(Stata13):ARDL Models and Bounds Test Estimations

[@CrunchEconometrix]

KZbin recently changed the way my content will be monetised. My channel now needs 1,000 subscribers. So it would be amazing if you show your support by both watching my videos and subscribing to my channel if you haven’t done so already. Monetising my videos allows me to invest back into the channel with some new equipment so this small gesture from you will be extremely huge for me. Many thanks for your support….CrunchEconometrix loves to teach, help me stay online.

[@oresterpaluma2669]

Econometrics is really an enjoying and interesting course with your tutorials,,

[@CrunchEconometrix]

Thanks, Orester for the encouraging feedback. Deeply appreciated! ❤️🙏

[@kofimawuli8940]

Explanation very clear. Good work. Keep it up.

[@CrunchEconometrix]

Thanks Kofi, for the encouraging feedback. Deeply appreciated! 🙏

[@LongLE-te8nq]

Thank you so much, Sir. That is a great lesson that I obtained!

[@CrunchEconometrix]

You are welcome, Sir Long🥰🙏

[@shahidulislam-xp8xs]

Thank you for making such helpful tutorials

[@CrunchEconometrix]

Thanks, Shahidul, for your encouraging feedback. Deeply appreciated 🙏 🥰

[@chelsealeibrandt5232]

You’re amazing!!! This was extremely helpful thank you

[@CrunchEconometrix]

Glad you find it helpful, Chelsea... thanks!

[@twikaleghemwalwanda3663]

Thanks for the videos, they are very helpful.

[@CrunchEconometrix]

Thanks, Twikaleghe for the encouraging feedback. Deeply appreciated! Please may I know from where (location) you are reaching me?

[@ange-jolinardaldrichtbetsa4407]

Hello Professor and thank you for your very instructive videos on time series and particularly on the ARDL model. However, I have two questions: 1/ In my study, when I ran the varsoc command (“varsoc LogPIBRH LogDEPS LogINV”), I obtained the following optimal lags : 4, with AIC = -12.124*, and LR = 39.13*; 2, with HQIC = -11.2733* and FPE = 9.6e-11*; and 1, with SBIC = -10.4585*. Do I deduce that the optimal lag is 2 because FPE has the smallest value, or do I retain 4 according to the AIC criterion? I'm a bit confused, knowing also that by running “varsoc” for each variable, I obtain the following optimal lags: 2 for LogPIBRH, and 1 for the other series. 2/ Why, in the “maxlags(p)” option of the ARDL syntax which precedes that of the matrix-list (in my study: “ardl LogPIBRH LogDEPS LogINV IDH, maxlags(2) aic”), didn't you consider that p = 4 which is the default maximum lag order is 4? NB: my data is in years Thank you again !

[@CrunchEconometrix]

There are different ways to do this, but the most convenient is to use the MAXLAGS option and allow Stata fix the appropriate lags. You can also perform individual VARSOC for each variable and incorporate into your model.

[@dalitsakadze4197]

Hi. Could you also tell me if you have videos on how to resolve serial correlation and instability in the ARDL model?

[@CrunchEconometrix]

Dali, adjust the lags to correct for serial correlation. Perform structural break analysis or modify regressors to resolve model instability.

[@tariqrahim1129]

Dear professor, Thank you very much for clarifying our concepts and making econometrics reachable. While doing my thesis estimations I applied the mentioned do file but the following are some of the commands that are not working in Stata 12. Can you please help me by suggesting a possible solution so that I can complete my estimations? The commands that are giving error includes : * matrix list e (lags) * estimates regstore ecreg * estat dwatson * estat bgodfrey , lags (4) *E estat lmtest, white * cusm6 lnco2 lngdp lnupop lntradebal lnindustry lnenergyuse, cs (cusm) lw (lower) uw(upper) * drop cusm upper lower Thank You!

[@CrunchEconometrix]

Hi Tariq, kindly post this on Statalist.org for constructive feedback. Thanks.

[@tariqrahim1129]

@@CrunchEconometrix Thanks Dr. for the feedback.

[@akindelemichaelojo8085]

Thanks a lot for this wonderful youtube trainning. I really benefit from it. Please I need the do-file for ARDL Models and Bounds test Estimation and that of ARDL + VECM estimate using time series data. I will appreciate it if I can have the dataset for practice also. thanks

[@CrunchEconometrix]

Thanks, Akin for the positive feedback on my videos. Deeply appreciated. Dofiles and datasets are available on my website. May I know from where (location) that you are reaching me?

[@akindelemichaelojo8085]

I am from Futminna

[@CrunchEconometrix]

@@akindelemichaelojo8085 Awesome! Kindly spread the word about my videos to your friends and academic community for awareness. They'll learn some useful tips and skills too...thanks 😊

[@varaquesabounjian5277]

Hello, Thank you for the helpful tutorial. I have a question, like in the video my bivariate ARDL model suffers from heteroskedasticity. How do you suggest to fix this problem ? Thank you

[@varaquesabounjian5277]

both of my variables are the difference of the log and I am using up to 4 lags

[@CrunchEconometrix]

Varaque, kindly watch my videos on HETEROSCEDASTICITY on ideas about how to resolve this.

[@MrNoman0409]

Hi, is there any video available on bootstrap ARDL (Eveiws)?

[@CrunchEconometrix]

Hi Mohd, I don't have such video at the moment. Thanks

[@alnas41a]

hello thank you Dr. for the video, please i have a question, if all variables are stationary at first difference do you use the main variables or the first difference variables? thank you

[@CrunchEconometrix]

Habib, I explained this in my ARDL videos. You may need to watch "This is how to specify ARDL models" and support that by reading the relevant literature listed at the end of ARDL videos. Thanks

[@rosariolodovice5125]

Dear Prof. Adeleye, Good day! How are you? I would like to ask question about my variables and model. I have two time series variables they are all I(1). I wanted to use ARDL model. Should I need to generate the first difference of each series before running ARDL or the ARDL command automatically differenced the variables in stata? Please what is the correct specification I use? I confused with the difference operator in the ARDL model actually. Thank you.

[@CrunchEconometrix]

Hi Rosario, kindly watch the introductory video "This is how to specify ARDL models". I gave some clear explanations. Thanks

[@yanuozhou6028]

Hi Rosario, do you have an answer now? Im struggling with the same question :(

[@CrunchEconometrix]

@Yanuo and @Rosario, there's no need to struggle. My ARDL videos are well explained in addition to showing you how to go through the estimation.

[@nabilaa.vashti2602]

Why do my STATA said "command ardl is unrecognized" ? I've already followed all the steps

[@CrunchEconometrix]

Hi Nabila, it implies that you are yet to install the syntax. Type help ardl in the Command Window and follow the Stata prompts.

[@sibylla553]

Hi Prof.! what to do if the CUSUM and CUSUMQ plotted points fall below the control limit?

[@CrunchEconometrix]

Kindly watch my videos on STRUCTURAL BREAKS. Thanks.

[@jeanyoung408]

Thank you for your detailed lecture. But I have some problem.when I command "ardl Deficit GINI GDP,lags(1,0,0)ec btest" is sucessful. I change the variable position "ardl GINI Deficit GDP,lags(1,0,0)ec btest ", it appear error instruction"Maximum number of iterations exceeded r(498)". How can I do? thanks

[@CrunchEconometrix]

Hi Jean, it simply means that Stata is unable to estimate the model you have specified. Change it. May I know from where (location) you are reaching me?

[@anilraj9908]

Thank you for the presentation. I ve a doubt please. If the p value of adj (ect) and long run variable in the output is higher than 0.05 should we interpret it as no statistical significance and no long run equation even if the f value and t value of bounds test shows counteraction?

[@CrunchEconometrix]

Hi Anil, it implies that the reversion to long-run equilibrium is not statistically significant.

[@Azam_Pakistan]

How to interpret the short run coefficients when there are lags? Do you have a video on that?

[@CrunchEconometrix]

Hi Azam, kindly watch my video on "Interpret regression output". Interpret the result in relation to the lag and incorporate the phrase "in the short-run". Thanks.

[@himanisharma5417]

Hi, thanks for the clear explanation. I was facing a question regarding maxlags(2) at 04:32. When all the criteria chose 1, why did we spectify maxlag(2)? Thanks a lot !

[@CrunchEconometrix]

I did that for the software to automatically determine the lag length.

[@raihansiddika9942]

Hi, Thanks for your video. I have a question. For the ARDL bound test, which critical value should we consider? In the present example, there are four value of I(0) and I(1)? Should we consider all four of them? What happens if any tabulated F value exceed one (upper bound I(1)) of the four critical F value, but does not satisfy the 3 other? what would be our decision? For the case where lnimp is a dependent variable, the F statistics of 5.861 did not exceed the upper bound of the critical value 6.36. But it exceed other three critical F value at I(1). On what basis, we ignored decision of the fourth column (where we failed to reject the null)? Thanks.

[@CrunchEconometrix]

Hi Raihan, I addressed your queries in this clip and in other ARDL videos. Kindly watch them all. Thanks.

[@raihansiddika9942]

@@CrunchEconometrix Thanks for your reply. But I did not find my answer although I have watched all of your video. Another question, if no cointegration is found which ARDL model should we consider? The following one? ardl lnmva lnimp rexch, maxlag(2) aic

[@CrunchEconometrix]

@@raihansiddika9942 Your response turned out to be untrue.

[@alhazi100]

Thanks for this video. it has help me

[@CrunchEconometrix]

Thanks, Abdoul for the positive feedback on my video. Deeply appreciated! May I know from where (location) you are reaching me?

[@alhazi100]

@@CrunchEconometrix I'm from Senegal

[@CrunchEconometrix]

@@alhazi100 Awesome! Kindly spread the word about my KZbin Channel to your friends, students and academic community in Senegal 🇸🇳...thanks 😊.

[@alhazi100]

@@CrunchEconometrix don't worry for that. Thanks agains

[@khemaram3790]

Hello Ma'am. My optimal lag selection is 3. what maxlags should I insert in this case? Please help

[@CrunchEconometrix]

If optimal lag is 3, then go ahead and analyse your data using the guide shown in the video.

[@khemaram3790]

thanks so much @@CrunchEconometrix

[@anilraj9908]

Hi, Is in't appropriate to select the BIC criteria instead of AIC while running ARDL command in stata if BIC has the least value as per the varsoc command table ?

[@CrunchEconometrix]

Of course, it is ideal to choose the criteria with the lowest values. But even at that, you can decide to use any only be consistent.

[@mrzadocknanyaro7584]

Hello can i run a single ARDL equation/model with only one dependent variable and around 6 explanatory variables? Also how many explanatory variables is too much?

[@CrunchEconometrix]

Hi Mr. Zadock, kindly watch my video on GENERAL-TO-SPECIFIC. It addresses your query. Thanks.

[@mrzadocknanyaro7584]

@@CrunchEconometrix thank you ,let me do that

[@anilraj9908]

Thanks for posting the video. Why are we using the difference operator for dependent variable and for short run equation? I actually didn't get it. If you could please clarify?

[@CrunchEconometrix]

Hi Anil, because that is the way the model is specified if you want to perform the Bounds test. It is a test for cointegration, hence the depvar takes the difference operator as the short-run equation of the model. To get into the econometrics details, kindly refer to either Gujarati and Wooldridge textbooks for the explicit explanations of the cointegration test. Because there you will come across several examples to aid your understanding.

[@CrunchEconometrix]

Hi Anil, because that is the way the model is specified if you want to perform the Bounds test. It is a test for cointegration, hence the depvar takes the difference operator as the short-run equation of the model. To get into the econometrics details, kindly refer to either Gujarati and Wooldridge textbooks for the explicit explanations of the cointegration test. Because there you will come across several examples to aid your understanding.

[@mulewachimanse86]

WHY DO I GET a response like this : unrecognized command :ardl When I input the ardl command on my stata 13

[@CrunchEconometrix]

Hi Mulewa, it implies that you are yet to install the command. Type any of these into the COMMAND BOX and follow the guides: ssc install ardl OR help ardl OR findit ardl.

[@mulewachimanse86]

thanks so much, does this also mean cusum6 applies?

[@CrunchEconometrix]

Yes, install cusum6 syntax.

[@rahmanakbi1905]

I have a database of 30 African countries from 1980 to 2017. The problem with the ARDL model does not give me results for analysis. Help me write the correct la commande de l'approche ARDL. Can you do that? the data are double-dimensional (countries) and time-based (periods) double-panel data. My question is: what are the correct commands to estimate my STATA data using the ARDL approach?

[@CrunchEconometrix]

Hi Rahma, you cannot get results because you have a panel data and the command you are using is for time-series ARDL. It won't work.

[@rahmanakbi1905]

@@CrunchEconometrix my data are data paneles ie the data are double dimension: individual dimension (the countries) and temporal dimension (the period from 1980 to 2017), and not time series.

[@fusonyBD]

The above-mentioned Google drive is not accessible.

[@CrunchEconometrix]

Yes. Check my website for resource files cruncheconometrix.com.ng/shop

[@anilraj9908]

Hi, I have run the ardl command without specifying the maxlags option and stata reverted with ARDL(4,3,4,4) regression for the variables used. But which is the correct way? Should we use varsoc command and run ardl with the lag mentioned by varsoc?

[@CrunchEconometrix]

Hi Anil, my Channel is still one of those that clearly explained how to estimate the ARDL model. So, simply follow the steps as shown. You also have access to my dofile located on my website (link is at the end of the video). Thanks.

[@omarahmed8957]

thank you very much for your very helpful videos. at the level of bounds test, it appears the message "expression too long". what is the solution please, thank you in advance.

[@CrunchEconometrix]

Hi Omari, I can't really say what the problem is. Better to use my dofile. It's available on my website.

[@reemelabd3994]

Thanks a lot for the videos.. I wanted to ask you, my model suffers from heteroskedasticity and serial correlation, how can I estimate ardl using HAC standard errors on Stata? Thanks again!

[@CrunchEconometrix]

I haven't used HAC standard errors but you can control these two problems using several combinations of measures: change regressors, take logs, estimate with higher-order lags etc. You can further explore options online.

[@reemelabd3994]

@@CrunchEconometrix Thanks a lot

[@arindashaman3648]

Thanks for the lecture. It has been so helpful. So in this case where the short run effects are not estimated by the ARDL model, what do you do?? do you conclude that there is no short term relationship?

[@CrunchEconometrix]

Arinda, your query is confusing. Know that if the Bounds test shows no cointegration, estimate the short-run ARDL model.

[@kevinongeri6487]

Greetings Dr.Ngozi I am very grateful for this video which is extremely educative. I have run an ardl model and i get an error message of Collinear variables detected. Is there a way of solving this issue? Thanks in advance

[@CrunchEconometrix]

Hi Kevin, watch my video on MULTICOLLINEARITY on guides to resolving this problem.

[@Ahmed-jl7uh]

When do we use the command ec1 instead of only ec?

[@CrunchEconometrix]

Ahmed, you may want to check out other online resources for more on that.

[@cessnasahu4483]

If my dependent variable is stationary at I(0) and all my independent variables are stationary at I(1) and my model is coming out to be integrated when I applied the johannson test, can I use a vecm model to interpret it or should I use ardl model?

[@CrunchEconometrix]

After watching this video, you should know that you can't use JCT.

[@cyrilchukwuka8378]

Hello Prof. I noticed that after performing the bounds test when the log of imp is the dependent variable, it showed cointegration but the command you used in performing the short and the long run model was still thesame with the command you used in performing only the short run(ardl). Any explanation for that? Thank you ma

[@cyrilchukwuka8378]

Basically I want to know what differentiates the stata command when there exists cointegration and when there is no cointegration

[@CrunchEconometrix]

The Stata ardl syntax are lumped up. So, you have to follow keenly for proper understanding. If there's no cointegration after using the "btest" syntax, run the ardl syntax without the "ec" option... Otherwise, invoke the "ec" option.

[@aaa89-u1p]

Hi, how did you decide whether the model has autocorrelation or not? Did you compare your DW value with DW critical value in DW table?

[@CrunchEconometrix]

Yes.

[@mohamedabdirahmanabdihamid8257]

Dear professor am confused why use 2 lags, although the optimum for aic showed only one lag. my optimum lag 4 when i tried to use 2 lags am seeing different results, what is the difference between maxlags and optimum lags can i use the varsoc optimum lags for the maxlags .

[@CrunchEconometrix]

Hi Mohd, please watch my video on OPTIMUM LAG SELECTION. Thanks

[@mohamedabdirahmanabdihamid8257]

@@CrunchEconometrix i did watched still i haven't found the answer am looking which is can i use the optimum lag for the maxlags when am running this code "ardl lnppi lngdp lnm3, maxlags(can i use my optimum lag here) aic" thank you.

[@CrunchEconometrix]

Mohd, the answer is in the video. Clear and quite understandable too.

[@iftekharimran9667]

Respected professor, thanks for your nice helpful tutorial. Is it mandatory to do a bound test for each of the variables individually? As in my model, there are 5 variables including dependent and independent variables.

[@CrunchEconometrix]

No need

[@iftekharimran9667]

@@CrunchEconometrix Thanks professor

[@iftekharimran9667]

Respected professor, I really appreciate your helpful tutorial. In my research, I found heteroskedasticity in the model stability checking part. So what will be the command for solving the heteroscedasticity problem? I also watched your heteroscadasticity part lectures but I don't know how can I solve it in the model stability part.

[@CrunchEconometrix]

Hi Iftekhar, thanks for the positive feedback. Deeply appreciated. You can adjust the lags, re-estimate the model and test for heteroscedasticity. You may also need to change your control variables if the problem persists.

[@CrunchEconometrix]

Hi Iftekhar, thanks for the positive feedback. Deeply appreciated. You can adjust the lags, re-estimate the model and test for heteroscedasticity. You may also need to change your control variables if the problem persists.

[@adisuabebaw5518]

How can we extract the long run and short run results from ARDL model using EViews 10?

[@CrunchEconometrix]

You obtain short-run results from the "ARDL" syntax, and long-run results from the "Bounds test and long-run form" syntax.

[@busaritajudeen9168]

Sorry what i tried to say is, How can heteroscedasticity and serial correlation be corrected in ARDL model in Stata?

[@CrunchEconometrix]

You can take logs and estimate model at higher-order lags.

[@seanh19954]

why do you do an aic for 2 if varsoc gave you optimal lags of 1 of Lnmva? as my model gives me 1 as well but if i run an AIC of 2 like you did, the number of lags i get is different to that of running an AIC of 1

[@CrunchEconometrix]

It's the way the Stata algorithm works. It drops a lag during estimation.

[@rupanandawidanage947]

Dear Professor I use STATA 15.1 , I used ardl lCGDP lCAGRIS, lags(1 1) to estimate the model. It says ardl cannot be recognized. Could you pls clarify

[@CrunchEconometrix]

Hi Rupananda, that is because you are yet to install the ARDL syntax. Type any of these in the COMMAND WINDOW and follow the prompts: ssc install ardl OR findit ardl OR help ardl.

[@rupanandawidanage947]

@@CrunchEconometrix Thank you so much professor I found it

[@amengizaw1970]

Hy Adeleye, i got your video very nice in addition to that the you teach and prepared video is a such very nice.you see i got a problem while doing my paper on determinant of current account deficit on ethiopian economy using ARDL.my problem is i am new for stata 14 commands regarding to performe my paper.10q inadvance i hope u will help me..

[@CrunchEconometrix]

Hi Gizaw, thanks for the positive feedback and kind remarks about my KZbin videos. Deeply appreciated! Kindly state your queries and I'll do my best to guide you correctly. Please may I know from where (location) you are reaching me?

[@amengizaw1970]

@@CrunchEconometrix from east africa ethiopia

[@sibylla553]

Hello Prof. Thanks for the informative tutorial. Im taking the log of variable and my F-statistics critical value in ARDL Bounds test is quite large as compare to the calculated value then is there any problem with the test?, Thank you in advance :)

[@CrunchEconometrix]

Hi Bylla, thanks for the encouraging feedback. Kindly interpret your result.

[@sibylla553]

@@CrunchEconometrix I observed through the period of 2000-2019 and used Stata's varsoc command with multiple variables (GDP, FDI, UNEM) and AIC, BIC, and SIC suggest 4 lags, but varsoc command with each variable, they suggest 1,0,1 respectively. I choose 4 lags in this case but the CUSUM is not stability, PLEASE ADVISE.

[@CrunchEconometrix]

Watch my video on the CHOW TEST about how to fix the problem.

[@sibylla553]

@@CrunchEconometrix Many thanks for your nice help.

[@anthonyclarke7993]

can the ardl model still be used even if the dependent variable is I(0)? the independent variables are a mixture of I(0) and I(1)

[@CrunchEconometrix]

Hi Anthony, that's a dicey question that the PSS (1999, 2001) left unclarified. I will say YES because I have published a paper with I(0) as the depvar. I've also seen a couple of papers do same too.

[@anthonyclarke7993]

@@CrunchEconometrix thank you so much for the speedy reply. I'm a Msc student and i have a paper working on. I have looked all over, finding info backing both against and for however, most of what I find says it won't pose an issue so i will work with it and use supporting literature to justify

[@anthonyclarke7993]

One last question, the same set of steps being shown in the video can be applied right? Same specifications at the first part and the same steps for running the tests in stata?

[@CrunchEconometrix]

Yes.

[@iftekharimran9667]

Respected professor, thanks for the nice helpful tutorial. Why you choose lags (1) in BGodfrey autocorrelation test? In my research, I found optimal lag (4), in this situation how many lags I need to choose in BGodfrey autocorrelation test?

[@CrunchEconometrix]

Hi Iftektar, you can run the test using the lag length for the model.

[@iftekharimran9667]

@@CrunchEconometrix Thanks professor for your nice comments.

[@joycentdouvi4082]

Good morning Madame. Please when I want to do the bounds test for cointegration test, the result coming is " expression too long" please I need your help

[@CrunchEconometrix]

Joycent, I'm not familiar with that error message.

[@joycentdouvi4082]

@@CrunchEconometrix Yes Madame, i have solved it. The error was coming from the moment I installed the Stata, i have chosen the option "small" without knowing. I changed it and it works. Thanks Madame for the answer

[@CrunchEconometrix]

Oh great! Glad to hear that it's all good.

[@busaritajudeen9168]

I have taken logs and higher-order lags in estimation of the models. lnPCE lnDpi and lnGDP Here is my diagnosis Results: .estat dwatson: Durbin-watson d-statistic (8, 287) = 1.893166 .estat bgodfrey, lags(1): chi2 = 3.734. Prob =0.0533 .estat imtest, white: chi2 (31) = 105.52 Prob=0.0000 Cusum graph was okay. Any solutions to heteroscedasticity and serial correlation? Hope I am not bothering my great tutor!

[@CrunchEconometrix]

Hi Busari, did you use the optimal lags during estimation?

[@benhatch1357]

Hi, as you ran the varsoc command and the optimal lag for the model came back as 1, how come you included 2 lags in the max lag command? Thanks

[@CrunchEconometrix]

Hi Ben, I did that because I wanted Stata to AUTOMATICALLY generate lags for the variables.

[@emblue000]

Can I use ARDL if my variables are I(3)

[@CrunchEconometrix]

Priscilla, kindly watch my video on "This is how to watch ARDL Model" for detailed answer to your query. Regards.

[@subhasishdas9045]

"command ardl is unrecognized"...now what should I do? Please help

[@CrunchEconometrix]

Hi Subhasish, it's an indication that you are yet to install the ARDL syntax. Do so with either "ssc install ardl" or "findit ardl".

[@mehdifarouki330]

Hello madmae thanks for this video please i have one question if you don't mind i have 5 variables when i use this lags (2 1 2 3 3) I don' find cointegration but when i use this lags ((4 4 0 2 2) i find F statistic > I 1 that mean i have cointegartion please Madame what lags i have to take? thank you

[@CrunchEconometrix]

Unfortunately I can't decide on that for you.

[@mehdifarouki330]

@@CrunchEconometrix it's incorrect

[@spinebuster9490]

So Prof, i wish to ask this. What did you use the varsoc optimal lags for? Also if there is no cointegration do you use the optimal lags for varsoc or from the matrix (e)lags?. Thank you.

[@CrunchEconometrix]

The lags from varsoc is used for the ARDL estimation....and yes, use the optimal lags from varsoc if there's no cointegration. In other words, you'll simply fall back on the ARDL model earlier estimated.

[@ChienNguyen-xr4mv]

Dear Prof, this video is so interesting. May you send to us this do-file? Thank you.

[@CrunchEconometrix]

Hi Chen, here's the link www.cruncheconometrix.com.ng/shop/ but the files are available upon payment after which you are allowed a one-time download. Thanks.

[@yanuozhou6028]

Dear sir, how do i export bounds test results? is there a command that i can use in stata?

[@CrunchEconometrix]

Yanuo, I don't export such results. I write them out. You can do same.

[@yanuozhou6028]

@@CrunchEconometrix Got it, Thank you so much!

[@benhalimaaziz7171]

please can explain how have you transforme the data to times series

[@CrunchEconometrix]

benhalima aziz Hi Benhalima, the data is initially a time series data so I did not have to transform it. Perhaps, you imply something else? Kindly clarify. Thank you.

[@benhalimaaziz7171]

the variale time is considred a s string variable

[@CrunchEconometrix]

Oh ok, use the dofile for data conversion on my website.

[@omduttdixit3433]

Hello, I am applying ardl model for model but the problem is that out my 7 variables only 3 are stationary even after first difference. Same problem with optimal lags can u help

[@CrunchEconometrix]

Hi Dixit, ARDL is not applicable. You can read up on the Toda-Yamamoto technique. It's the most applicable with I(2) series.

[@omduttdixit3433]

@@CrunchEconometrix thank you for reply. I Recheck my calculation and analysis tables and I found out my mistakes and found that all my variables are stationary in i(0) and i(1) . But after applying bounds test 3 out of 7 variables showing cointegration. Is ok to have more than 1 cointegrated variables in your modal.

[@CrunchEconometrix]

@@omduttdixit3433 You can only have ONE cointegrating relationship. But you can have more than one that is if you want to build a vector by taken each variable as a depvar.

[@omduttdixit3433]

@@CrunchEconometrix thank you for replying again . In one of your videos on ardl after checking for cointegration we have to use long run ardl model without difference operator. Can u explain this part of running ardl model without difference operator

[@busaritajudeen9168]

How to correct heteroskedasticity, serial correlation, in constitution diagnosis in stata

[@CrunchEconometrix]

...and what is constitution diagnosis?

[@sanjoysaha8149]

@@CrunchEconometrix can you give the data nad the do file of this analysis

[@CrunchEconometrix]

Hi Sanjoy, they are available on my website upon payment. Here's the link CrunchEconometrix.com.ng/shop

[@LonemaliZena]

Thank you Mam for such insightful videos. I have watched almost every video on time series data but I am still stuck. after testing for stationarity I used bound testing, the results show that I now have to estimate the short and long run relationships using ADRL and VECM, should I use all your commands in your DO FILE?

[@CrunchEconometrix]

Hi Zenani, the procedure is straightforward and well-explained. Follow what I did and do likewise.

[@LonemaliZena]

Thank you for the response. Durbin Watson shows that there is serial correlation between the variables. How to remove multicollinearity?

[@CrunchEconometrix]

@@LonemaliZena Remove multicollinearity? Or serial correlation? Both are not the same.

[@LonemaliZena]

Serial correlation

[@CrunchEconometrix]

@@LonemaliZena Perform the test with higher-order lags.

[@mariamsyed9935]

can i get the do file plz

[@CrunchEconometrix]

Hi Mariam, due to abuse and unethical conduct Stata dofiles used in my videos are no longer free but available on my website upon payment. Here's the link cruncheconometrix.com/view/datashop.php

[@zerubabbeldawit797]

Thank you for your smart lecture on youtube of ardl model. I want to have all vedios do files including the data. could you send me? I look forward to receive it soon.

[@CrunchEconometrix]

Hi Zerub, thanks for the compliments. You have access to all Stata dofiles used in my videos but not for all the data. They are already uploaded on my website. The link is at the end of the video. I'll also appreciate if you return the favour by sharing my YT Channel links with your students and colleagues. Deeply appreciated...

[@hameedkhan-de6rt]

how to run granger causality after we know that there is long run relationship

[@CrunchEconometrix]

Hi Hameed, please watch my videos on Causality Tests. You will find them very helpful. Thanks.

[@hameedkhan-de6rt]

@@CrunchEconometrix thank you so much dear

[@CrunchEconometrix]

hameed khan U're welcome Hameed😀...may I know from where (location) you are reaching me?

[@hameedkhan-de6rt]

@@CrunchEconometrix I am PhD student in China dear.

[@sawikifadhili148]

Helpfully video. .please send for me your do file

[@CrunchEconometrix]

Thanks for the encouraging feedback, Sawiki. Deeply appreciated! However, dofiles are no longer free but available on my website upon payment of a token fee after which you are allowed a one-time download. Here's the link cruncheconometrix.com.ng/shop/. Please may I know from where (location) you are reaching me?

[@yanuozhou6028]

Hello sir, i got error: "Number of gaps in sample = 1 sample may not contain gaps" after cusum6 command. What should i do in this situation? Thank you.

[@CrunchEconometrix]

Yanuo, this error message points to missing observations in your data.

[@yanuozhou6028]

@@CrunchEconometrix Thank you for your reply, sir! The missing observations are due to the fact that some of my original data has the value 0. After taking logs and differencing, the values are missing. In this case, I could not perform the cusum6 command. What should I do? Thanks again!

[@yanuozhou6028]

@@CrunchEconometrix Also, my dataset is quite small, so I don't think deleting the original data that has value 0 is an option for me. I was thinking about changing the missing values for my dlnY to 0, so that i can perform cusum6 command. I think it make sense mathematically, but is it an appropriate thing to do?

[@CrunchEconometrix]

You may want to change the variables with missing observations to their closer proxies. For instance, if "broad money" has too many "holes" you can use "liquid liabilities" or "financial system deposit" as replacement.

[@yanuozhou6028]

@@CrunchEconometrix Thank you!

[@busaritajudeen9168]

How to correct heteroskedasticity, serial correlation, in constitution diagnosis in stata

[@CrunchEconometrix]

I don't understand what you imply by "constitution diagnosis".

[@coordinatormdefcm3440]

Thank you for your smart lecture on youtube of ardl model. I want to have all vedios do files including the data. could you send me? I look forward to receive it soon.

[@CrunchEconometrix]

Thanks for your positive feedback. Deeply appreciated! Dofiles and datasets are available on my website at a token. Kindly find the link to the Store www.cruncheconometrix.com.ng/shop/

[@sawikifadhili148]

Helpfully video. .please send for me your do file

[@CrunchEconometrix]

Thanks for the positive feedback, Sawiki. Deeply grateful! I have responded to your request on a different thread.