mortgages, and derivatives. The Vasicek and CIR models:
- Are milestones in term structure theory and pioneered the no-arbitrage interest rate modeling paradigm
- The CIR model is a representative of the affine model family and remains a core tool for central banks and financial institutions today
- Provide a theoretical foundation for understanding the shape and dynamics of the yield curve
- Are the basis for more complex models (such as Hull-White and the HJM framework)
Application Scenarios
- Bond pricing: calculating the theoretical prices of various fixed-income securities
- Interest rate risk management: asset-liability management (ALM) for banks and insurance companies
- Monetary policy analysis: central banks simulate interest rate paths and evaluate policy effects
- Derivatives valuation: pricing interest rate options, interest rate swaps, and cap/floor contracts
- Scenario analysis: generating interest rate scenarios in stress testing
- Academic research: empirical testing of term structure theory
Deficiencies and Limitations
Vasicek model:
- Interest rates may become negative (in a low interest rate environment this actually became an "advantage," but historically it was regarded as a flaw)
- The constant volatility assumption is inconsistent with market observations (actual volatility changes with the interest rate level)
- It has only one stochastic factor and cannot capture complex deformations of the yield curve
CIR model:
- In discrete simulation it may still produce negative values (a known issue with Euler discretization)
- Square-root diffusion limits the flexibility of volatility and may underestimate volatility when interest rates are high
- It is also a single-factor model and cannot fit an arbitrarily shaped initial yield curve
- Parameter estimation is more complex than for Vasicek