Command for the Predicted Value of ∆Yt

[dy1107]

Please tell me, what Eviews’ command I should use to operate the predicted values of ∆Yt. For example, in STATA, “deltaYhat” is used in the equation for the predicted value of ∆Yt, such as regress y deltaYhat x dif_x. For Eviews, what command should I substitute for deltaYhat. - ls y (??? instead of deltaYhat) x d(x). I would greatly appreciate it if anyone can tell me. Thanks!

[EViews Gareth]

I don't understand what you're asking.

[dy1107]

I am trying to use the single equation error correction model. Since the long term effect is a ratio of two coefficients, the single equation ECM does not report the standard error and confidence level of the total long term effect of an IV on DV. Thus, I need to estimate the standard error and confidence level of the total long term effect of X on Y using the Bewley transformation regression. The following procedures will be adopted. First, I estimate ΔYt= α+ βY_(t-1)+ βX_t+βΔX_t+ ε_t. Second, I should take the predicted values of ΔYt from the previous procedure and estimate Yt= α+ δ_0 ΔY_t+ δ_1 X_t- δ_2 ΔX_t+μ_t. In STATA, I can use the following command to estimate the second equation: Regress y deltaYhat x dif_x. I wonder what commend I should use to estimate the second equation in Eviews. I hope this explanation is clear enough. Please let me know! Thanks!

[EViews Gareth]

Use equation.fit or equation.forecast to get the fitted values from first equation.

[dy1107]

My first equation in Eviews’ words would be d(y) c d(x) y(-1) x(-1), and I am not sure how I can transcribe the second statistical algorism into Eviews’ words? Do you mean this? - y c equation.fit x d(x). I did but some error message appear about “equation.fit.”

[EViews Gareth]

Assuming you have an equation called "eq1", you would produce fitted values of Y by typing:

Code: Select all

eq1.forecast yf

See the equation section of the Object Reference for more details.

[dy1107]

I found that the variables of interest are nonstationary and integrated of I(1), but they are cointegrated. Can I estimate the equation in levels using OLS? I read somewhere that the equation of cointegrate variables is less likely to produce spurious results. I just wonder whether I can run the analysis in levels if the variables are conitegrated without differencing the series or building error correction model. Thank you!