Talk:Hildreth–Lu estimation

Method of selecting rho

Currently the article says

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

$y_{t}-\rho y_{t-1}=\alpha (1-\rho )+\beta (X_{t}-\rho X_{t-1})+e_{t}.\,$

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

But if I recall correctly, one actually selects a value of rho for which the resulting residuals are serially uncorrelated. Should the above be changed to this?:

.... From all these auxiliary regressions, one selects the one that yields the smallest residual sum of squares from among those for which we accept the null hypothesis of no serial correlation of residuals.

Loraof (talk) 01:04, 29 December 2017 (UTC)

• Never mind – I found a couple sources that agree with what the article says. Loraof (talk) 21:54, 30 December 2017 (UTC)