Let yₜ be an order p autoregressive process of the form yₜ + ∑ᵖₛ₌₁ βₛ yₜ₋ₛ = uₜ, where the uₜ's are i.i.d. variables with a symmetric distribution F such that E log⁺ |uₜ| < ∞. For the Yule-Walker version βT^ of the least-squares estimate of β = (β₁,⋯, βₚ), it is shown that T¹/2(βT^ - β) is bounded in probability.
No takes yet. Share an insight, caveat, or question.
Yohai et al. (1977) studied this question.