Title

Confidence Sets in Regressions with Highly Serially Correlated Regressors

Author(s)

James H. Stock James Stock (Harvard University)

Mark W. Watson Mark Watson (Princeton University)

Abstract

Small deviations from exact unit roots can product large coverage rate distortions for conventional confidence sets for cointegrating coefficients (Elliott [1994]). We therefore propose new methods for constructing confidence sets for long-run coefficients with highly serially correlated regressors which do not necessarily have a unit root. Although the standard bootstrap is shown to be asymptotically invalid, a modified, valid bootstrap is developed. invariant confidence sets that are option (highest average accuracy) are obtained but are difficult to implement in practice. An approximately optimal invariant method is proposed; this works almost as well as the optimal method, at least for a single persistent regressor.

Creation Date

1996-12

Section URL ID

Paper Number

1996-1

URL

http://www.princeton.edu/~mwatson/papers/boot5.pdf

File Function

Jel

C01

Keyword(s)

Cointegration, Local to Unit Roots, Money Demand

Suppress

false

Series

13