Page 1 of 1

Forecasting ARDL in EViews 10

Posted: Thu May 16, 2019 3:44 pm
by mikebeeonthekop
My dependent variable is a ddlog number of passengers at an airport and so the forecast is of ddlog passengers. How do I convert back into a real number of passengers ? I think in EViews 11, you have the option to have the forecast produced in real numbers? But not in E-Views 10? Thanks for your help.

Re: Forecasting ARDL in EViews 10

Posted: Thu May 16, 2019 6:19 pm
by EViews Gareth
Use d(dlog(passenger)) as your dependent variable.

Re: Forecasting ARDL in EViews 10

Posted: Fri May 17, 2019 5:46 am
by mikebeeonthekop
Many thanks, Gareth.

How do I get a forecast of passengers for subsequent months using the lagged independent variable coefficients? For example, I want to get the forecast passengers for 12 months beyond the last month with an actual independent variable value based on the lagged coefficients?

Mike

Re: Forecasting ARDL in EViews 10

Posted: Fri May 17, 2019 7:39 am
by EViews Gareth
I don't understand the question.

Isn't that just what a forecast is?

Re: Forecasting ARDL in EViews 10

Posted: Sat May 18, 2019 4:18 am
by mikebeeonthekop
Gareth

Here's the equation. Pax is the dependent variable and searches is the independent variable. I have data for the independent variable up to May 2019. What I would like the forecast to do is to estimate the dependent variable for 8 months after May 2019 using the lagged coefficients but it will only forecast to May 2019.

Thanks

Mike

Dependent Variable: D(DLOG(PAX))
Method: ARDL
Date: 05/17/19 Time: 08:54
Sample (adjusted): 2015M11 2019M03
Included observations: 41 after adjustments
Dependent lags: 8 (Fixed)
Dynamic regressors (8 lags, fixed): DDLOGSEACHESYLWINBC
Fixed regressors: C

Variable Coefficient Std. Error t-Statistic Prob.*

D(DLOG(PAX(-1))) -1.266072 0.189484 -6.681682 0.0000
D(DLOG(PAX(-2))) -1.790391 0.274034 -6.533469 0.0000
D(DLOG(PAX(-3))) -2.078512 0.353590 -5.878316 0.0000
D(DLOG(PAX(-4))) -2.296622 0.429315 -5.349507 0.0000
D(DLOG(PAX(-5))) -1.584443 0.442105 -3.583862 0.0016
D(DLOG(PAX(-6))) -1.630204 0.341099 -4.779274 0.0001
D(DLOG(PAX(-7))) -0.688054 0.278615 -2.469555 0.0214
D(DLOG(PAX(-8))) -0.265106 0.177810 -1.490951 0.1496
DDLOGSEACHESYLWINBC 0.096754 0.032105 3.013658 0.0062
DDLOGSEACHESYLWINBC(-1) 0.128840 0.035171 3.663267 0.0013
DDLOGSEACHESYLWINBC(-2) 0.187252 0.040185 4.659723 0.0001
DDLOGSEACHESYLWINBC(-3) 0.130813 0.044496 2.939885 0.0074
DDLOGSEACHESYLWINBC(-4) 0.250937 0.045652 5.496779 0.0000
DDLOGSEACHESYLWINBC(-5) 0.144448 0.051890 2.783722 0.0106
DDLOGSEACHESYLWINBC(-6) 0.129838 0.051059 2.542881 0.0182
DDLOGSEACHESYLWINBC(-7) 0.142372 0.040723 3.496081 0.0019
DDLOGSEACHESYLWINBC(-8) 0.083309 0.029395 2.834107 0.0094
C -0.002394 0.007899 -0.303030 0.7646

R-squared 0.964200 Mean dependent var 0.002249
Adjusted R-squared 0.937739 S.D. dependent var 0.199664
S.E. of regression 0.049821 Akaike info criterion -2.860807
Sum squared resid 0.057088 Schwarz criterion -2.108507
Log likelihood 76.64655 Hannan-Quinn criter. -2.586861
F-statistic 36.43846 Durbin-Watson stat 1.870587
Prob(F-statistic) 0.000000

*Note: p-values and any subsequent tests do not account for model selection.

Re: Forecasting ARDL in EViews 10

Posted: Sat May 18, 2019 6:33 am
by startz
Your model says you need the contemporaneous value of the independent variable to forecast the dependent variable in a given period. You don't have that. The forecast can't be done with this model.

Re: Forecasting ARDL in EViews 10

Posted: Sun May 19, 2019 4:07 am
by mikebeeonthekop
OK. Thanks