Hildreth–Lu estimation

Hildreth–Lu estimation, named for Clifford Hildreth and John Y. Lu, is a method for adjusting a linear model in response to the presence of serial correlation in the error term. It is an iterative procedure related to the Cochrane–Orcutt estimation.

The idea is to repeatedly apply non-linear least squares to:

for different values of ρ between −1 and 1. From all these auxiliary regressions, one selects the one that yields the smallest residual sum of squares.