ARDL Model

ARDL model (Bound test)
Today we will discuss about the ARLD model and preconditions of this Approach using
EVIEWS 9. Following is a chart which will clarify about the model section on the base of
data stationary.
In above chart ill focus on ARDL I have mention in above that we can Run ARDL when we
have our data stationary mix i.e. Few variables are stationary at level and few ones at first
difference but it’s also important to know that ARLD also can be run if our variables are
purely stationary at level or purely at first differences.
I have few assumptions for ARDL approach data

 Data must be free from autocorrelation
 Data must be free from heteroskedasticity (HSK)
 Data must be normal distributed
 No one variable stationary at I(2)
 Data should be stationery purely at level (0) or purely first difference (1) or mixture of
level and first difference.
Now lets us start applying ARDL using EVIEWS 9.

1. Step one drag you excel file on Eviews 9 icon
2. Check stationary level of your series if your series fulfill the assumptions of ARDL
then apply ARDL otherwise move for other tests.
3. Now we have checked stationary level of our variables let suppose variables are
stationary on mixture pattern few ones stationary at level and few stationary at first
difference.
4. Go to quick ------------> select estimate equation and from bottom and from drop down
Manu select ARDL, when you will select ARDL new window will be open.
From here I selected
ARDL option
Dynamic
selection
Option Table
You also can
chose fixed
From automatic
selection chose
maximum lags for
dependent and List of fixed
independent Regressor
variables
The Specification tab gives you option for lags of variables in the above screenshot we have
two options for lags, first one is automatic and second is fixed,
1. if you select automatic lags then Eviews automatically chose appropriate lags for your
variables but you have to choose maximum lags for your variables.in automatic lags
selection option u can specify different lags length for dependent variables and
independent variable let suppose I can select 6 lags for dependent and 8 for
independent it’s all up to and next Eviews automatically select what must be the suit
able lags length for dependent and independent variables.
2. But if we select fixed lags length then we must specify same lags length for the both of
dependent as well as independent variables.
3. List of fixed Regressor this option gives us opportunity to use all those variables
which are fixed or static variables, i.e. variables without lags.
4. Trend specification from this option u can add trend or trend and constant or any static
variable also can be specify
5. Option tab; you can chose the Akaike Information Criterion (AIC), (SIC), (HQ), or the
Adjusted R-squared.
6. Dynamic selection in this area we write our equation dependent first and then all
independent variables.
Let suppose I write my equation and go-ahead.
I selected automatic lags lent criteria and used maximum 4 lags and all things remain
unchanged because which option I was required selected by default but you can change
according to different situation, it’s all depend u how much maximum lags u should include,,
after this Eviews automatically select lags, I do ok and proceed.
Here are results we can see , I have include
4 lags but Eviews automatically select 4 for
dependent variables, while three for FDI
independent variables etc.
In above results we can see that we have use AIC for the optimal lag lengths we see how AIC
chose these lags ,we will check it with the help of graph, go to views of above resulted
window---model summary—graph.
See the numbers of lags and same our

results are indicating that for dependent
On these lags AIC
variables 4 lags suitable while same for
is minimum
FDI, GDP and LIFE.
Now we have need F-STATISTICS value so that we can conform we can move further or not,
For F statistics we will go in above resulted windows views-----coefficient diagnostic---bound
test.
F-statistic value tells about the
cointegration among variables if F value
comes less than critical bound values then
we conclude that there is no cointegration
among variables.
There is different critical value of bound on

different level of confidence; here our F
value is above from upper and lower bound
test so we can say there is cointegration in
our variables. Not if our F values come less
than critical value must do add or remove
variables or adopt any other way.
Conclusion from above results we can conclude that there is cointegration in our variables.
So we move on for long run relationship. Go to view of resulted window or above window----
coefficients diagnostic----cointegration and long run form
Here we can see long run relation

Note:coineq(-1) must be negative and
significant,
Further we can check serial correlation, heteroskedasticity or normality of data etc. Thank you
so much for being with me…

ARDL Model

Uploaded by

Copyright:

Available Formats

ARDL Model

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

ARDL Model

Uploaded by

Copyright:

Available Formats

ARDL model (Bound test)

I have few assumptions for ARDL approach data

Now lets us start applying ARDL using EVIEWS 9.

Let suppose I write my equation and go-ahead.

See the numbers of lags and same our

There is different critical value of bound on

Here we can see long run relation

You might also like