Stochastic Interest Rate Models and Their Impact on Duration and Convexity

In fixed income markets, interest rate risk is a primary concern for bond investors and portfolio managers. Duration and convexity are key metrics used to measure a bond’s sensitivity to interest rate changes. Traditional interest rate models often assume deterministic rates or simple dynamics; however, stochastic interest rate models offer a more realistic representation by allowing interest rates to evolve randomly over time, incorporating uncertainty and mean-reversion properties.

Stochastic interest rate modeling is important for pricing interest rate derivatives, risk management, and constructing hedging strategies that account for the changing structure of interest rate volatility and term dynamics. This article elucidates the mechanics of prominent stochastic interest rate models, quantifies their impact on duration and convexity, and illustrates practical applications.

Overview of Interest Rate Risk Measures

Duration

Duration measures the linear sensitivity of a bond’s price ( P ) to small parallel shifts in the yield curve. For a yield-to-maturity ( y ) or an instantaneous interest rate ( r ), Macaulay Duration ( D ) is:

[ D = -\frac{1}{P} \frac{\partial P}{\partial y} = \frac{\sum_{i=1}^n t_i \cdot CF_i \cdot e^{-y t_i}}{\sum_{i=1}^n CF_i \cdot e^{-y t_i}} ]

Where ( CF_i ) denotes the cash flow at time ( t_i ).

Modified Duration ( D^* ) refines this by accounting for compounded yield changes:*

[ D^* = \frac{D}{1 + y / m} ]*

with ( m ) as the number of compounding periods per year.

Convexity

Convexity captures the curvature in the price-yield relationship and measures the non-linear price change with respect to yield shifts:

[ C = \frac{1}{P} \frac{\partial^2 P}{\partial y^2} = \frac{\sum_{i=1}^n t_i^2 \cdot CF_i \cdot e^{-y t_i}}{\sum_{i=1}^n CF_i \cdot e^{-y t_i}} ]

Accurate assessment of convexity is important for portfolios vulnerable to large rate changes.

Shortcomings of Deterministic Assumptions

Traditional duration and convexity calculations assume yields are deterministic and price sensitivity is to parallel shifts. However, interest rates exhibit stochastic behavior characterized by mean-reversion, volatility clustering, and shifting term structures. This limits the efficacy of static measures in capturing realistic risk profiles, especially for long-dated or embedded option bonds and interest rate derivatives.

Stochastic interest rate models integrate the randomness of interest rate processes, enabling dynamic estimation of risk sensitivities.

Fundamental Stochastic Interest Rate Models

Vasicek Model

The Vasicek model (1977) models the short rate ( r_t ) with mean reversion to a long-run average ( \theta ):

[ dr_t = a(\theta - r_t) dt + \sigma dW_t ]

• ( a ): speed of mean reversion
• ( \theta ): equilibrium rate
• ( \sigma ): volatility of rate changes
• ( W_t ): Wiener process (Brownian motion)

The Vasicek model produces normally distributed rates with a closed-form solution for zero-coupon bond prices:

[ P(t, T) = A(t, T) e^{-B(t, T) r_t} ]

where

[ B(t, T) = \frac{1 - e^{-a(T-t)}}{a}, \quad A(t, T) = \exp\left{ \left(\theta - \frac{\sigma^2}{2 a^2}\right) \left( B(t, T) - (T-t) \right) - \frac{\sigma^2}{4 a} B(t, T)^2 \right} ]

Limitations: Allows negative rates due to normal distribution assumption.

Cox-Ingersoll-Ross (CIR) Model

The CIR model (1985) modifies the Vasicek by making volatility proportional to the square root of the rate, preventing negative rates:

[ dr_t = a(\theta - r_t) dt + \sigma \sqrt{r_t} dW_t ]

Bond prices still have closed forms, but involve noncentral chi-square distributions. The CIR process is mean-reverting and produces strictly positive rates if ( 2 a \theta > \sigma^2 ).

Hull-White Model

The Hull-White model extends Vasicek by allowing time-dependent parameters:

[ dr_t = [ \theta(t) - a r_t ] dt + \sigma dW_t ]

The deterministic function ( \theta(t) ) is chosen to fit the initial term structure exactly. This flexibility helps in calibration to market data.

Impact on Duration and Convexity

Traditional duration and convexity assume deterministic or constant rates, but stochastic interest rates introduce:

1. Randomness in discount factors: Bond prices are no longer deterministic functions of fixed yields; they depend on the stochastic evolution of ( r_t ).

2. Term structure dynamics: Sensitivities vary with the state and path of interest rates, affecting risk management.

3. Volatility of risings and fallings: The distribution of future rates impacts the convexity and embedded optionality.

Redefining Duration and Convexity in Stochastic Frameworks

Bond price ( P(t,r_t) ) for zero-coupon bond with maturity ( T ) is:

[ P(t, r_t) = \mathbb{E}^{\mathbb{Q}} \left[ e^{-\int_t^T r_s ds} \mid \mathcal{F}t \right] ]

Here, ( \mathbb{Q} ) denotes the risk-neutral measure.

The sensitivity of ( P ) to changes in the short rate ( r_t ) can be expressed as:

• Short-Rate Duration:

[ D_{sr} = -\frac{1}{P} \frac{\partial P}{\partial r_t} ]_

• Short-Rate Convexity:

[ C_{sr} = \frac{1}{P} \frac{\partial^2 P}{\partial r_t^2} ]_

Example: Vasicek Model Duration & Convexity

For zero-coupon bond price in Vasicek:

[ P(t, T) = A(t, T) e^{- B(t, T) r_t } ]

Differentiating w.r.t. ( r_t ):

[ \frac{\partial P}{\partial r_t} = -B(t, T) A(t, T) e^{- B(t, T) r_t } = -B(t, T) P(t, T) ]

Second derivative:

[ \frac{\partial^2 P}{\partial r_t^2} = B(t, T)^2 P(t, T) ]

Therefore,

[ D_{sr} = B(t, T), \quad C_{sr} = B(t, T)^2 ]

These expressions differ from classical duration since they relate to changes in the underlying short rate, embedding stochastic properties.

Duration and Convexity for Coupon-Bearing Bonds

Coupon bonds can be viewed as portfolios of zero-coupon bonds. Given ( n ) cash flows ( CF_i ) at ( t_i ), price under the Vasicek model is:

[ P(t, r_t) = \sum_{i=1}^n CF_i \cdot P(t, t_i; r_t) ]_

Duration and convexity are weighted sums over zero coupon components, where:

[ D_{sr} = -\frac{1}{P} \sum_{i=1}^n CF_i \frac{\partial P(t, t_i; r_t)}{\partial r_t}, \quad C_{sr} = \frac{1}{P} \sum_{i=1}^n CF_i \frac{\partial^2 P(t, t_i; r_t)}{\partial r_t^2} ]

Implications for Hedging Strategies

Hedging Interest Rate Exposure

A hedge constructed based on classical duration and convexity may underestimate or overestimate interest rate risk if the stochastic nature of rates and their volatilities are ignored. Using stochastic models:

• One can adjust hedge ratios dynamically according to the current state of ( r_t ).
• Incorporate volatility smiles and convexities observed in option-implied volatility.
• Manage the risk of nonlinearities introduced by stochastic paths more effectively.

Example: Hedging a Portfolio with Hull-White Dynamics

Assume a portfolio sensitive to ( r_t ), modeled by Hull-White:

• Use model parameters ( a = 0.1, \sigma = 0.01 )
• From current rate ( r_t = 0.03 ), compute zero-coupon bond prices and their derivatives
• Calculate ( D_{sr} ) and ( C_{sr} ) per bond constituent
• Determine hedge positions in futures or swaps that adjust as ( r_t ) varies

Dynamic hedging involves recalculating ( D_{sr} ) and ( C_{sr} ) over time and adjusting exposure—a significant improvement over static hedges.

Multi-Factor Models and Their Effects

Single-factor models like Vasicek focus only on the short rate. Multi-factor models (e.g., two-factor Hull-White) introduce more sources of randomness, modeling both short rate and long-term components separately:

[ dr_t = (\theta(t) - a r_t - b x_t) dt + \sigma dW_t^{(1)} ] [ dx_t = -c x_t dt + \eta dW_t^{(2)} ]

with ( W_t^{(1)} ) and ( W_t^{(2)} ) correlated Brownian motions.

This adds depth to duration and convexity analysis:

• Duration becomes vector-valued, representing sensitivities to multiple factors.

[ D = \left( -\frac{1}{P} \frac{\partial P}{\partial r_t}, - \frac{1}{P} \frac{\partial P}{\partial x_t} \right) ]

• Convexity is a matrix of second derivatives (Hessian), capturing cross sensitivities:

[ C = \frac{1}{P} \begin{bmatrix} \frac{\partial^2 P}{\partial r_t^2} & \frac{\partial^2 P}{\partial r_t \partial x_t} \ \frac{\partial^2 P}{\partial x_t \partial r_t} & \frac{\partial^2 P}{\partial x_t^2} \end{bmatrix} ]

This multidimensional sensitivity framework enables more precise risk attribution and hedging in complex interest rate environments.

Practical Example: Impact on Portfolio Duration

Consider a 5-year fixed coupon bond with annual coupon ( C = 5% ), face value 100, price ( P ) derived under the Vasicek model with parameters:

• ( a = 0.1 )
• ( \theta = 0.05 )
• ( \sigma = 0.01 )
• current short rate ( r_t = 0.03 )

Calculate zero-coupon bond prices ( P(t, t_i; r_t) ) for ( i=1,\ldots,5 ) using Vasicek ( A(t, t_i) ) and ( B(t, t_i) ):

[ B(t, t_i) = \frac{1 - e^{-a(t_i - t)}}{a}, \quad A(t, t_i) = \exp\left( \left( \theta - \frac{\sigma^2}{2 a^2} \right) \left( B(t, t_i) - (t_i - t) \right) - \frac{\sigma^2}{4a} B(t, t_i)^2 \right) ]

Compute:

[ P(t, t_i; r_t) = A(t, t_i) e^{-B(t, t_i) r_t} ]

Sum cash flows discounted by these stochastic zero curves:

[ P = \sum_{i=1}^5 5 \times P(t, t_i; r_t) + 100 \times P(t, t_5; r_t) ]_

Duration estimate (sensitivity to ( r_t )):

[ D_{sr} = \frac{1}{P} \sum_{i=1}^5 CF_i \cdot B(t, t_i) \cdot P(t, t_i; r_t) ]

Compared to classical duration ( D = \sum t_i \times PV(CF_i)/P ), this duration reflects the rate stochasticity and mean reversion.

Summary and Recommendations

• Stochastic interest rate models, such as Vasicek, CIR, and Hull-White, provide superior frameworks for modeling interest rate evolution, incorporating randomness and mean reversion.

• These models affect the calculation and interpretation of duration and convexity, shifting from static derivative measures with respect to yield to dynamic sensitivities with respect to the instantaneous rate or multiple factors.

• Practical use involves differentiating model-derived bond prices with respect to state variables, enhancing hedging and risk management strategies.

• Multi-factor models increase realism but complicate sensitivity analysis, demanding advanced hedging techniques based on vector durations and convexities.

• Traders and risk managers must adjust duration and convexity metrics dynamically, according to the stochastic process parameters, rather than relying solely on classical calculations.

References for In-Depth Study

1. Brigo, D., & Mercurio, F. (2006). Interest Rate Models—Theory and Practice. Springer Finance.
2. Hull, J. (2018). Options, Futures, and Other Derivatives (10th Edition). Pearson.
3. Vasicek, O. (1977). An equilibrium characterization of the term structure. Journal of Financial Economics, 5(2), 177-188.
4. Cox, J.C., Ingersoll, J.E., & Ross, S.A. (1985). A Theory of the Term Structure of Interest Rates. Econometrica, 53(2), 385-407.
5. Hull-White, J.C. (1990). Pricing Interest-Rate-Derivative Securities. The Review of Financial Studies, 3(4), 573-592.

Understanding the precise effects of stochasticity on duration and convexity equips traders with the ability to model interest rate risk with higher fidelity, aligning risk metrics more closely with market realities.