Interpreting the Volatility Process in a Factor Interest Rate Model

Summary

The document asks how to identify the volatility process in a factor Heath-Jarrow-Morton interest rate framework when observations of spot rates across maturities are available. It distinguishes the model’s volatility process from the observed volatility of spot or forward rates. The response clarifies that the matrix process is the diffusion coefficient of the factor vector: it specifies how the factors load on the Brownian shocks in the model’s stochastic dynamics.

This distinction matters because rate volatility observed in the yield curve is not automatically the same object as factor volatility. The response points the reader to further equations in the referenced paper, but the excerpt does not reproduce them or explain an estimation procedure. It therefore provides a useful interpretation of the model notation, while leaving open how to infer the process empirically from rate data, including choices about factors, Brownian dimension, and model calibration.

Key ideas

• The volatility process is the diffusion loading matrix for the model’s factor vector.
• Its dimensions correspond to the number of factors and Brownian shocks.
• Observed spot or forward rate volatility is distinct from factor volatility.
• The excerpt clarifies notation but does not describe a data-based estimation method.

Tags

# What does volatility process mean and how is it different from volatility?


# What does volatility process mean and how is it different from volatility?












I have been reading the paper "Bridging P-Q Modeling Divide with Factor HJM Modeling Framework" by Lyashenko and Goncharov (2022). On Equation 5 of page 4 of the paper, I came across the volatility process $\Sigma_{t}$ which is a K x N matrix where K is the dimension of basis vector and N is the dimension of Brownian motion. Assuming I have the data for spot rates at many maturities and at every time, how can I find this volatility process term?

I know that it can't be the same as volatility of spot rate or the forward rate. For example, given the screenshot below from page 18 of the paper, I have $R_{t}$ and can find out $B_{R}$. The issue is to find out the volatility processes $\Sigma_{t}$ or $\Sigma_{t}^{R}$.

## Answer by Alex (score 1)

https://quant.stackexchange.com/a/79307

$\boldsymbol{\Sigma}_t$ is a $K \times N$ matrix volatility process for the factor vector $X_t$, and $dX_t = \ldots dt + \boldsymbol{\Sigma}_t dW_t$.

See equations (19) and (20)

Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.