Generating CUSUM and CUSUMsq charts for ARDL model in EVIEWS 9

I had received recent comments regarding Eviews 9 unable to show CUSUM and CUSUMsq stability charts after the estimation of ARDL. Which is illustrated in here.. Note you can get eviews 9 here by filling the request form.

In this blog I will show how we can generate these stability charts. If you follow from the previous ARDL blog and come up to following stage where you have already estimated the ARDL model and going further for the F bounds test.



Here you can press the view button and press representations like in picture  below



It will show the equation form and the representation which this ARDL is using in the background. here you have to highlight the estimation equation and copy it like in the image below.



then go in Object option above and then new object and then press OK for equation as new object. Here it will ask about equation specification paste the estimation equation which you copied earlier and press OK. It will show the same ARDL output in OLS format here.


From here press the view button and go in stability diagnostics and press Recursive Estimates (OLS only).. You will get following options.


here select the CUSUM test and press OK to get the CUSUM chart for the ARDL and then do it again for CUSUM sq. This way you can generate your CUSUM and CUSUMsq charts, which were not there in the default package as compared to the Microfit outcome shown here..

Note: The new update of Eviews has enabled users to generate CUSUM charts directly.


34 thoughts on “Generating CUSUM and CUSUMsq charts for ARDL model in EVIEWS 9

  1. Faisal says:

    Dear Sir, indeed helpful.
    When i go in Object option above i did not see the new object. can u explain it kindly sir

  2. Niraj says:

    Thank you so much Noman Arshed sir!! i was really wondering about CUSUM test in eviews 9. it helped me to plot CUSUM and CUSUMSQ.

  3. Maxim Harper says:

    Hello Sir,

    Forgive me if I am wrong but why not sure the CUMSUM/CUMSUMSQ’s generated buy using View / Stability Diagnostics / Recursive Estimates / CUMSUM ?

    Many thanks,

      1. Maxim Harper says:

        Sorry I realise now in hindsight that was not clear,

        What I was saying is at least in my version of EVIEWS9 (Nov 2015 Build) after estimating an ARDL model, one does not have to go through this procedure above, just follow the path View / Stability Diagnostics / Recursive Estimates / CUMSUM. Image here:

        Is this procedure different to the one you’ve described above?

        Many thanks,

  4. ibn-e-adam says:

    Sir Noman Arshad , it’s very helpful but I’ve got another method too suggested by Dave Giles. He said save residuals from ARDL then regress these residuals just on a constant using ols and then go from there and get cusum.

    Sir please tell me if this method is also fine or not ?


  5. massy says:

    Hi thanks for your information first of all what is our criteria for choosing rest constant and unrest constant also rest linear trend ? for trend I am checking it is prob significant or not

    second question in eviews 9 with new aproach for ardl how can we apply best second model LİKE ardl (3.2.4) fixing regressors max lag is not solving this problem using fixed regressors part is solving estimation but after applied long rung coefficients coming with 1 lag

  6. massy says:

    it is possible with new update to get directly cusum test gtaphs but do u know how to apply fix regressors pls dont say use fixed regressor on ardl model because of that fixing methed is fixing all variables at the same level

  7. Mohamed says:

    thank you sir this tutorial helps me very much. i want to ask how run Toda and Yamamoto procedure to test the Granger causality

  8. femi Jones says:

    in a short run form result from Cointegration, d(texp), d(texp(-1)) and d(texp(-2)). I want to test for the joint significance of texp on gdp for instance using Wald test in the short run. do I exclude d(texp) from the test or I should include it as well i.e c1=c2=c3=0 (dtexp included) or c2=c3=0 (dtexp excluded)

  9. hafizh says:

    dear sir,

    its really a great sharing. However, i have some doubt regarding the CUSUM test. the CUSUM test actually based on ECM or ARDL? from what i understand, we diagnose the ECM model instead of ARDL model. but from your tutorial, the diagnose is based on ARDL model. Often paper shows diagnostic is based on ECM not ARDL. in addition, from what i learn, we first extract the ECT series then create a ECM model. next we diagnose it.

    thank you sir. really nice sharing

    1. Noman Arshed says:

      ARDL model is a type of ECM model which incorporates I(0) variables. and Eviews does extraction of ECT in the background and runs ECM to show the short run coefficients.

  10. Khoa says:

    Dear sir,
    I run ARDL in Eviews 9. when I run CUSUM, my eviews menu only has Ramsey Reset test, Leverage Plots, Influence Statictis, but does not have Recursive Estimates in VIEW/stability Diagnostics. What should I do to run Recursive Estimates? Thank you

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s