HAC Standard Errors for a Yield Regression

Summary

The document asks how to estimate standard errors robust to heteroskedasticity and serial correlation in an OLS regression of monthly yield changes on the current yield. This is a time-series inference problem: conventional OLS standard errors may be unreliable when residual variance changes over time or errors are autocorrelated.

The document gives no answer, estimator, diagnostics, or empirical evidence, so it does not establish a particular correction or lag choice. It identifies a fixed-income setting and a clear statistical question, but readers would need additional guidance to choose and implement an appropriate covariance estimator.

Key ideas

• The regression relates changes in a one-month zero-coupon yield to its current level.
• The question concerns inference robust to both heteroskedasticity and residual autocorrelation.
• The document does not provide a proposed estimator or supporting analysis.

Tags

# Robust standard errors OLS for term structure


# Robust standard errors OLS for term structure












Suppose i have estimated the following model with OLS: $y_{1,t+1} - y_{1,t} = \alpha + \beta y_{1,t} + \epsilon_{t+1}$. Where $y_{1,t}$ is the 1 month zero-coupon yield at time t. What would be an appropriate way to obtain standard errors robust to heteroskedasticity and residual autocorrelation?

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.