Serena Ng

Overview

Serena Ng is an econometrician at Columbia University (Université de Montréal at the time of Perron-Ng 1996), specializing in unit root testing, spectral density estimation, and large-dimensional factor models. Her work with Pierre Perron on modified unit root tests identified the autoregressive spectral density estimator as the critical component that determines whether PP-type tests have correct size in problematic error specifications.

Key Contributions / Features

• M statistics for unit root testing (1996, with P. Perron): local asymptotic analysis showing that the AR spectral density estimator $s^2_{AR}$ — formulated on first differences of $y_t$ — decouples $\sigma^2$ estimation from the unreliable $\hat\alpha$, giving M tests ($MZ_\alpha$, $MSB$, $MZ_t$) robust size properties across three near-unit-root error models.
• Autoregressive spectral density properties (1995 working paper, with P. Perron): derived the behavior of $s^2_{AR}$ at frequency zero in local asymptotic frameworks; showed $\hat b(1)$ diverges but $s^2_{ek}$ converges to zero at the same rate, giving a finite non-zero limit for $s^2_{AR}$.

Related

• Unit Root Tests
• Pierre Perron
• Peter C. B. Phillips
• Perron-Ng (1996) Useful Modifications to some Unit Root Tests with Dependent Errors and their Local Asymptotic Properties