|
Durbin s two-step method of estimating ρ Write the generalized difference equation (10.14) in a slightly different but equivalent form as follows:
Yt = B1(l - ρ) + B2Xt - ρB2Xt-1 + ρYt-1 + vt
In step 1 Durbin suggests estimating this regression with Y as the dependent variable and Xt, Xt-1, and Yt-1 as explanatory variables. The coefficient of Yt-1 will provide an estimate of ρ. The ρ thus estimated is a consistent estimator; that is, in large samples it provides a good estimate of true ρ.
In step 2 use the p estimated from step 1 to transform the data to estimate the generalized difference equation (10.14).
Apply Durbin s two-step method to the U.S. import expenditure data discussed in Chapter 7 and compare your results with those shown for the original regression. |
| New search. (Also 1294 free access solutions) |