Key points are not available for this paper at this time.
A first-order autoregressive process, Yₜ = Yₓ - ₁ + ₜ, is said to be nearly nonstationary when is close to one. The limiting distribution of the least-squares estimate bₙ for is studied when Yₜ is nearly nonstationary. By reparameterizing to be 1 - /n, being a fixed constant, it is shown that the limiting distribution of ₙ = (ⁿₓ = ₁Y²ₓ - ₁) ^1/2 (bₙ -) converges to L () which is a quotient of stochastic integrals of standard Brownian motion. This provides a reasonable alternative to the approximation of the distribution of ₙ proposed by Ahtola and Tiao (1984).
Chan et al. (Tue,) studied this question.