Interpreting a Negative Constant in a GARCHX Variance Equation

Summary

The document raises a model-validity concern while fitting an ARX mean equation with a GARCHX conditional-variance equation. The variance specification includes a constant, lagged variance, lagged squared residuals, and an external regressor. The estimated constant is negative, leading the author to ask whether the model can imply negative conditional variance and how to address that outcome.

The author says the variables are already logged and that standardization did not resolve the estimate. No answer, data, parameter estimates beyond the sign of the constant, or diagnostics are provided, so the cause and an appropriate remedy cannot be determined from the document. The question highlights that a variance model must be assessed for nonnegative conditional variance across relevant observations; the displayed form alone does not establish whether the fitted process violates that requirement.

Key ideas

• The model combines an ARX mean specification with an external regressor in a GARCH variance equation.
• A negative variance-equation constant raises concern about whether fitted conditional variance remains nonnegative.
• Logging and standardizing the variables did not resolve the reported negative estimate.
• The document provides no fitted parameter values or answer that would identify a remedy.
• Model validity depends on the behavior of the full conditional-variance process, not the constant alone.

Tags

# Negative constant in GARCHX model


# Negative constant in GARCHX model












I am fitting the following ARX(1,1)-GARCHX(1,1,1):

\begin{align*} y_t&=c+a_1y_{t-1}+\gamma_1x_t+\varepsilon_t\\ h_t&=\delta+\omega_1h_{t-1}+\theta_1\varepsilon_{t-1}^2+\pi_1x_{1,t} \end{align*}

Delta is negative, suggesting the conditional variance can theoretically be negative. Does anyone know how to get around this problem? My variables are already in logarithmic form, and I have tried standardising them to no avail. Any help would be much appreciated.

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.