Fact-checked by Grok 2 weeks ago

Short-rate model

A short-rate model is a mathematical framework in financial modeling that describes the stochastic evolution of the instantaneous short interest rate, denoted as r(t), which represents the spot rate for an infinitesimally short period and serves as the state variable for the term structure of interest rates.^[1] These models specify the dynamics of r(t) typically via a stochastic differential equation of the form dr(t) = \mu(t, r(t)) \, dt + \sigma(t, r(t)) \, dW(t), where \mu is the drift, \sigma is the volatility, and W(t) is a Wiener process, enabling the simulation of future interest rate paths under risk-neutral measures.^[1] They are fundamental in quantitative finance for pricing interest rate derivatives, such as bonds, options, caps, floors, and swaptions, by ensuring consistency with observed market prices of zero-coupon bonds.^[2]^[1] Short-rate models emerged in the late 1970s as part of the broader development of term structure modeling, with early contributions focusing on equilibrium-based approaches that incorporate economic factors like mean reversion to reflect real-world interest rate behavior.^[3] They are classified primarily into one-factor and multi-factor variants; one-factor models assume a single source of randomness driving the short rate, simplifying computations but potentially limiting their ability to capture correlations across the yield curve, while multi-factor models introduce additional stochastic factors to better replicate complex market dynamics.^[2]^[1] Prominent examples include the Vasicek model (1977), which features constant mean reversion and Gaussian dynamics but allows negative rates; the Cox-Ingersoll-Ross (CIR) model (1985), an affine model with square-root volatility that ensures non-negative rates under certain parameter conditions; and the Hull-White model, a flexible extension of Vasicek with time-dependent parameters for exact calibration to the initial term structure.^[1]^[3] These models are calibrated to current market data, such as yield curves and volatilities, and are widely applied in risk management for instruments like mortgages and credit derivatives, though they face challenges in accurately forecasting long-term rates or handling stochastic volatility.^[2]^[3]

Core Concepts

Definition and Role of the Short Rate

In interest rate modeling, the short rate r_t is defined as the instantaneous spot interest rate at time t, which represents the interest rate applicable to borrowing or lending funds over an infinitesimally short period starting at that time.^[4] This rate serves as the fundamental building block for describing the dynamics of interest rates in continuous-time frameworks, capturing the risk-free return over negligible time intervals.^[1] The concept of the short rate emerged in continuous-time finance through seminal works that applied stochastic calculus to interest rate processes, most notably Vasicek (1977), who introduced an equilibrium model where the short rate evolves stochastically to characterize the term structure.^[5] This approach shifted focus from deterministic rates to probabilistic models, enabling the analysis of uncertainty in interest rate movements.^[6] The short rate plays a central role in the term structure of interest rates, forming the basis for deriving the yield curve. Specifically, the instantaneous forward rate f(t,T), which indicates the rate agreed at time t for a loan starting at T, equals the expected future short rate under the risk-neutral measure:
f(t,T) = \mathbb{E}^Q [r_T \mid \mathcal{F}_t].
This relationship highlights how expectations of short rate paths underpin longer-term rates, ensuring consistency across maturities.^[7] A key application of the short rate is in bond pricing under no-arbitrage conditions. The price P(t,T) of a zero-coupon bond paying 1 at maturity T, observed at time t, is given by the risk-neutral expectation of its continuously discounted payoff:
P(t,T) = \mathbb{E}^Q \left[ \exp\left( -\int_t^T r_s \, ds \right) \mid \mathcal{F}_t \right].
This formula derives from the fundamental theorem of asset pricing in continuous time: under the risk-neutral measure \mathbb{Q}, where all assets earn the instantaneous risk-free return r_t, the value of any payoff is its expected value discounted along the short rate path. The integral \int_t^T r_s \, ds accumulates the cumulative interest over the period, reflecting the total discounting factor for paths from t to T, conditional on the filtration \mathcal{F}_t up to time t.^[8] Short-rate models presuppose this risk-neutral pricing framework and no-arbitrage principles to ensure model consistency with observed market prices.^[9]

Mathematical Framework of Short-Rate Models

Short-rate models describe the dynamics of the instantaneous short rate r_t through a stochastic differential equation (SDE) under the physical probability measure:

dr_t = \mu(t, r_t) \, dt + \sigma(t, r_t) \, dW_t,

where \mu(t, r_t) represents the drift term capturing the expected change in the short rate, \sigma(t, r_t) denotes the instantaneous volatility, and W_t is a standard Brownian motion.^[5] This formulation allows for mean-reverting behavior or other dynamics depending on the functional forms of \mu and \sigma. To price interest rate derivatives, the model is typically analyzed under the risk-neutral measure \mathbb{Q}, obtained via Girsanov's theorem by adjusting for the market price of risk \lambda(t, r_t). Under \mathbb{Q}, the SDE becomes

dr_t = \bigl( \mu(t, r_t) - \lambda(t, r_t) \sigma(t, r_t) \bigr) dt + \sigma(t, r_t) \, dW_t^\mathbb{Q},

where W_t^\mathbb{Q} is a Brownian motion under \mathbb{Q}. This change ensures that discounted asset prices are martingales, enabling arbitrage-free pricing.^[5] The price P(t, T; r_t) of a zero-coupon bond maturing at time T with face value 1 is given by the risk-neutral expectation

P(t, T; r_t) = \mathbb{E}^\mathbb{Q} \left[ \exp\left( -\int_t^T r_s \, ds \right) \Big| \mathcal{F}_t \right],

conditional on the filtration \mathcal{F}_t up to time t. By the Feynman-Kac theorem, this expectation satisfies the partial differential equation (PDE)

\frac{\partial P}{\partial t} + \bigl( \mu(t, r_t) - \lambda(t, r_t) \sigma(t, r_t) \bigr) \frac{\partial P}{\partial r} + \frac{1}{2} \sigma^2(t, r_t) \frac{\partial^2 P}{\partial r^2} - r_t P = 0,

subject to the terminal boundary condition P(T, T; r_T) = 1. This PDE arises from applying Itô's lemma to the bond price process and imposing the no-arbitrage condition that the discounted bond price is a \mathbb{Q}-martingale. Solving this PDE yields bond prices and, by extension, the entire term structure of interest rates.^[5]^[10] A significant class of short-rate models features an affine term structure, where bond prices take the exponential-affine form

P(t, T; r_t) = \exp\bigl( A(t, T) - B(t, T) r_t \bigr).

This structure emerges when the drift and squared volatility are affine functions of r_t, specifically \mu(t, r_t) - \lambda(t, r_t) \sigma(t, r_t) = \delta_0(t) + \delta_1(t) r_t and \sigma^2(t, r_t) = \alpha_0(t) + \alpha_1(t) r_t, under the risk-neutral measure. Substituting this form into the bond pricing PDE reduces it to a system of ordinary differential equations (ODEs) for A(t, T) and B(t, T):

\frac{\partial B}{\partial t} = -\delta_1(t) B + \frac{1}{2} \alpha_1(t) B^2 - 1, \quad B(T, T) = 0,

\frac{\partial A}{\partial t} = \delta_0(t) B - \frac{1}{2} \alpha_0(t) B^2, \quad A(T, T) = 0.

The equation for the coefficient of r_t is satisfied by the ODE for B, ensuring consistency. These ODEs often admit closed-form solutions, facilitating efficient computation of yields and derivative prices.^[11] Short-rate models are classified as endogenous or exogenous based on their handling of the initial yield curve. Exogenous models, such as the Hull-White model, incorporate time-dependent parameters in the drift and volatility to fit the observed initial term structure exactly by construction, ensuring no-arbitrage consistency from the outset. In contrast, endogenous models, like the Vasicek model, specify time-homogeneous parameters that generate an implied term structure, which generally does not match the market curve and thus requires calibration through parameter adjustment or additional procedures.^[6]^[10]^[5]

Model Classifications

One-Factor Short-Rate Models

One-factor short-rate models describe the evolution of the instantaneous interest rate r_t using a single stochastic factor, typically governed by a stochastic differential equation (SDE) under the risk-neutral measure. These models originated in the early 1970s with Robert Merton's formulation of a normal (Gaussian) process for the short rate, dr_t = \alpha dt + \sigma dW_t, where \alpha is a constant drift and W_t is a Wiener process, allowing for analytical bond pricing but lacking mean reversion to reflect long-term stability.^[12] This approach evolved in the late 1970s and 1980s toward mean-reverting dynamics, driven by empirical observations of interest rate behavior during periods of high volatility, such as those following the 1970s economic turbulence, where rates tended to revert to a long-term equilibrium rather than following pure diffusion. Seminal contributions include the Vasicek, Cox-Ingersoll-Ross (CIR), Hull-White, and Black-Derman-Toy (BDT) models, each introducing refinements to address limitations like negative rates or term structure fitting while maintaining tractable solutions for derivative pricing. The Vasicek model, proposed in 1977, represents a foundational mean-reverting framework, modeling the short rate as an Ornstein-Uhlenbeck process:

dr_t = \kappa (\theta - r_t) dt + \sigma dW_t,

where \kappa > 0 is the speed of mean reversion, \theta is the long-term mean, and \sigma > 0 is the volatility.^[5] This Gaussian model admits closed-form solutions for zero-coupon bond prices, expressed as P(t,T) = \exp[A(t,T) - B(t,T) r_t], with explicit functions B(t,T) = \frac{1 - e^{-\kappa (T-t)}}{\kappa} and A(t,T) involving integrals of the forward rate and volatility terms, enabling efficient computation of yields and options.^[5] Its simplicity facilitates calibration and simulation, making it suitable for introductory applications in fixed-income derivatives, though it permits negative rates—a drawback in low-rate environments—as the normal diffusion lacks boundary constraints. Building on Vasicek's structure, the CIR model of 1985 incorporates a square-root diffusion to ensure non-negativity of rates, addressing a key empirical feature:

dr_t = \kappa (\theta - r_t) dt + \sigma \sqrt{r_t} dW_t.

This SDE, derived from an intertemporal equilibrium framework, guarantees r_t \geq 0 when the Feller condition $2\kappa\theta > \sigma^2 holds, preventing negative rates while retaining mean reversion.^[13] Bond prices lack fully explicit affine forms but can be computed via the non-central chi-squared distribution, with the price P(t,T) involving modified Bessel functions and parameters reflecting the stochastic volatility proportional to the square root of the rate level.^[14] The model's square-root volatility captures humped term structures observed in data, enhancing realism for pricing interest rate caps and floors, though it requires numerical methods for calibration due to the non-Gaussian nature.^[13] The Hull-White model, introduced in 1990, extends the Vasicek framework with time-dependent parameters to fit the observed initial yield curve exactly:

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

where a is constant mean reversion speed, and \theta(t), \sigma(t) are chosen to match market data.^[10] Retaining the Gaussian structure, it yields closed-form bond prices analogous to Vasicek's, P(t,T) = \exp[A(t,T) - B(t,T) r_t], with time-varying B(t,T) = \frac{1 - e^{-a (T-t)}}{a} and A(t,T) incorporating integrals of \theta and \sigma.^[10] This flexibility allows arbitrage-free pricing of European options on bonds via Black's formula adjusted for the model's dynamics, making it widely adopted in practice for its balance of tractability and market consistency, despite the potential for negative rates.^[15] Also from 1990, the Black-Derman-Toy (BDT) model adopts a lognormal process for the short rate, specified under the discrete-time binomial framework but approximable continuously as d \ln r_t = [\mu(t) - \frac{1}{2} \sigma^2(t)] dt + \sigma(t) dW_t, ensuring positivity and lognormal distribution.^[16] Implemented via a recombining binomial tree calibrated to the yield curve and volatility structure, it prices bonds and options by backward induction, with node values reflecting risk-neutral probabilities.^[17] The model's time-varying drift and volatility enable fitting of the entire term structure and implied volatilities, facilitating applications to Treasury bond options, though the discrete nature increases computational demands compared to continuous affine models.^[18] Within one-factor models, normal (Gaussian) processes like Vasicek and Hull-White allow unbounded rates, including negatives, which aligns with empirical low-rate regimes but fails to capture volatility smiles in option prices, as the symmetric diffusion produces flat implied volatilities.^[19] In contrast, lognormal models such as BDT prevent negative rates through multiplicative noise but risk explosive behavior—rapid rate increases to infinity—under high volatility, as seen in regimes where the short rate's log-process leads to fat-tailed distributions; this can generate skews or smiles in caplet implied volatilities, better matching market data during volatile periods.^[20] These differences highlight a trade-off: Gaussian models offer simplicity and closed forms but limited smile dynamics, while lognormal variants enhance positivity and smile fitting at the cost of potential instability and numerical complexity.^[20]

Multi-Factor Short-Rate Models

Multi-factor short-rate models address the shortcomings of one-factor models, which struggle to replicate observed yield curve dynamics such as twists and humps, by incorporating multiple stochastic factors that separately influence short-term and long-term interest rate movements.^[21] These models typically feature two or more state variables, allowing for more realistic representations of term structure variations, including non-parallel shifts and changes in curvature.^[22] A seminal example is the two-factor model developed by Longstaff and Schwartz in 1992, which combines a mean-reverting short rate with stochastic volatility of changes in the short rate. The dynamics are governed by the following stochastic differential equations:

dr_t = \kappa (\theta - r_t) \, [dt](/page/DT) + \sqrt{v_t} \, [dW](/page/DW)_{1t},

dv_t = \lambda (\gamma - v_t) \, [dt](/page/DT) + \eta \sqrt{v_t} \, [dW](/page/DW)_{2t},

where the Wiener processes W_{1t} and W_{2t} are correlated with correlation \rho.^[21] Bond prices in this framework can be derived semi-analytically by solving the associated partial differential equation, facilitating pricing of interest rate derivatives while capturing volatility clustering.^[23] The Chen model from 1996 extends this to a three-factor framework, modeling the short rate level, slope, and curvature through a system of non-linear stochastic differential equations:

dr_t = \kappa (\theta - r_t) \, dt + \sigma_r \, dW_{rt},

ds_t = \lambda (\mu_s - s_t) \, dt + \sigma_s \, dW_{st},

along with a third equation for the curvature factor, where the Wiener processes W_r, W_s (and the third) are correlated via a specified correlation matrix.^[24] This setup allows the model to fit empirical yield curve shapes more accurately, including humps, by disentangling distinct sources of risk.^[25] In general, multi-factor affine term structure models posit a state vector X_t following

dX_t = K (\theta - X_t) \, dt + \Sigma \, dW_t,

where K is the speed-of-adjustment matrix, \theta the long-run mean, and \Sigma the volatility matrix, with the short rate affine in X_t.^[11] Bond prices are then expressed using matrix exponentials through the solution to Riccati equations, though obtaining closed-form solutions becomes challenging for higher factor counts, often requiring numerical approximations.^[26] Historically, the development of multi-factor models gained momentum in the post-1990s era, driven by the need for better empirical fit to market data, as exemplified by the canonical representations introduced by Dai and Singleton in 2000, which systematize affine models into up to three-factor forms for enhanced tractability and testing.^[26]

Practical Aspects

Calibration Techniques

Calibration techniques for short-rate models involve estimating parameters by fitting the model's implied prices for bonds, options, or other derivatives to observed market data, ensuring the model accurately reproduces the current term structure of interest rates. This process is crucial for practical applications, as it aligns the theoretical dynamics of the short rate with empirical yield curves and volatility surfaces. Typically, calibration begins with adjusting drift parameters to match the initial yield curve, followed by tuning volatility parameters to fit option prices like caps, floors, or swaptions. Least-squares methods are commonly used to minimize the squared differences between model-implied zero rates or bond prices and their market counterparts, providing a straightforward approach for initial curve fitting. In the Hull-White model, the time-dependent drift function \theta(t) is determined to exactly fit the observed term structure, often via spline interpolation of the forward rates derived from market zero-coupon bonds. For instance, cubic splines can interpolate the instantaneous forward rate curve f(0,t), allowing \theta(t) to be solved analytically as \theta(t) = \frac{\partial}{\partial t} f(0,t) + \kappa f(0,t), where \kappa is the mean-reversion speed. This method ensures zero arbitrage with the initial curve while keeping computational costs low.^[27] Maximum likelihood estimation (MLE) estimates parameters under the physical measure using historical time series of short rates or yields, maximizing the likelihood of observing the data given the model's stochastic differential equation. For the Vasicek model, dr_t = \kappa (\theta - r_t) dt + \sigma dW_t, MLE jointly estimates the speed \kappa, long-term mean \theta, and volatility \sigma by assuming discretized Gaussian increments and solving the resulting optimization problem, often via numerical methods like Newton-Raphson. To shift to the risk-neutral measure for pricing, Girsanov's theorem adjusts the drift by the market price of risk, transforming physical parameters into risk-neutral ones. This approach is particularly useful for capturing long-term dynamics from historical data.^[28] For volatility parameters, calibration often targets implied volatilities from swaptions and caplets, which provide market views on future rate movements. In the Black-Derman-Toy (BDT) model, a binomial tree is constructed where node probabilities and short rates are adjusted iteratively to match the term structure and the Black volatilities of caplets or swaptions. For example, starting from the root, the tree's volatility structure is scaled at each step to reproduce observed caplet prices, ensuring the model prices European options consistently with market quotes. This tree-based adjustment allows for flexible lognormal dynamics and is widely implemented in practice for Bermudan option valuation.^[29]^[30] Recent advances since 2020 have incorporated machine learning, particularly neural networks, for non-parametric calibration, enabling faster and more robust fits to complex yield curves without assuming specific functional forms. For instance, deep neural networks can approximate the mapping from market data to model parameters in the Hull-White or Cheyette short-rate models, reducing calibration time from hours to seconds while handling high-dimensional inputs like full volatility surfaces. Empirical studies demonstrate that these methods achieve lower root-mean-square errors in yield curve reproduction compared to traditional least-squares, especially in volatile markets, by learning surrogate pricing functions trained on simulated paths.^[31]

Numerical Implementation

Numerical implementation of short-rate models is essential when closed-form solutions are unavailable, particularly for complex derivatives like American-style options or path-dependent instruments. These methods discretize the underlying stochastic differential equations (SDEs) or associated partial differential equations (PDEs) to enable computational pricing and risk management. Common approaches include Monte Carlo simulation for flexible path generation, finite difference methods for solving pricing PDEs, and tree-based lattices for efficient backward induction, each suited to different model characteristics and computational demands. Monte Carlo simulation generates multiple interest rate paths to estimate expectations under the risk-neutral measure, making it ideal for multi-factor models or high-dimensional problems. Path generation typically employs the Euler-Maruyama discretization scheme, approximating the SDE dr_t = \mu(r_t, t) dt + \sigma(r_t, t) dW_t as

\Delta r \approx \mu(r, t) \Delta t + \sigma(r, t) \sqrt{\Delta t} \, Z,

where Z \sim \mathcal{N}(0,1) is a standard normal random variable, and paths are simulated forward in time from initial conditions. For the Vasicek model, this becomes r = r[t-1] + \kappa (\theta - r[t-1]) \Delta t + \sigma \sqrt{\Delta t} \, Z, enabling bond prices via averaging discounted payoffs over many paths. To reduce variance in bond pricing, techniques like antithetic variates pair paths with -Z to exploit symmetry, halving the effective simulation error without biasing results.^[32] Finite difference methods solve the pricing PDE derived from the short-rate SDE, discretizing the state space (e.g., rate r and time t) into a grid and approximating derivatives with difference operators. For a general bond pricing PDE \frac{\partial P}{\partial t} + \mu(r,t) \frac{\partial P}{\partial r} + \frac{1}{2} \sigma^2(r,t) \frac{\partial^2 P}{\partial r^2} - r P = 0, implicit schemes are preferred for unconditional stability, especially in the CIR model where the \sqrt{r} diffusion term introduces a singularity at r=0. The fully implicit scheme advances the solution via a tridiagonal matrix solve, ensuring convergence even with larger time steps, and boundary conditions (e.g., reflecting at zero) handle the non-negativity constraint. In practice, Crank-Nicolson variants balance accuracy and efficiency, yielding second-order convergence in both space and time for CIR bond pricing.^[33] Tree-based methods construct recombining lattices to approximate the short-rate process, facilitating backward recursion for derivative valuation. For the Hull-White model, a trinomial tree is built by matching moments of the normal-distributed rate increments at each node, with up, middle, and down probabilities ensuring no-arbitrage and term-structure fitting; node values evolve as r_{i,j+1} = r_{i,j} + \delta_j + \sqrt{\Delta t} \sigma \sqrt{3} (up), r_{i,j+1} = r_{i,j} + \delta_j (middle), and symmetric down, where \delta_j adjusts for mean reversion. The Black-Derman-Toy (BDT) model uses a similar binomial or trinomial structure but enforces lognormal dynamics to prevent negative rates. Recombining trees reduce complexity from O(n^2) to O(n) nodes per layer, enabling efficient American option pricing by checking early exercise at each node. Post-2010s negative interest rate environments challenged traditional models like CIR, which assume non-negative rates via the Feller condition. Adjustments include the shifted CIR model, where the short rate is r_t = x_t + \lambda, with x_t following an extended CIR process (as the difference of two independent CIR processes) and \lambda < 0 a constant shift calibrated to market data, preserving affine structure and closed-form bond prices while allowing negative values. This exogenous extension fits observed yield curves exactly and maintains analytical tractability for swaptions via Riccati solutions, validated through Monte Carlo consistency checks.^[34] In multi-factor short-rate models, high dimensionality amplifies computational costs, prompting advancements like GPU acceleration for parallel path simulations and quasi-Monte Carlo (QMC) integration. GPUs exploit the independence of Monte Carlo paths, achieving over 100x speedups in quantitative finance simulations by vectorizing Euler steps across thousands of cores. QMC replaces pseudo-random numbers with low-discrepancy sequences (e.g., Sobol), improving convergence from O(1/\sqrt{M}) to nearly O(1/M) for smooth integrands in multi-factor bond option valuation. These techniques address real-time hedging needs in 2025 trading systems.^[35]

Broader Context

Applications in Finance

Short-rate models play a pivotal role in pricing interest rate derivatives, such as caps and floors, where the Vasicek model approximates Black's formula to value these instruments by simulating mean-reverting short-rate paths that capture the volatility of caplet and floorlet payoffs.^[36] In the Hull-White model, swaptions are priced using Jamshidian's decomposition, which decomposes the swaption into a portfolio of zero-coupon bond options, enabling efficient computation of European swaption values under the model's Gaussian dynamics.^[37] In fixed income portfolio management, immunization strategies leverage duration sensitivities derived from the Cox-Ingersoll-Ross (CIR) model to match asset and liability durations, protecting portfolios against parallel shifts in the yield curve while accounting for the model's square-root diffusion to prevent negative rates.^[38] Yield curve forecasting employs short-rate models like CIR to infer future short rates from cross-sectional bond prices, providing probabilistic projections that inform portfolio rebalancing and investment horizon planning.^[39] For risk management, short-rate models facilitate Value-at-Risk (VaR) computation through Monte Carlo simulations of interest rate paths, generating scenarios to estimate potential losses in fixed income portfolios under varying volatility regimes.^[40] Stress testing incorporates these models to assess impacts from negative rates. Recent developments as of 2025 include short-rate models with stochastic volatility, which enhance the modeling of interest rate dynamics during turbulent periods by allowing volatility to evolve as a separate stochastic process, improving accuracy in derivative pricing and risk assessment.^[41]

Comparisons with Other Models

Short-rate models, which focus on the dynamics of the instantaneous short rate as a Markov process, differ fundamentally from the Heath-Jarrow-Morton (HJM) framework in their approach to term structure evolution. Introduced by Heath, Jarrow, and Morton in 1992, the HJM model specifies the dynamics of the entire forward rate curve directly, with the evolution given by

df(t,T) = \alpha(t,T) \, dt + \sigma(t,T) \, dW_t,

where f(t,T) is the forward rate at time t for maturity T, \alpha(t,T) is the drift term, \sigma(t,T) is the volatility, and W_t is a Brownian motion. To ensure no-arbitrage, the drift \alpha(t,T) must satisfy a specific condition derived from the risk-neutral measure, linking it to the volatility structure: \alpha(t,T) = \sigma(t,T) \left( \int_t^T \sigma(t,u) \, du \right). This direct modeling of forward rates allows HJM to capture the full term structure without relying on a single state variable like the short rate, providing greater flexibility for multi-horizon dynamics but at the cost of increased computational complexity compared to the parsimonious Markovian structure of short-rate models.^[42] In contrast to short-rate models, the LIBOR Market Model (LMM), also known as the Brace-Gatarek-Musiela (BGM) model, models the evolution of discrete forward LIBOR rates under a log-normal distribution, ensuring consistency with market-quoted rates. Developed by Brace, Gatarek, and Musiela in 1997, the LMM posits that each forward LIBOR rate L_k(t) for the period [T_k, T_{k+1}] follows a stochastic process calibrated directly to observed caplet volatilities, with dynamics under the appropriate forward measure given by dL_k(t) = L_k(t) \mu_k(t) \, dt + L_k(t) \sigma_k(t) \, dW_t^k, where the drift \mu_k(t) arises from measure changes. This setup excels in pricing path-dependent derivatives like Bermudan swaptions, where short-rate models often struggle due to their Markovian assumptions leading to non-Markovian paths in discrete tenor structures; empirical implementations show LMM yielding more accurate valuations for such instruments by aligning with market conventions for swaptions and caps.^[43]^[44] Shadow rate models extend short-rate frameworks to handle the zero lower bound (ZLB) on nominal rates, a limitation not inherent in unconstrained short-rate models like Vasicek or CIR. Fischer Black's 1995 conceptualization treats the observed short rate as a floored version of an underlying "shadow" rate that can go negative: r_t = \max(r_t^*, 0), where r_t^* follows a standard short-rate diffusion such as an Ornstein-Uhlenbeck process. Post-2010 developments, amid quantitative easing (QE) policies, have refined these models to better fit term structures during low-rate environments; for instance, extensions incorporating QE effects demonstrate superior empirical performance in capturing yield curve distortions from 2008 to 2022, where traditional short-rate models without flooring overestimate rates near the ZLB.^[45]^[46] Key trade-offs between short-rate models and alternatives like HJM and LMM revolve around simplicity versus consistency. Short-rate models offer advantages in their intuitive state-variable representation and ease of implementation for European-style derivatives, facilitating quick calibration and analytical tractability in one-factor settings. However, they often underperform in capturing multi-horizon volatility structures and term structure consistency, particularly in volatile regimes. HJM and LMM address these by modeling the full curve or market observables directly, ensuring no-arbitrage across maturities, but require more sophisticated calibration due to higher dimensionality and non-Markovian features. Recent empirical studies, including analyses up to 2024, highlight short-rate models' underperformance during the 2022 inflation spikes, where elevated volatility led to poorer fits for long-end yields compared to HJM/LMM frameworks that better accommodated curve shifts.^[47]^[48]

References

[1]
[PDF] Interest Rate and Credit Models - Baruch MFE Program
Dec 11, 2019 · Short rate models, in which the stochastic state variable is taken to be the instantaneous forward rate. Historically, these were the earliest ...
[2]
Short Rate Model - Overview, Importance, and Types
A short rate model is a mathematical model used to evaluate interest rate derivatives and illustrate the evolution of interest rates over time.
[3]
A technical primer on short rate models and interest rate derivatives ...
Apr 6, 2025 · If the short rate is above its long-term mean, the model pushes it back down whilst if the rate is below the long-term mean, the model ...
[4]
[PDF] Parameterizing Interest Rate Models - Casualty Actuarial Society
The short-term rate (also called the short rate or instantaneous rate) is ... Vasicek formulates the interest rate model in terms of changes in the short-term (or.
[5]
An equilibrium characterization of the term structure - ScienceDirect
View PDF; Download Full Issue. Elsevier · Journal of Financial Economics · Volume 5, Issue 2, November 1977, Pages 177-188. Journal of Financial Economics. An ...
[6]
An Overview of the Vasicek Short Rate Model - SSRN
Aug 15, 2014 · The Vasicek model is an early stochastic model of interest rates, with analytical tractability and mean reversion features, and is a short rate ...
[7]
[PDF] Interpreting implied risk-neutral densities: the role of risk premia
Sep 2, 2003 · forward-T measure. That is, the expected future short rate under the T-measure, is equal to the instantaneous forward rate for the relevant ...<|control11|><|separator|>
[8]
[PDF] Interest rate models
Apr 15, 2015 · So in this lecture, we will discuss modeling the short rate R(t) under the risk neutral measure and leave the connection with the forward rate ...
[9]
[PDF] Real-World Interest Rate Models and Current Practices - SOA
Jul 1, 2015 · These models define the instantaneous interest rate (i.e., the short rate) using stochastic differential equations:
[10]
Pricing Interest-Rate-Derivative Securities - Oxford Academic
PDF. Views. Article contents. Cite. Cite. John Hull, Alan White, Pricing Interest-Rate-Derivative Securities, The Review of Financial Studies, Volume 3 ...
[11]
[PDF] A YIELD-FACTOR MODEL OF INTEREST RATES DARRELL DUFFIE
This paper presents a consistent and arbitrage-free multifactor model of the term structure of interest rates in which yields at selected fixed maturities ...
[12]
An Intertemporal Capital Asset Pricing Model - jstor
7. Merton [25] has shown in a number of examples that portfolio behavior for an intertemporal maximizer will be significantly different when he faces a changing.
[13]
[PDF] A Theory of the Term Structure of Interest Rates - NYU Stern
This paper uses an intertemporal general equilibrium asset pricing model to study the term structure of interest rates. In this model, anticipations, risk ...
[14]
A Theory of the Term Structure of Interest Rates
Mar 1, 1985 · A Theory of the Term Structure of Interest Rates ... This paper uses an intertemporal general equilibrium asset pricing model to study the term ...
[15]
(PDF) Pricing Interest-Rate-Derivative Securities - ResearchGate
Aug 9, 2025 · This article shows that the one-state-variable interest-rate models of Vasicek (1977) and Cox, Ingersoll, and Ross (1985b) can be extended
[16]
A One-Factor Model of Interest Rates and Its Application to Treasury ...
Dec 31, 2018 · (1990). A One-Factor Model of Interest Rates and Its Application to Treasury Bond Options. Financial Analysts Journal: Vol. 46, No. 1, pp.
[17]
[PDF] A One-Factor Model Of Interest Rates And Its Application To
A One-Factor Model Of Interest Rates And Its Application To. Black, Fischer; Derman, Emanuel; Toy, William. Financial Analysts Journal; Jan/Feb 1990; 46, 1; ...
[18]
A One-Factor Model of Interest Rates and Its Application to Treasury ...
Aug 6, 2025 · The study [12] introduced the Black-Derman-Toy (BDT) model, a yield-based model applied for pricing bonds and interest-rate options. The BDT is ...
[19]
hullwhite - Normal vs Lognormal Short Rate models
Jul 16, 2013 · Lognormal models are preferred for pricing models due to more precise distribution properties, and cannot describe negative interest rates. ...
[20]
[1104.0322] Explosive behavior in a log-normal interest rate model
Apr 2, 2011 · For sufficiently low volatilities the caplet smile is log-normal to a very good approximation, while in the large volatility phase the model ...
[21]
[PDF] Interest Rate Volatility and the Term Structure: A Two-Factor General ...
In this paper, we develop a two-factor general equilibrium model of the term structure of interest rates using the CIR (1985a) framework and apply it to the ...
[22]
[PDF] Multifactor Term Structure Models
General Affine Models a. Duffie and Kan, “Mathematical Finance”. 1996, generalize the CIR and Vasicek models to a class of models called affine. b. The zero ...
[23]
Interest Rate Volatility and the Term Structure: A Two‐Factor ...
We develop a two-factor general equilibrium model of the term structure. The factors are the short-term interest rate and the volatility of the short-term ...
[24]
A bond pricing formula under a non-trivial, three-factor model of ...
In this paper a bond pricing formula is derived under a non-trivial, three-factor model of interest rates. In the model the future short rate depends on (1) ...
[25]
A bond pricing formula under a non-trivial, three-factor model of ...
Three-factor stochastic models have been proposed in the literature for bond pricing. For example, Chen (1996a) suggested a pricing formula “under a non-trivial ...
[26]
Specification Analysis of Affine Term Structure Models - Dai - 2000
Dec 17, 2002 · This paper explores the structural differences and relative goodness-of-fits of affine term structure models (ATSMs). Within the family of ...
[27]
[PDF] Hull-White term structure model
The Vasicek and CIR models incorporate only one stochastic factor (the short-term rate), so that bond prices of all maturities are subsequently related to the ...Missing: mathematical | Show results with:mathematical
[28]
[PDF] Calibrating Interest Rate Models - SOA
The focus of this paper is narrow: calibrating and validating an interest rate model for default-free bond yields in a single economy. Out of scope are ...
[29]
[PDF] Calibration of the Interest Rate Models: BDT, HL, String and BGM
As expected, the BDT model can be fitted to value the caplets quite well using the term structure of volatilities.
[30]
[PDF] Term Structure Lattice Models
Example 8 (Pricing a Payer Swaption in a BDT Model). We would like to price a 2 − 8 payer swaption in a BDT model that has been calibrated to match the observed.
[31]
High-dimensional parameter calibration of interest rate model for the ...
Mar 13, 2025 · High-dimensional parameter calibration of interest rate model ... neural networks to approximate pricing functions and accelerate the calibration ...1. Introduction · 3. Methodology · 4.2 Experimental Results
[32]
Vasicek Model Simulation with Python - QuantStart
The Vasicek model suggests that the short-term interest rate will be pulled towards the mean interest rate μ at a rate governed by θ , while still being ...
[33]
[PDF] Boundary values and finite difference methods for the term structure ...
The numerical method is implicit in time and second order accurate. The flexibility of a finite difference method makes it easy to change the drift and ...
[34]
On the Deterministic-Shift Extended CIR Model in a Negative ... - MDPI
May 20, 2022 · In this paper, we propose a new exogenous model to address the problem of negative interest rates that preserves the analytical tractability of the original ...
[35]
GPU-Accelerate Algorithmic Trading Simulations by over 100x with ...
Mar 4, 2025 · Statistical Monte Carlo simulations need to be accelerated with GPUs because they are a good fit to the problem and make the researchers more ...
[36]
[PDF] Models for Interest Rates and Interest Rate Derivatives - e d o c . h u
Caps and Floors are derivatives which can be used to insure the holder against interest ... Hull and White (1990) proposed an extended Vasicek model to address ...
[37]
[PDF] Hull-White One Factor Model: Results and implementation
May 9, 2012 · The standard pricing formula for physical delivery swaption in the model uses the Jamshidian decomposition proposed in Jamshidian [1989]. The.<|separator|>
[38]
[PDF] A New Stochastic Duration Based on the Vasicek and CIR Term ...
Given the poor immunization performance caused by the misspecification of the one-factor equilibrium term structure models, this paper proposes an alternative ...
[39]
https://www.worldscientific.com/doi/10.1142/S2010495216500044
[40]
[PDF] A Quantitative Metric to Validate Risk Models - SOA
Abstract. The article applies a back-testing validation methodology of economic scenario–generating models and introduces a new D statistic to evaluate the ...Missing: SOFR | Show results with:SOFR
[41]
Bank business models at negative interest rates
Nov 22, 2017 · This article shows how changes in the yield curve and reductions in the ECB's deposit facility rate (DFR) to negative values have affected ...
[42]
Validating Structured Finance Models - RiskSpan
Nov 11, 2020 · Short-rate models: A short-rate model describes the future evolution of the short rate (instantaneous spot rate, usually written). LIBOR ...
[43]
[PDF] Modelling yields at the lower bound through regime shifts
After the global financial crisis of 2008-09, short-term nominal interest rates have reached levels close to their lower bound (henceforth, LB) in many advanced ...
[44]
Pricing of sustainability-linked bonds - ScienceDirect.com
We find that the SLB market is efficient: the prices of SLBs depend strongly on the size of the potential penalty and there is no evidence of mispricing.
[45]
[PDF] A Macroeconomic Model of Central Bank Digital Currency
We develop a quantitative New Keynesian DSGE model with monopolistic banks to study the macroeconomic effects of introducing a central bank digital currency.
[46]
Bond Pricing and the Term Structure of Interest Rates - jstor
This paper presents a unifying theory for valuing contingent claims under a stochastic term structure of interest rates. The methodology, based on the ...
[47]
The Market Model of Interest Rate Dynamics - Brace - 1997
Jan 5, 2002 · Abstract. A class of term structure models with volatility of lognormal type is analyzed in the general HJM framework. The corresponding market ...
[48]
[PDF] Market Models
In their original paper, Brace, Gatarek and Musiela (BGM) succeeded in deriving Black's formula for caplets and thereby demonstrated its consistency with ...
[49]
Interest Rates as Options - BLACK - 1995 - Wiley Online Library
The nominal short rate is an option. Longer term interest rates are always positive, since the future short rate may be positive even when the current short ...
[50]
[PDF] A shadow policy rate to calibrate US monetary policy at the zero ...
The idea is to extract shadow rates that can turn negative, driven by the dynamics of the term structure of interest rates. Extending the Black (1995) framework ...
[51]
(PDF) Review Paper. Interest–rate term–structure pricing models
Aug 10, 2025 · The paper presents a review of interest-rate term structure modelling from the early short-rate-based models to the current developments.
[52]
Comparative Analysis of Single-Factor and Multi-Factor Short Rate ...
Jun 26, 2024 · This research paper delves into the dynamic landscape of interest rate modeling and its implications for effective risk management strategies.Missing: empirical 2020-2025