Hi Everyone! Thanks for watching! ✅ You can get the material for this video at: jdeconomicstore.com/b/var-stata-forecast ✅ You can contact me, or follow me on social media (IG, twitter, linkedin, etc.) at: juandamico.start.page Best Regards!
Hello Juan. Thank you for sharing an excellent video. I want to know the difference between "one standard error innovation" and "percentage innovations (impulses)". You mentioned in this video that you use percentage change (multiplying by 100). I saw many papers mainly using EViews use one SD innovation, and two SD innovations. Can we generate percentage impulses in EViews? I need your clarification. Thanks again for your informative videos.
My pleasure. A standard deviation is not the same than a percentage point. A SD cam be any percentage. Eviews I believe has an extra that you can download to express it in perdentage. But idk. Not sure about it. I mentioned in the video I used percentage (x100) so the results iand graphs would read 0.1 instead of 0.001. Im sure in the forums there may be a lot of info about this. Note: In our particular case, one standard deviation shock is 1 percent. Check the top left (That is a to money supply. Its 1%) and check the bottom right. Bottom right corresponds to prices and is also 1%. Sometimes the main diagonal is not 1%. Maybe its 2, 3, whatever. I hope that helps. Regards
Hi,whenever i am trying to do the forecast in stata i am getting r(198) everytime even though i have set the time variable "fcast compute fc,step(10) the time variable may not be missing r(198); can you please help ?
Hello Juan, thank you very much for your videos. I have a question for you, please. What platform do you prefer for time series analysis - EViews, Stata, R, or any other? Also, is EViews still relevant in the labor market?
I think that Stata is good for educational purposes. My experience is that most universities use Stata. R and Python are very practical as you can use it in a professional setting too. However requires more software knowledge than Stata or EViews. Finally, I know many firms who still use EViews.
Hi! I don’t recall if I tested for normality of errors in this video. Check the other videos I got too just in case. But to summarize, it’s not strictly necessary the normality of errors in var models. Regards
17:57 Hi! can I use as p-value threshold 1% instead of 5% in this autocorrelation test with Varlmar? Because in my case I get as p-value 0.012. therefore if I were to use as threshold 1% of p value i would be able to exclude the presence of autocorrelation, by not rejecting H0, right?
9:22 My lecturer said that if the trace statistics is greater than its critical value, so there is cointegration on the model. But why do you say in the video it's not cointegrated? please help me, i'm really confused. which one is true???
Hi. I would encourage to ask your lecturer. Chances are that the null hypothesis he is using differs from the null hypothesis that Stata uses. You can check Stata's criteria in the manual. It's here, check page number 8: www.stata.com/manuals13/tsvecintro.pdf ; Good luck!
Hello Juan, can you make an introduction to the software R. like you dot it with stata, eviews, and matlab_Dynare, ii think it would be good if you did the same for R. let me tell you that because of you I know how to use python learn! so many thanks!