Key points are not available for this paper at this time.
Consider a first-order autoregressive process Xₜ = Xₓ - ₁ + ₜ, where \ₜ\ are independent and identically distributed random errors with mean 0 and variance 1. It is shown that when = 1 the standard bootstrap least squares estimate of is asymptotically invalid, even if the error distribution is assumed to be normal. The conditional limit distribution of the bootstrap estimate at = 1 is shown to converge to a random distribution.
Basawa et al. (1991) studied this question.