(EViews10): Estimate and Interpret VECM (2)

Поділитися
Вставка
  • Опубліковано 11 січ 2025

КОМЕНТАРІ •

  • @CrunchEconometrix
    @CrunchEconometrix  6 років тому +4

    UA-cam 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, support my Channel with your subscription and sharing my videos with your cohorts.

    • @bachblues2
      @bachblues2 4 роки тому

      Looking forwards to your video on how to posit restrictions in VECM modelling.

  • @kojomensah4309
    @kojomensah4309 4 роки тому +1

    Thank you very much. So far, this is one of the best video on ADF, cointegration and VECM I have watched.

  • @gabpat7657
    @gabpat7657 5 років тому +3

    your videos are great, you explain everything very clearly and with a good mix of theory and how to do it in practice in Eviews. thank you very much!

  • @jannatulmaua4603
    @jannatulmaua4603 4 роки тому +1

    I am really thankful to you. Tomorrow, I have to given a presentation in zoom about inflation and economic growth in Bangladesh.I use all the econometric method in eviews and your videos are really helpful for me.your interpretation's are very helpful. I have several papers I want to work with you .

    • @CrunchEconometrix
      @CrunchEconometrix  4 роки тому

      Great 👍...let me know what you have in mind but I have strict rules about collaboration (based on past experiences). If you have a Scopus publication send to cruncheconometrix@gmail.com for evaluation...that's one of my conditions. Thanks and take care.

  • @Jaminn
    @Jaminn 6 років тому +2

    Thank you for this clear presentation. Currently writing my undergrad dissertation and it is very helpful

    • @CrunchEconometrix
      @CrunchEconometrix  6 років тому

      Thanks for the positive feedback on my video, deeply appreciated! 💕 Kindly share my Channel link with your students and academic networks. May I know where you are reaching me from?

    • @Jaminn
      @Jaminn 6 років тому

      @@CrunchEconometrix I'm from England

  • @SamWall93
    @SamWall93 5 років тому +1

    Congratulations, great channel! Thank you for being so clear! You have my support.

    • @CrunchEconometrix
      @CrunchEconometrix  5 років тому +1

      Wow! Thanks for the positive feedback and support for my content. Deeply appreciated! May I know from where (location) you are reaching me?

    • @SamWall93
      @SamWall93 5 років тому +1

      @@CrunchEconometrix I am from Milan, Italy. And I did my studies in Frankfurt, Germany. It's a just while I got the passion for econometrics, so cool and useful.

    • @CrunchEconometrix
      @CrunchEconometrix  5 років тому +1

      @@SamWall93 Awesome!!! Kindly share my YT Channel link with your colleagues and academic community in Italy and Germany. They will find the content on my Channel very helpful...thanks!

  • @BilalAhmad-em7no
    @BilalAhmad-em7no 5 років тому

    Respected Madam, I have Learnt a lot from your valuable
    videos specially in VAR environment. I am preparing electricity demand forecast
    using conventional multiple regression analysis since long. Now I am shifting
    to VAR model, I have gone through all steps
    1.
    lag selection criteria for each variable
    2.
    unit root test for each variable
    3.
    found I (1) Cointegration
    4.
    Combined lag selection criteria found to be I (1).
    5.
    Johansen cointegration showed one integrating
    equation with using 1 Lag
    6.
    Then I ran VECM with 1 lag
    7.
    Speed of adjustment come negative and
    significant
    8.
    No serial correlation, no heteroskedasticity,
    normality test is ok
    9.
    It also passed cusm stability test
    My VECM model is statistically ok.
    In Johansen Equation, we interpret it by changing the
    sign, while after changing the sign results are according to econometric
    relationships e.g. if price is increased, the consumption will decrease. My
    question is that in the VECM Long Term equation. the results are not coming
    according to econometric theory i.e. Price elasticity is not negative. so, is
    it ok? But in short Term Model in VECM the electricity elasticity is coming to
    negative.
    My next question is, now i have
    computed both Long run and short run equation from the VECM model and, I can
    compute the residuals ECT t-1 from the long-term equation and also,
    I know the coefficient of ECT i.e. speed of adjustment
    My data was from 1985 to 2019 and
    my independent variables are total GDP, real price of domestic category
    electricity and population of Pakistan and my target variable is domestic
    electricity consumption. I have taken
    all these variables in logarithmic form.
    now my request is how I can get the forecast of my target variables for
    my next 10 years while I can project the independent variables like GDP, price
    and population for next 10 years. but how can I make projection of ECT t-1
    for next 10 years by transforming this short run equation in a form that
    I can easily calculate forecast of my target variables. currently i am facing the
    issue that in equation my target variable is in differenced form which is the
    form of VECM equation.
    In OLS regression analysis after
    finding out the elasticity of the best equation we can easily transform this
    equation into
    Electricity Sale = (1 + Growth Rate GDP) GDP
    elasticity x (1 + Growth Rate Price) Price
    elasticity x (1 + Growth Rate Lag
    sale) Lag elasticity
    Is there some method to transform
    VECM equation like the above-mentioned equation to compute the forecast?
    Can you tell me if
    there is any option in EVIEWS that it can give us forecast for the period
    beyond the data if we provide to EVIEW the projection of independent variables?
    Normally this projection work doing in excel.
    Is it possible can
    you make a video on VECM forecast for next 10 years to understand the concept?
    Best regards

    • @CrunchEconometrix
      @CrunchEconometrix  5 років тому

      Hi Bilal, your query is TOO LONG. Kindly go straight to the point. Thanks.

  • @preetigarg2k7
    @preetigarg2k7 3 роки тому +1

    your videos are very helpful ...thanks

  • @TinaTina-xn9on
    @TinaTina-xn9on 4 роки тому +1

    oH my god!! I was in a real trouble in making a research assignment. You helped me so much, thank you.

  • @glennndjadila5124
    @glennndjadila5124 3 роки тому +1

    Thank you very much... from Namibia

    • @CrunchEconometrix
      @CrunchEconometrix  3 роки тому

      My pleasure, Glenn! Thanks for the encouraging feedback, deeply appreciated!

  • @MaskimReal
    @MaskimReal Рік тому +1

    Hello m'am. One question: In the short-run part, the CointEq1 of the Error Correction, If I have a value of 0.0783 where you have -0.067 in min 02:00, is this correct? I read that the value should be between -1 and 0 but I don't know if this is correct. The long-run coefficients are significative, but this 0.0783 it's not.
    This 0.0783 is placed where you have the -0.067 in the short-run equation.
    Thank you very much m'am.
    Best regards!!!

  • @SuperReddevil23
    @SuperReddevil23 6 років тому

    Thank you so much professor, amazing explaination. You have made things so easy for me. Please continue posting videos on empirical econometrics, you are too good. Cheers and godspeed

    • @CrunchEconometrix
      @CrunchEconometrix  6 років тому

      SuperReddevil23 Hahahaha, u're welcome SuperReddev (but I'm a Super bright RedGunner😊)...and please tell others about my UA-cam channel🥂

    • @SuperReddevil23
      @SuperReddevil23 6 років тому

      Will surely do. One other question i wanted to ask you is that what is the implication of a statistically insignificant error correction term ?

    • @CrunchEconometrix
      @CrunchEconometrix  6 років тому

      SuperReddevil23 I will interpret it that to mean the evidence of a long-run equilibrium is not significant. You can also do a little Google search on it.

    • @SuperReddevil23
      @SuperReddevil23 6 років тому

      ok thanks

  • @deeptysarder6797
    @deeptysarder6797 4 роки тому +1

    U r perfect at explaining everything... U made my day

  • @happydays2419
    @happydays2419 4 роки тому +1

    thank you so much! your way of explaining has been such a great help.
    for my cointegration test- trace stat indicates 1 and max indicates 0. following the trace stat result i went ahead and estimated a VECM on lag 1 (i.e.2-1) .
    while cointegrating eq coefficients are statistically significant, i find ECT and all 6 variables' coefficients are statistically insignificant in the error correction estimation.
    1.how to interoperate this?
    2. do i go ahead and alternatively estimate unrestricted VAR instead (according to max stat result of no cointegration).

    • @CrunchEconometrix
      @CrunchEconometrix  4 роки тому +1

      Hi there, thanks for the encouraging words and feedback. Deeply appreciated! Try option 2 and see what the outcome is.

  • @sadafayesha7763
    @sadafayesha7763 4 роки тому +1

    Really helpful for my thesis....thank you sooo much

  • @mohammedseid4182
    @mohammedseid4182 5 років тому +2

    Thank you really you clarify my problem on ECM interpretation

    • @CrunchEconometrix
      @CrunchEconometrix  5 років тому

      Thanks for the positive feedback, Muhd. Deeply appreciated! 💕 May I know from where (location) you are reaching me?

  • @huiruchen5664
    @huiruchen5664 5 років тому +2

    Thank you so much for the videos!!! Extremely helpful!

    • @CrunchEconometrix
      @CrunchEconometrix  5 років тому

      Thanks for the positive feedback, Huiru. Deeply appreciated! 💕 May I know from where (location) you are reaching me?

    • @huiruchen5664
      @huiruchen5664 5 років тому +1

      @@CrunchEconometrix Arkansas :) I'm a phd student of University of Arkansas.

    • @CrunchEconometrix
      @CrunchEconometrix  5 років тому

      @@huiruchen5664 Good to hear, Huiru! 💕 Kindly spread the word about my videos to your friends, students and academic community in Arkansas. Thanks! 😊

    • @huiruchen5664
      @huiruchen5664 5 років тому +1

      @@CrunchEconometrix Will do! Thanks again for your videos!

  • @sthmdy8859
    @sthmdy8859 Рік тому +1

    1. In the long run part of ECT model, the coefficients of regressors must be interpreted in reverse signs. 2. When we run VECM practically, reduce one lag for D(Endogenous).

  • @chikezieezenna1629
    @chikezieezenna1629 11 місяців тому +1

    Thank you very much for the class, please how do we get the durbin watson and p-value for each variables on the e-view?
    Because authors provide these info in the data presentation.

    • @CrunchEconometrix
      @CrunchEconometrix  11 місяців тому +1

      Thanks for your encouraging feedback...deeply appreciated. EViews displays the Durbin Watson statistic in most its results...as for p-values, you may have to use the t-stat to determine the statistical significance of the coefficient.

  • @SherMehta
    @SherMehta Рік тому +1

    Great tutorial on VECM. Only one query, the model is showing that in the long run, a 1% change in GDP is associated with 0.099% decrease in PDI. While statistically, the model is appropriate. However, in economics, a rise in GDP in the long run or the short run should result in an increase in personal disposable income (PDI), rather than a decrease in PDI. Can you please comment on this. It will help a lot.

    • @CrunchEconometrix
      @CrunchEconometrix  Рік тому

      Hi Sher, results are a function of several factors: variables, scope, and technique used. Besides, there are tons of empirical findings that contradict a priori expectations. Yours could be one of such.

  • @jhabindrapokharel
    @jhabindrapokharel 3 роки тому +2

    I watched most of the videos from CrnchEconometrix and found very much useful to the researchers in secondary data. Now i am interested in structural VAR estimation. I request you to make one video about i, if possible.

  • @SherMehta
    @SherMehta Рік тому +1

    Great videos on VECM. However, there is a clarification I seek. In economics, whether in the short run or the long run an increase in GDP should lead to a rise in PDI (personal disposable income). Whereas, this VECM model is showing a negative relatiopship i.e. a rise in GDP results in a decrease in PDI. Could you please explain this? Or was your attempt more to show the econometrric/statistics side of VECM rather than economic.

    • @CrunchEconometrix
      @CrunchEconometrix  Рік тому

      Hi Sher, I have responded to you on this on a another thread.

  • @babaywakhe
    @babaywakhe 5 років тому +1

    I love all your videos. very clear and easy to follow. I need the link to a video on step 4 (When there is no cointegration)

    • @CrunchEconometrix
      @CrunchEconometrix  5 років тому

      Thanks Zama, estimate VAR when JCT shows no cointegration. Click on the Playlists to watch my videos on the procedure. May I know from where (location) you are reaching me?

    • @babaywakhe
      @babaywakhe 5 років тому +1

      @@CrunchEconometrix Thank you. i am watching your videos all the way from Eswatini (formerly known as Swaziland)

    • @CrunchEconometrix
      @CrunchEconometrix  5 років тому +1

      @@babaywakhe Awesome! Please spread the word about my videos to your students, friends and academic community in Eswatini! 💕 😊

  • @denistiyo7193
    @denistiyo7193 4 роки тому +1

    How do you proceed when you obtain non-uniform results during johansen cointegration test i.e you find that there is no cointegration under trace stats but there is cointegration under max eigen value? will you estimate a VAR or VECM?

    • @CrunchEconometrix
      @CrunchEconometrix  4 роки тому +1

      Hi Dennis, I mentioned in my video on Johansen Cointegration Test that researchers take decision when the results conflict which guides subsequent approach.

    • @denistiyo7193
      @denistiyo7193 4 роки тому

      Thanks.

  • @warcatbattalion
    @warcatbattalion 4 роки тому +1

    4:27 I don't understand this part. the coefficients have low t-stats, so how am I allowed to put those coefficients into the equation when they're not significant?

    • @CrunchEconometrix
      @CrunchEconometrix  4 роки тому

      Hi Ecksdee, you construct your equation regardless of whether the coeff is statistically significant or not. Confirm this from any econometrics text.

    • @danaeellina2645
      @danaeellina2645 4 роки тому

      @@CrunchEconometrix hi ! any text book suggestions to find this?

    • @CrunchEconometrix
      @CrunchEconometrix  4 роки тому

      @@danaeellina2645 You mean you don't have any econometrics textbook that you use for studying?

    • @danaeellina2645
      @danaeellina2645 4 роки тому

      @@CrunchEconometrix I do but I have not found one that says its OK that the error correction in 4.27 is insignificant. I need to back this up somehow so I would love a suggestion on one that says that we must include the error term irrespective of its significance!

    • @CrunchEconometrix
      @CrunchEconometrix  4 роки тому

      Alright.

  • @williammbiakop4438
    @williammbiakop4438 4 роки тому +1

    please how do we interpret VECM with 2 cointagrations equations? which equation to choose and how to interpret variable with1.000

    • @CrunchEconometrix
      @CrunchEconometrix  4 роки тому

      Hi William, as explained in my VECM and Johansen cointegration test videos, keep it simple by using 1 CE except you are on top of your interpretation then you can use more than 1 CE....I still use only 1 CE regardless of if the JCT yields more than 1.

  • @meriemgharsallah2692
    @meriemgharsallah2692 3 роки тому +2

    Dear Madam, thanks for your effort and explanation, my question is can i use pedroni and kao for panel data instead of using johansen ???

    • @CrunchEconometrix
      @CrunchEconometrix  3 роки тому

      Yes, you can. Johansen is NOT applicable to panel data. Kindly watch my videos on panel ARDL. Thanks.

  • @Unknown-ic1ko
    @Unknown-ic1ko Рік тому +1

    How do you know whether it is a significant relationship or not

  • @robert18081995
    @robert18081995 6 років тому +12

    Thank god its not heteroskedastic

  • @baovietle5958
    @baovietle5958 2 роки тому +1

    I have a question. if Granger test shows that there's no casuality relationship between two variables then do we have to stop the research?

  • @najeebkhan4246
    @najeebkhan4246 6 років тому +1

    coefficient values shows e (t-1) values in the running model as you hove shown after performing VECM ?. If it is significant and negative it means there is long run relationship exist.

  • @simeonlucky6703
    @simeonlucky6703 3 роки тому +1

    Thank for this inestimable lecture it is very precise and timely.
    If the long run relationship is to be discussed upon, is it the normalized Johansen cointegration that will be adopted or the lag ECT cointegration equation?

    • @CrunchEconometrix
      @CrunchEconometrix  3 роки тому

      Hi Simeon, I'm not quite clear about your query but the Normalization captures the long-run content of the model.

  • @bachblues2
    @bachblues2 4 роки тому +1

    Great video. Loved it. Got one question just to be sure. You described the lambda (speed of adjustment) and its meaning, but no mention as to its sign. But then noticed that in your example it was negative. Must it be negative to have some meaning? What happens when it's positive?

    • @CrunchEconometrix
      @CrunchEconometrix  4 роки тому +1

      Hi Douglas, I appreciate the positive feedback. Deeply appreciated. You skipped the prerequisite videos that contained the information you seek. Kindly watch them. But note that theoretically the sign must be negative. if positive it implies there is no convergence to long-run equilibrium and the model becomes explosive. Thanks.

  • @owenedgson7302
    @owenedgson7302 2 роки тому +1

    Hello, thank you for the videos. Are we not supposed to change the signs for the long-run model? So should it not be (+) 3.63? Or do we only do that in the Johansen part? I was told to flip the signs in the long run part.

    • @CrunchEconometrix
      @CrunchEconometrix  2 роки тому

      Hi Owen, please watch my video on Johansen Cointegration Test. It answers your question. Thanks

  • @chintamanigautam2523
    @chintamanigautam2523 3 роки тому +1

    Dear ADELEYE, would you please answer these two queries?
    (1) In my model all variables (both IDV & DV) are cointegrated at I(1). Thus I used VECM. While running varsoc , lag one [1] is suggested by all criteria. Now, while checking cointegration, what lag should I use? one or two? And what lag should I use in estimating vecm?
    (2) In my other model, cointegration exists but while running ecm error correction term is found insignificant, in this situation what should I do?

    • @CrunchEconometrix
      @CrunchEconometrix  3 роки тому +1

      Hi Chinta, for (1) I gave clear explanations. Kindly watch my VECM videos again. For (2), simply interpret your results. There is a long-run relationship (cointegration) but reversion to long-run equilibrium (EECM) is not statistically significant. Thanks.

  • @edmundudemba9031
    @edmundudemba9031 6 років тому +2

    Hello My Sister! Thumbs up!! you are making us proud here. I used this video as a specimen to estimate my vecm model and arrived at the same conclusion with you but I still need clarification concerning the issue of normality. You rightly said there that while some of the variables are normally distributed others are not and in general, the model is not normally distributed but seems you were a bit relief with the outcome of the heteroscedacity . Please, what should I do as regards the normality issue? should I just leave it or correct?

    • @CrunchEconometrix
      @CrunchEconometrix  6 років тому +1

      Thanks Ed, for the positive feedback! Normality is of no consequence compared to serial correlation and heteroscedasticity. You can ignore.

    • @1717ratul
      @1717ratul 5 років тому

      @@CrunchEconometrix Hi, Can you please give me some references for what you just mentioned so that I can mention in my research paper, I have a VECM which is free of autocorrelation and heteroskedasticity but two equations are not normally distributed. Thanks.

    • @CrunchEconometrix
      @CrunchEconometrix  5 років тому

      @@1717ratul You can always cite published papers (tons of them) that used the procedure.

    • @1717ratul
      @1717ratul 5 років тому

      @@CrunchEconometrix I couldn't find, that's why I asked you. Sorry to bother you, Ma'am.

    • @CrunchEconometrix
      @CrunchEconometrix  5 років тому

      @@1717ratul You mean you found no papers with VECM approach? That's very difficult to believe. Anyways, do a Google search.

  • @KavehAAdli
    @KavehAAdli 4 роки тому

    thank you for your great explanations about VAR and VECM. I have a problem with heteroskedasticity in residuals of my VAR and VEC model residuals, could you please guide me on how to deal with this problem?

    • @KavehAAdli
      @KavehAAdli 4 роки тому

      and according to your responses to other people, you suggested choosing higher lags to remove heteroskedasticity, therefore this is a violation of choosing suitable lag regarding information criteria

    • @KavehAAdli
      @KavehAAdli 4 роки тому

      in my data, the suitable lag for var model is 4, and the lag in which heteroskedasticity disappears is 25!

    • @CrunchEconometrix
      @CrunchEconometrix  4 роки тому

      Hi Kaveh, thanks for the encouraging feedback on my videos. Deeply appreciated! It appears that the choice of your variables does not respond to heteroscedasticity. You may request to change them to closer proxies and re-estimate.

  • @eleniyuan1921
    @eleniyuan1921 3 роки тому +1

    Morning Prof, may i know Error Correction Term (ECT) stand for any used?
    With it be same term to use for mine any other VECM analysis test??

    • @CrunchEconometrix
      @CrunchEconometrix  3 роки тому

      Hi Eleni, there's a lot to read up about regarding the ECT. Kindly get any econometrics textbook and other online resources. Thanks.

  • @NehaRajput-tv5op
    @NehaRajput-tv5op 2 роки тому +1

    What we should do in vecm lag selection if lag is 1

    • @CrunchEconometrix
      @CrunchEconometrix  2 роки тому

      Hi Neha, retain the one lag and put a note in your work on why you did that.

  • @Jm1__
    @Jm1__ 4 роки тому +1

    hi professor, thank you for the great video, it really helped, ill be sure to share your channel. I have a question regarding the optimal lag for the VECM model. my optimal lag is 7. does this mean for the VECM my optimal lag would be 6 (p - 1). If so which short-run dependent variable lag should I use. aka d(GDP(1)). D(GDP(2), D(GDP(3) etc..

    • @CrunchEconometrix
      @CrunchEconometrix  4 роки тому +1

      Hi there, yes use 6 lags for the VECM. Follow the guide as explained in the clip on how to proceed.

  • @srishtibatra9991
    @srishtibatra9991 2 роки тому +1

    Ma'am, why did we take t-1 subscript for the ECT model, not t minus anything else

    • @CrunchEconometrix
      @CrunchEconometrix  2 роки тому

      Srishti, that's the way the ECT is specified. I suggest that you scour the literature for detailed explanation about the ECT. Thanks.

  • @delphinehounye4328
    @delphinehounye4328 2 роки тому +1

    Hello! Thanks for videos... it's very helpful for me.
    I am doing panel vecm right now and my errors are non-normality distributed. Please help me to correct that non-normality.
    Thanks!

    • @CrunchEconometrix
      @CrunchEconometrix  2 роки тому +1

      Delphine, I don't have videos on panel VECM. You may want to check out other online resources. Thanks.

    • @delphinehounye4328
      @delphinehounye4328 2 роки тому

      @@CrunchEconometrix Alright! Thanks!

  • @takundamugwira7747
    @takundamugwira7747 3 роки тому +1

    Thank you very much for the videos!! Is it a problem because in my model, the probabilities are less than 5% level so there is serial correlation and on normality all components showed a high prob value hence are distributed normally however the total shows that residuals are not normally distributed. And on heteroskedasticity the prob value is 0.44 so the model is not heteroskedastic. So what should i do or its fine like that?

    • @CrunchEconometrix
      @CrunchEconometrix  3 роки тому

      Takunda, correct the model for serial correlation. Estimate the model at higher-order lags.

    • @takundamugwira7747
      @takundamugwira7747 3 роки тому

      @@CrunchEconometrix Thank you, much appreciated!! I used 6 lags for serial correlation only and starting from the 2nd lag up to the 6th, the model is now showing that there is No serial correlation. However, i wanted to know is it not a problem because from Var lag order selection criteria i was using 2 lags but i now change them to 6 lags on residual tests

    • @CrunchEconometrix
      @CrunchEconometrix  3 роки тому

      No, it's not. Put a note note in the text explaining what you did about the diagnostics.

  • @5魏
    @5魏 4 роки тому +1

    Thank you!! But what can we do if there is heteroskedasticity.....?

    • @CrunchEconometrix
      @CrunchEconometrix  4 роки тому

      You can initiate a battery of measures: re-estimate at higher-order lags, and/or change regressors.

    • @5魏
      @5魏 4 роки тому

      CrunchEconometrix Thank you! Can I re-estimate the equation by using the LS method but changing the covariance method to Huber-White?

    • @CrunchEconometrix
      @CrunchEconometrix  4 роки тому

      Not advisable. Use the VAR routine to estimate.

  • @mohapatraful
    @mohapatraful Рік тому +1

    As error term distribution does not follow normality so what can be done

    • @CrunchEconometrix
      @CrunchEconometrix  Рік тому

      Hi Arup, NORMALITY TEST under VAR-VECM can be overlooked because it's SYSTEM of equations. Focus more on the model passing both HETEROSCEDASTICITY and AUTOCORRELATION tests.

  • @BilalAhmad-em7no
    @BilalAhmad-em7no 5 років тому +1

    Thank You for your reply
    Now my specific questions are as follows
    Q # 1
    If in Johanson test lag length criteria tells us lag = p= 1 . So for VECM as you said in your vedio take lag as ( p-1). What should I do
    Q. # 2
    I have computed the VECM conventional equation with 4 variables like electricity sales, GDP, electricity real price, population, which is also pasted below. Kindly guide me with a video or with some other means, how I can transform this equation to compute the forecast of the electricity sales. The target variable is LDOM8519 electricity sales which is in differenced form.
    ∆ LDOM8519t = 0.611109 + 0.185012 ∆ LDOM8519t-1 + 0.5014 ∆ LGDPTOT8519(t-1) - 0.014722 ∆LDOMRP8519t-1 -24.51419 ∆ LPOP8519t-1 - 0.48071 ect(t-1
    )

    • @CrunchEconometrix
      @CrunchEconometrix  5 років тому

      Hi Bilal, my responses:
      Q1: In this case, use 1 lag since VECM cannot be estimated as a static model.
      Q2: My videos on "forecast error variance decomposition" will be helpful.
      Thanks.

  • @amilcartaddei4718
    @amilcartaddei4718 3 роки тому +1

    could you tell me some bibliographic references please? love the vid

    • @CrunchEconometrix
      @CrunchEconometrix  3 роки тому

      Thanks Amilcar, for the positive feedback. There are several papers on VAR/VECM procedure online that are freely accessible.

  • @moomoosowon3911
    @moomoosowon3911 2 роки тому +1

    Thank for your video 😍 I have a question. If there is serial correlation or the residuals are not normally distributed, what would we do? Can we do anything to fix the model?

    • @CrunchEconometrix
      @CrunchEconometrix  2 роки тому

      Hi MooMoo, there are several measures you can deplo: readjust the lags, change regressors etc and re-estimate the model.

  • @statistics5391
    @statistics5391 2 роки тому +1

    When interpreting result we have to change the sign

    • @CrunchEconometrix
      @CrunchEconometrix  2 роки тому

      You change the sign only for the Johansen long-run equation.

  • @pedromrfernandes
    @pedromrfernandes 5 років тому +1

    Thank you for the video. From this, how can I proceed to assess causality?

    • @CrunchEconometrix
      @CrunchEconometrix  5 років тому +1

      U're welcome, Pedro. Simply watch my videos on VECM and Causality Tests. May I know from where (location) you are reaching me?

    • @pedromrfernandes
      @pedromrfernandes 5 років тому +1

      @@CrunchEconometrixThank you very much for your answer. In the mean time, I've already watched almost all your videos and understood that. I'm from Portugal. Thank you

  • @mohapatraful
    @mohapatraful 2 роки тому +1

    Vecm will run level data or first difference data

    • @CrunchEconometrix
      @CrunchEconometrix  2 роки тому

      Hi Arup, difference variables. I explained and showed this in my VECM videos.

  • @ekundayoibrahim7865
    @ekundayoibrahim7865 6 років тому +1

    Thank you for your good work. Please, I have a time series, all are I(1) and cointegrated but my optimal lag is 1, can I still go on and run VECM given p-1 condition

    • @CrunchEconometrix
      @CrunchEconometrix  6 років тому +2

      Hi Dayo, this is not a problem since VECM cannot be a static model but must contain lags of the regressors. All you have to do is to maintain the 1 lag and put a footnote to explain why....and thanks for the kind words on my UA-cam Channel. Humbly appreciated!

  • @utamadima7767
    @utamadima7767 4 роки тому +1

    Thank you for your explanation, I really appreciate it!
    Further, I want to ask, what if the Cointegration Equation is not significant? Does it mean that both in the long run and short run/only short-run have no relationship between the dependent and independent variables?

    • @CrunchEconometrix
      @CrunchEconometrix  4 роки тому +1

      Hi Utama, if that is the case then the long run relationship is statistically not significant.

  • @dr.datetsovini1190
    @dr.datetsovini1190 2 роки тому +1

    Thank you, Mama

  • @sabrososh
    @sabrososh 4 роки тому

    Hello CrunchEconometrix. I appreciate a lot your contribution with these videos, they amazing. Thank you!
    Please can you help me with a question: Why ECT=-0.0670 is interpreted as "an adjustment speed of 6.7%", but for the short-run effect of PCE=0.44 you say "a percent change in PCE is associated with a 0.44 percent increase in PDI"?. Why using a different rule for interpreting long-run and short-run coefficients, or was it a small lapsus? Thank you again for your great videos!

    • @CrunchEconometrix
      @CrunchEconometrix  4 роки тому +1

      Hi Sabrososh, thanks for the encouraging feedback. Deeply appreciated! Those are the generic interpretations. Look up the the references listed at the end of the video or my papers listed on the Community Page. Thanks

  • @emirarefa
    @emirarefa 3 роки тому +1

    Thank you mam,
    May I ask about the t-statistics? How do we know the value in the t table? And how do we know the degree of freedom in VECM?

    • @CrunchEconometrix
      @CrunchEconometrix  3 роки тому

      Hi Emira, from the output the t-stats are in [ ]. On how to get that from the t-table, I will advise you read up HYPOTHESIS TESTING from any econometrics textbook for detailed guidance. Thanks.

  • @Waleedkhancooldevil93
    @Waleedkhancooldevil93 3 роки тому +1

    Mam if we found 3 cointegrating equations, then while running vecm, should we enter 3 or 1 in rank option? If i put one, will it make any difference. Or if i put 3 then how to interpret three cointegrating equations?

    • @CrunchEconometrix
      @CrunchEconometrix  3 роки тому +1

      Waleed, I mentioned that you keep it simple and use 1 CE.

  • @kenechukwunwisienyi6262
    @kenechukwunwisienyi6262 4 роки тому

    Thank you Dr. for this video. Pls how do I get to know the P-values of the coefficients since what is provided is the standard errors and t-statistics? thanks

    • @CrunchEconometrix
      @CrunchEconometrix  4 роки тому

      I can't recall. Use the t-stats.

    • @kenechukwunwisienyi6262
      @kenechukwunwisienyi6262 4 роки тому

      Thanks for the prompt response. But sorry to draw you back. What process am I to take with the T-stat? I really don't know

    • @CrunchEconometrix
      @CrunchEconometrix  4 роки тому

      Alright, Kenechukwu. You made me go into EViews10 at 2.30am. To obtain pvalues from VAR output: Go to ->>Proc, Make System, Order by variable, Estimate, 🆗.

    • @kenechukwunwisienyi6262
      @kenechukwunwisienyi6262 4 роки тому

      Lolzz. So sorry about the inconvenience Dr. I appreciate. Thanks for the info. I will try it out.

  • @praisedee7696
    @praisedee7696 4 роки тому

    Thank you so much ma for a good presentation. In a case where there are two cointegrating equations (that is none and at most 1 are asterisked) , do we interpret the D(...)(-1) or the D(...)(-2) of each variables for the VEC estimates?
    Also following the video preceding this one, in a case where we have the variables stationery at first difference but different max-lag lengths for the unit root test, is it still acceptable to use or not acceptable to use if the are different unit root lag lengths for the variables?

    • @CrunchEconometrix
      @CrunchEconometrix  4 роки тому

      Hi Praise, thanks for the encouraging feedback. Deeply appreciated! Follow the guide and explanations. Use 1 cointegrating equation. Same lag length is used for all variables.

  • @olesialebedivna7763
    @olesialebedivna7763 4 роки тому

    Thank you very much for your lessons! It is incredible and very helpful for me!
    Also, I would very much like to clarify a few points.
    Please tell me what we should do if after Lag Length Criteria and Lag Exclusion Test my lag intervals are 1 1 4 4 7 7? How exactly can I reduce these intervals by 1 lag for the VECM?
    In this case, I would also like to ask what exactly the number of lags I should use for the LM-test (if lag intervals are 1 1 4 4 7 7)?
    Thank you very much in advance for your answer!

    • @CrunchEconometrix
      @CrunchEconometrix  4 роки тому +1

      Hi Olesia, kindly watch the clip again and follow the guides shown. Do same for the video on diagnostics. Thanks.

    • @olesialebedivna7763
      @olesialebedivna7763 4 роки тому +1

      Thank you very much!!

  • @manojwickramasinghe5921
    @manojwickramasinghe5921 3 роки тому +1

    I have a question, Suppose if we get the no. of lags for cointegration as 1, then what are we going to use as the no. of lags for VECM, we can't use 0 (since we get an error) isn't it madam?
    P.s. I love your videos, they're very beginner-friendly and clear also they're really helpful for my analyses

    • @CrunchEconometrix
      @CrunchEconometrix  3 роки тому +1

      Hi Manoj, VECM cannot be estimated in static mode. Retain the 1 lag and put a note to that in your work. Thanks.

  • @macro_finance
    @macro_finance 4 роки тому +1

    Thank you for the video. So all the steps are done on stationary variables, even cointegration test?

    • @macro_finance
      @macro_finance 4 роки тому

      I have created d(ln_variable) for transformation purposes and further on used d(ln_variables). But should cointegration test be done with d(ln_variables) or level variables?

    • @CrunchEconometrix
      @CrunchEconometrix  4 роки тому +1

      Hi Ksenija, kindly watch my video on Johansen Cointegration Test. Well explained, thanks.

  • @jelsinrubenmatamoroscondor5413
    @jelsinrubenmatamoroscondor5413 3 роки тому

    How can I correct for autocorrelation, heteroscedasticity and normality problems?

  • @martinamarino8700
    @martinamarino8700 5 років тому

    Hello,
    Thank you for all your uself videos.
    Can you please clarify if you use a Normal distribution to assess the variables significance and if you use a two tail test?

    • @CrunchEconometrix
      @CrunchEconometrix  5 років тому

      Hi Martina, regression outputs come out as 2-tail tests.

    • @martinamarino8700
      @martinamarino8700 5 років тому

      @@CrunchEconometrix
      Thank you for the advice.
      Are you using a Student T-distribution to check the critical values?

    • @CrunchEconometrix
      @CrunchEconometrix  5 років тому

      @@martinamarino8700 I use the corresponding pvalues.

  • @md.abdullah-al-mamun1897
    @md.abdullah-al-mamun1897 4 роки тому

    I have some issues about this model..
    1.Is it necessary to include the ECM equation (raw equation,not computed one) in my article
    ?
    2. Though my speed of adjustment is negative but it is statistically insignificant ( p value is higher than .05) ,is it a matter of concern?
    3.Relevance of Macro Economic factors for the Indian Stock Market
    by Aman Srivastava
    (2010) order the equation in another way, was it wrong?

    • @CrunchEconometrix
      @CrunchEconometrix  4 роки тому

      Hi Abdullah, I've read your queries over and over again I still don't understand what you mean?

  • @mouanguissalagesnadege830
    @mouanguissalagesnadege830 5 років тому

    How to do if there are evidence of heteroscedasticity and the residuals are not normaly distributed in the model? And i have annual data with lag 1.

    • @CrunchEconometrix
      @CrunchEconometrix  5 років тому

      Hi Mouanguissa, you can explore a combination of measures: use logs, estimate with the "robust" option, change regressors etc. May I know from where (location) you are reaching me?

  • @pedromrfernandes
    @pedromrfernandes 5 років тому

    Thank you very much for your videos. Why did you interpret the ECT coefficient term from the VECM differently (as a log-log rather than as a log, I guess) from the other coefficients? It was a lapsus-linguae?

  • @mehdifarouki330
    @mehdifarouki330 4 роки тому

    I have two integrated variables of order 1 and two integrated variables of order 0 and I used the ARDL model.
    please how can i use causality between these variables

    • @CrunchEconometrix
      @CrunchEconometrix  4 роки тому

      Watch my video on "ARDL-ECM and 3 ways causality checks".

  • @nutansingh4752
    @nutansingh4752 3 роки тому

    Thank u so much as u have explained the concept very well. I have a query that my vecm is not fitted in residual diagnosis. There is no autocorrelation but rest two conditions are not met. What can I do ? Kindly help me out.

  • @diegopillihuaman2550
    @diegopillihuaman2550 6 років тому +1

    hello profesor, i have monthly data for 14 years. in Johansen test I have at most 2 cointegration. when i run with the optimized lag, i have a lot of coeficients

    • @CrunchEconometrix
      @CrunchEconometrix  3 роки тому

      I explained in my video on JCT to keep it simple by using 1 CE.

  • @manalisharma9367
    @manalisharma9367 3 роки тому +1

    Mam I have 5 variables and 4 cointegrating equations. I have taken 2 exogenous variables too. Mam can I proceed with the model and how to interpret 4 cointegrated equations

    • @CrunchEconometrix
      @CrunchEconometrix  3 роки тому +1

      Hi Manali, kindly watch my video on Johansen Cointegration and take my advise to "keep it simple". Use 1 CE. Thanks.

    • @manalisharma9367
      @manalisharma9367 3 роки тому +1

      @@CrunchEconometrix Thank you Ma'am

  • @ann_vocals
    @ann_vocals 2 роки тому

    After doing the VECM test I ran the diagnostic to see if there was heteroskedasticity and the residuals were found to be heteroskedastic. How can I correct it?

    • @CrunchEconometrix
      @CrunchEconometrix  2 роки тому

      Hi Ann, re-estimate the model at higher-order lags.

    • @ann_vocals
      @ann_vocals 2 роки тому

      @@CrunchEconometrix ok Thank you :)

  • @bentoumathew6240
    @bentoumathew6240 6 років тому

    I have a problem with my regression. I'm using Eviews as the econometric software and I'm running Vector Error Correction Model (VECM). I have a problem of heteroskedasticity and autocorrelation. I don't know how to correct them in Eviews using VECM.

    • @CrunchEconometrix
      @CrunchEconometrix  6 років тому

      Hi Bentou, correction will require several measures not one. You can try increasing or decreasing the lags, changing regressors or control variables.

  • @rezagod7951
    @rezagod7951 5 років тому

    Thanks for the very useful video.
    I was wondering if we can use Granger causality and VECM for the variables which are stationary at their level?

    • @CrunchEconometrix
      @CrunchEconometrix  5 років тому

      Reza, you have muddled up this query. You can't do VECM with level-stationary variables but Granger causality can be done regardless of the the level of integration of the series.

    • @rezagod7951
      @rezagod7951 5 років тому

      Now, how can I measure the long run causality? And, what should I do with them? Simple regression?

    • @CrunchEconometrix
      @CrunchEconometrix  5 років тому

      @@rezagod7951 Watch my videos on "VECM and Causality Checks". Well-explained and practice along with your data.

  • @moroccan1001
    @moroccan1001 3 роки тому +1

    Hi. I have 5 variable with cointegration.
    Can i choose between VAR and VECM ?
    Or only VECM ?
    Thanks

    • @CrunchEconometrix
      @CrunchEconometrix  3 роки тому

      Please watch my VAR and VECM videos. Well explained.

    • @moroccan1001
      @moroccan1001 3 роки тому

      I watched the video. I see in video that if the variable I(1) are cointegrated yo estimate var AND evcm.
      But its (and ) or (OR)

    • @moroccan1001
      @moroccan1001 3 роки тому

      @@CrunchEconometrix so i choose beteween var and evcm or both

    • @CrunchEconometrix
      @CrunchEconometrix  3 роки тому

      You may decide to show the VAR estimates before the VECM.

  • @ahmedtrabelsi3589
    @ahmedtrabelsi3589 4 роки тому

    hi , you said "a percentage change in pce is associated with 0.44 % increase in pdi on average ceteris paribus in the short run" but what you mean with change in pci increase or decrease ??

  • @duartegracamoreira
    @duartegracamoreira 5 років тому

    Hello!
    I have been following your videos and I want to thank you for the help they have given me in my dissertation.
    One question: How do you resolve heteroscedasticity if it arises on White's test?

    • @CrunchEconometrix
      @CrunchEconometrix  5 років тому +1

      You can take logs or change regressors.

    • @duartegracamoreira
      @duartegracamoreira 5 років тому

      @@CrunchEconometrix Thanks for your feedback! By the way, since the dummy variables are exogenous, how can we measure their impulse/response function?
      Thanks!

    • @CrunchEconometrix
      @CrunchEconometrix  5 років тому +1

      @@duartegracamoreira You cannot.

  • @tallyskalynkafeldens1753
    @tallyskalynkafeldens1753 4 роки тому

    Dear teacher, is there a heteroskedasticity test like that in Stata as well? Also, you have chosen the (No-Cross Terms) option in your test. There is the (With cross terms) option too. Would you kindly explain when should we use each of those two options?

    • @CrunchEconometrix
      @CrunchEconometrix  4 роки тому

      Hi Revolta, you can do further online readings on both concepts. Tx

  • @bentoumathew6240
    @bentoumathew6240 6 років тому

    Hello Prof, many thanks for the youtube lectures. I have a question please: my step 1 and 3 are all met but i have two integrating equations after doing step 3. I am a bit confused, shouldn't i have one cointegrating equation? many times i see one cointegrating equation. what do i do next? below text is copied from the table in step 3.
    Trace test indicates 2 cointegrating eqn(s) at the 0.05 level
    * denotes rejection of the hypothesis at the 0.05 level
    **MacKinnon-Haug-Michelis (1999) p-values

    • @CrunchEconometrix
      @CrunchEconometrix  6 років тому

      Keep it simple, use 1 CE unless you can interpret your results with 2 CEs.

  • @jackylin6281
    @jackylin6281 4 роки тому

    Hi professor. My three variables are all I(1). According to Johensen test, there are 2 cointegrations. However,when I construct VECM, C(1) is not significant and sometimes not negative.
    I have no idea where the mistake exists because I followed your method step by step.

    • @CrunchEconometrix
      @CrunchEconometrix  4 роки тому

      Hi Jacky, the outcomes we get have a lot to do with the variables used for analysis not necessarily because the econometric procedure is wrong....assuming things are done properly.

  • @TheMaxxit
    @TheMaxxit 4 роки тому

    Is the interpretation of the ECT coefficients in percentage because of the logs you take (lndpi, lnpce etc..)? Or would this be a percentage for normal variables as well?

    • @CrunchEconometrix
      @CrunchEconometrix  4 роки тому

      Adapt the interpretation given in the video.

    • @TheMaxxit
      @TheMaxxit 4 роки тому

      @@CrunchEconometrix At 4:30 you say that 'the previous years deviations from long-run equilibrium is corrected at a speed of 6.7%.' If it wouldn't be a log variable but a regular variable, would it just be 'corrected with a value of -0.067'? Thank you in advance

    • @CrunchEconometrix
      @CrunchEconometrix  4 роки тому

      Max, since you are not paying attention to my guide, you can easily clarify your doubts by reading interpretation of the ECT from any publication.

    • @TheMaxxit
      @TheMaxxit 4 роки тому

      @@CrunchEconometrix okay, I was hoping for just a short answer yes or no

    • @CrunchEconometrix
      @CrunchEconometrix  4 роки тому

      Max, you need to read to be sure of the information you need. You may not get a YES or NO answer all the time. I've done my best to guide you but you aren't paying attention. So, my honest and candid advice is that you read. I've made that clear in almost all my clips: video tutorials are not substitutes for reading. Please READ to strengthen your academic confidence.

  • @jamiepalborn5916
    @jamiepalborn5916 2 роки тому

    Hi, really appreciate this video! How do we know which of the variables are cointegrated with the dependent variable? or am I missing/misunderstanding something?

    • @CrunchEconometrix
      @CrunchEconometrix  2 роки тому

      Hi Jamie, kindly watch the clip again. Well explained. Thanks

  • @sharon246787
    @sharon246787 5 років тому

    How do you handle non-normal residuals as shown by JB test?

    • @CrunchEconometrix
      @CrunchEconometrix  5 років тому +1

      Hi Sharon, I'll give my usual response. I don't. I'm more concerned about serial correlation and heteroscedasticity. Once those are good, I downplay non-normality. May I know from where (location) you are reaching me?

  • @rizka_khr
    @rizka_khr 3 роки тому +1

    Hello maam I am here again. Thanks for the video again and I really thankfull, but I have another question. How to interpret the ECT Coefficient that has (+) positive sign? So, I have a significant ECT coefficient at 5% and the coefficient is 0.534676. Is the way to interpret it different compare to ECT coefficient with negative sign?

    • @CrunchEconometrix
      @CrunchEconometrix  3 роки тому

      Hi Rizka, positive ECT is interpreted in the opposite. No convergence to long-run equilibrium. Model is explosive.

    • @rizka_khr
      @rizka_khr 3 роки тому +1

      @@CrunchEconometrix thanks for the answer mam. You are so humble and answering each of the question here in comment section, I already transform my variabel into log natural now it shows there is one cointegrating equation, but from the VECM it shows that the ECT not significant from t statistics also p-value and the ECT has negative sign. So should I interpret like " there is long equilibrium between variabel but not statistically significant?"..Thank you again mam 🙏🏼

    • @CrunchEconometrix
      @CrunchEconometrix  3 роки тому

      Correct. Or simply "no significant reversion to long-run equilibrium".

    • @rizka_khr
      @rizka_khr 3 роки тому +1

      @@CrunchEconometrix thank you so much mam, really appriciate it, I am working on my undergraduate thesis..thanks a lot for all of your video about this method 🙏🏼💙

  • @noku_m6026
    @noku_m6026 5 років тому

    Hi again . 1. Is it possible to run Cointegration test using lag order 1 to 2 and VECM with lag order of 1. The different selection criterion suggested lag 1 & lag 2 for the model.

    • @CrunchEconometrix
      @CrunchEconometrix  5 років тому +1

      You have to stick with a selection criteria and use its optimal lag.

  • @elliechew618
    @elliechew618 4 роки тому

    In my cointegrating equation I also have '@Trend(71)' - what does this mean or how should it be interpreted? Thank you.

    • @CrunchEconometrix
      @CrunchEconometrix  4 роки тому

      Hi Ellie, I have no idea what that is. You may need to seek further clarity online.

  • @aristidemurwanashyaka8496
    @aristidemurwanashyaka8496 5 років тому

    regards how do interpret a positive short run equation coefficient

    • @CrunchEconometrix
      @CrunchEconometrix  5 років тому

      Same way you interpret long-run coefficients.

    • @aristidemurwanashyaka8496
      @aristidemurwanashyaka8496 5 років тому

      @@CrunchEconometrix i have watched the video over again and i cant seem to find the part where you interpret the long run coefficients. you just state that this is the equation that signifies long run cointergration

  • @kidanemariamgidey1046
    @kidanemariamgidey1046 6 років тому +2

    Thanks sister!!

    • @CrunchEconometrix
      @CrunchEconometrix  6 років тому

      You're welcome, Gidey...please share with your students and colleagues! :)

  • @jyotinair5529
    @jyotinair5529 5 років тому

    Hi Prof...thanks for this video...so many doubts of mine have been cleared and the way you demonstrated the interpretations was outstanding. I have one question - how do you decide on a target variable, when you have many variables in the Johanssen method? Should we go for a causality to determine the target variable? Look forward to hear from you. I am based in India...

    • @CrunchEconometrix
      @CrunchEconometrix  5 років тому

      Thanks for the positive feedback, Jyoti. Deeply appreciated! The target variable is the 1st variable listed in the VAR system. It is the variable upon which the JCT is normalized. Hope this helps. Kindly share my UA-cam Channel link with your friends and academic community in India 🇮🇳. They will find the content helpful too 😊.

    • @jyotinair5529
      @jyotinair5529 5 років тому

      @@CrunchEconometrix Thanks a lot Prof...I will certainly..all the best

    • @CrunchEconometrix
      @CrunchEconometrix  5 років тому

      Awesome! Thanks a million! 😀

  • @nesiya3493
    @nesiya3493 Рік тому +1

    ma'am can we do VECM in panel data ?

    • @CrunchEconometrix
      @CrunchEconometrix  Рік тому

      Nesiya, to the best of my knowledge, NO. But you can check out other online resources for more information.

  • @napariayuba4941
    @napariayuba4941 6 років тому

    Please What do I do if I conduct johannsen cointegration test for non-stationary series and r=n, that is full rank. should i just go on and conduct the VECM? Thanks

    • @CrunchEconometrix
      @CrunchEconometrix  6 років тому

      Napari, kindly watch the video on Johansen Cointegration, before this video to get a better understanding on what to do before performing a VECM.

  • @achrafbenssassi9946
    @achrafbenssassi9946 4 роки тому

    How can I interpret the coefficient of ECT if the sign is positive ?? Thank u in advance

    • @CrunchEconometrix
      @CrunchEconometrix  4 роки тому

      It implies that the model diverges. There's no reversion to long run equilibrium. Model is explosive.

    • @achrafbenssassi9946
      @achrafbenssassi9946 4 роки тому

      CrunchEconometrix so in this case I should interpret that like this or may I change my model . But after researching on other plateforme I found that it is possible if there’s a structural change or break is this right .

    • @CrunchEconometrix
      @CrunchEconometrix  4 роки тому +1

      There is no right or wrong. The result is what it is. Surf the web and you will find papers with positive ECT. You may also decide to re-model by changing or removing some regressors.

  • @sherifftouray3159
    @sherifftouray3159 5 років тому

    Thank you, Ma for the videos. It is possible for a VECM to be unstable when all the variables are I(1)?

  • @Kndzu
    @Kndzu 5 років тому

    Hi. Thank you so much for your assistance. I have been following your tutorial with no problem until I got to the part where I need to run the White Heteroskedasticity Test. Eviews is not responding and just gives me an error message saying: Positive or non-negative argument to function expected. Then the output I get from Eviews is gibberish in a way. Would you be able to advise what to do? I notice that when I run it with optimal lags of 1 (even though my optimal lags are 6 per the AIC) I don't have this problem.

    • @CrunchEconometrix
      @CrunchEconometrix  5 років тому

      Hi Khanya, then I advise you estimate with 1 lag. May I know from where (location) you are reaching me?

    • @Kndzu
      @Kndzu 5 років тому +1

      @@CrunchEconometrix from south africa. Thanks so much for your response :)

    • @CrunchEconometrix
      @CrunchEconometrix  5 років тому

      @@Kndzu Awesome, Khanya! Please spread the word about my UA-cam Channel to your friends, students and colleagues in South Africa for awareness. Thanks 😊.

    • @Kndzu
      @Kndzu 5 років тому +1

      @@CrunchEconometrix I already am. Thanks again:)

    • @Kndzu
      @Kndzu 5 років тому

      Hi Bosede. I am in need of your assistance again. I am running a VECM for my study and it appears the data has many outliers. To deal with this, my plan is to add dummy variables.
      Would you mind advising if there is any best practice on how to decide where to add dummy variables?

  • @bentoumathew6240
    @bentoumathew6240 6 років тому

    Hello Prof, how can I solve the problem of heteroskedasticity and autocorrelation in VECM?

  • @ismashasyaizzati3822
    @ismashasyaizzati3822 5 років тому

    Hello, may I know how to estimate the unrestricted var model? Because from johansen cointegration test, I got no cointegration.

    • @CrunchEconometrix
      @CrunchEconometrix  5 років тому

      Hi Ismashasya, watch my videos on VAR and follow my guide. Please may I know from where (location) you are reaching me?

  • @Reine111
    @Reine111 5 років тому

    Ma, in the preceding video, you did not difference the variables before running the Johansen test, after they proved to be I(1). May I please know why? Thank you

    • @CrunchEconometrix
      @CrunchEconometrix  5 років тому

      Hi Reine, JCT is performed on the variables in their natural and not transformed form. Refer to any econometric textbook on the fundamentals of cointegration. May I know from where (location) you are reaching me?

  • @babaywakhe
    @babaywakhe 5 років тому

    Hi Bosede, I just want to find out if the sign in the VECM output matters?

    • @CrunchEconometrix
      @CrunchEconometrix  5 років тому

      Hi Zama, I will say it does because you have a priori expectations for the signs and significance of the coefficients. May I know from where (location) you are reaching me?

  • @mminakazemi
    @mminakazemi 6 років тому

    Hi, what if we came all the steps correctly but then we have nonnormal residuals and heteroskedasticity? I appreciate if you help with it. regards

    • @CrunchEconometrix
      @CrunchEconometrix  6 років тому

      Hi Kazemi, there's no straight answer to your query since you've not estimated any model. But some corrective measure may require that you simulate several functional forms of the model, drop a variable and replace with closer proxies etc

  • @michaeljeremi9493
    @michaeljeremi9493 5 років тому

    thank you for your video, can i ask u something, i still confused about how many max lag i should take. If i have annual data, should i take 1 max. lag?

    • @CrunchEconometrix
      @CrunchEconometrix  5 років тому

      Hi Tante, nothing to be confused about. If you have sufficient time-span, use 2 to 4 maxlags. But better still, you should watch my video on "Optimal Lag Selection" for more information. Please, may I know from where (location) you are reaching me?