Hello Everyone! I hope you liked the video. Please watch my VAR models videos for a deeper explanation on the estimation procedure and IRF/Variance decomposition analysis. ✅You can buy the Eviews Worfile + Video slides + dataset at: payhip.com/b/Uo7b 📈 Download the dataset for free and replicate the content of the video: www.jdeconomics.com/eviews-tutorials/Structural-VAR-models-in-EViews ✅ Visit my website to see all my FREE tutorials: www.jdeconomics.com ✅Subscribe to my channel by clicking: ru-vid.com/show-UC5P21WGFO4WRUlAiGLcwymg Thanks a lot!
thank you... you turned my 10 hours paper reading to 6 hour paper reading... I am a sloth reader plus i check the literature review on the go.... you helped me understand the VAR models. i cannot thank you enough...😇😊
How to estimate a svar under the Beaudry and Portier (2006) method, in which short-term and long-term restrictions are imposed at the same time? exists in Stata? Magnific work! Thanks so much!
I am very grateful for you, this video is very helpful. I want to ask about the difference between this long run structural VAR and structural cointegrating VAR? What is the difference between them?
Thank you for sharing the video. I have a query, for a model with 7 variables, how to impose short and long-run restrictions simultaneously in SVAR Matrix ( also, there are variables for which theoretically it is not possible to identify the impact)?
Hey, Thanks for your message. I would have to see what you are doing, but I would check with other similar papers. 7 variables seems a lot. You need to impose 21 resitrctions in the matrix if you use 7 variables. I don't recall many papers with 7. With four variables you need to impose 6 restrictions. Lastly, for the identification you would select it in the structural factorization option. Short run restrictions is just the cholesky decomposition, you can check my var videos for more details of the identification strategy of the cholesky decomposition. Finally, if you need to do any manual identifications, select the option manual and you can freeformat the matrix. Those values that have to be estimated, type "NA" in the matrix, and those that have no effects, 0. I hope that was helpful! Feel free to subscribe for more content! Good luck! JD Ec.
hello, thx for the video, it is very helpful and informative, hopefully, you will keep uploading new ones. Watching your video I noticed, that in your Var model, the R squared is approximately 10% respectively, which is very low and probably partly because of the differencing. My question is, how big of a role does R squared or adjusted R squared play in examining model accuracy in those kinds of models or more specifically in Vars to examine dynamic relationships between variables. thank you very much
Thanks for your comment! As long as my channel grows, I'll keep creating more content. In terms of your question, when differencing data you are removing that trend which may make two variables (x,y) have a common sense. Thats why R2 drops drastically. When working with complex models you are looking for the model to fit the data and how well it fits the data. Regards, JD
Thanks for your feedback! Unfortunately I haven't estimated one myself, so I couldn't tell you. I am planning to do a Johansen cointegration test + VECM tutorial in EViews sometime soon. Subscribe to my channel if you haven't so you get notified of the updates. Kind Regards, JDEc.
Hello. I love you videos! So helpful for my university classes if i don't understand my professor. But I have a question for you. I decided to make a economic growth model for my theseis and with all the scientific articles that I read i cannot tell which model is the best for economic growth. What is your opinion on this ? I thought at first at ARIMA but my professor said he recommends using an econometrics model and I am stuck..
Hey! Thanks Andrea. I see your channel has been growing since you first ever commented on my videos! haha. Congratulations. I strongly suggest you to email me at jdeconomics.inquiries@gmail.com so I can give you some ideas. Best Regards! JD.
Hello Juan, this is an excellent tutorial, And I've watched quite a few! Please keep this up! I have a question: by imposing the long-run restriction that the a nominal shock does not have any effect on the real exchange rate, aren't we sort of 'forcing' the model to give the result we want? I mean, we wanted to understand whether nominal variables affects real variables in the long run and in order to do so we intentionally impose the long-run restriction of no effect. Obviously, the model says that a nominal shock accounts for just about 2% of the variation in the real exchange rate (variance decomposition): but again, isn't this because of the restriction we imposed? How can we be sure that the no-nominal-impact is 'genuine' and not just 'artificially created' by our restriction? Hope it makes sense what I'm trying to ask you. Again, thanks a lot for your videos.
It's a very good point. And that's why you require an ad hoc assumption. It should have economic sense and when you obtain results, you always have to ensure it respects stylized facts. I have given very specific examples where long run restrictions are easier to see, (i.e., Money growth and output. We are ad hoc imposing our model that in the long run, printing money will only affect prices but not real output. Otherwise our model may suggest that you can print the stairs to heaven. The other case is inflation and unemployment. Short run vs Long Run Phillips.Curve. In the long run the curve is vertical. Changes in inflation only result in changes in wages). It is up to the econometrist to identify when the assumption is appropriate. It is always a good exercise to read existing literature because in the literature review section, the authors always cover and answer existing questions in the methodology. King and Watson paper and Blanchard and Quah are good starting points. Have a nice day! Best Regards! JD
Thank you for sharing your knowledge. May I make a question? When we apply structural decomposition in VAR models, the coefficients of lagged terms should be affected by that transformation (am I right?). However, these results aren't shown in the Eviews' output. How can I recover the structural parameters associated with lagged variables after the structural transformation?
yes, you will need at least one. Weird though that you would have so many I(0). Are you working with rates? I will be posting more tutorials soon, feel free to subscribe and stay tuned if interested! Regards, JDEcon.
Hello, thank you for your message and sorry for the delay in the response! I could include historical decomposition in a future video. To view the historical decomposition just click: View/Historical Decomposition. That's all. Leave as default the settings, and it will be displaying the graph for the whole data range. Best regards, JD.
@@JDEconomics Thank you . can you do it after a SVAR for example after a sign restriction or a short run /long run SVAR . If you do this that will be great Many thanks
@@rosenchowdhury66 Sure! Feel free to subscribe if you haven't so you are alert to my new videos. This week I will be teaching cointegration. Regards, JD!
Hi, it seems that the initial parameter values are producing a near singular matrix which results in the error you are getting. This can be fixed either by changing the assumptions/restrictions in your model, or changing the initial parameter values that EViews is using to solve the optimization problem. To change the initial parameter values, you go to structural factorization, (PROC>Structural factorization) and there should be a tab called "optimization control". You will find there the optimization values. By defaults, EViews uses 0.1, but there are other options you can try. Good luck! JD.
Hi...thanks for great video...so you think is it possible to apply this svar for panel data? please answer me it is really important and I will be thankful if you have any guide for me in this regard.
Hi, very good content! I am trying to model an SVAR of two variables, log of the real cooper price (I(1)) and mining investment as a percentage of GDP (I(0)), both series are non-cointegrated. My idea is to be able to see the effects of shocks in the price of commodities in mining investment, and for that I would like to be able to differentiate between permanent and transitory shocks to the price of commodities, but I don't know how to do that in Eviews. Can you help me? Thank you!
Hi! Thanks for the tutorials! I learned a lot. What happens if one of the variables is I(2)? Can we still use SVAR when all the rest are I(1)? Thanks again!
I understand that we have data for the nominal and real exchange rate, but what does it mean to apply a nominal and real shock to each? What is the actual shock? is it drawn from the data?
So, to clarify, while IRFs are computed using data to estimate the model’s parameters, the specific responses to shocks are not directly drawn from the data but are rather a result of the model’s structure and parameter estimates. They provide insights into how the variables in the model interact and respond to various shocks, helping you analyze the system’s dynamics. I hope the helps! Good luck, JD
@@JDEconomics thanks for the response! I read about something called the narrative approach. Would it be possible to apply dummy variables with sign restrictions in eviews to directly specify shock for certain time periods in which the dummy assumes the value 1?
Very easy way to explain SVAR..but the thing is you have checked the stationarity of the variables only using intersept but most of the time series have trend what you say about this?
Hello doctor, I hope you are well. Could you please put us a video related to the svar form for the short term in the eviews12 program that takes into account the s form and not the AB form?
Very good tutorial. One query : The graphs (~11) show a significant divergence of the real and nominal rates after 1995 or so. In the light of that, would it be appropriate to use a LR restriction ? I understand that we are trying to replicate the authors' model. However, there seem to be some structural break that needs to be analyzed ? Probably, the LR restriction worked till '95 and the the relation changed thereafter? Thanks again for the content
Hello! Thank you for the video. I am working on the paper How does monetary policy respond to exchange rate movements (Bjornland and Halvorsen,2014) They use a mix of sign and short-term restrictions. Is it possible to use this factorization decomposition? I have been trying to apply the sign restriction but Eviews does not understand commands such as (-, -NA). Do you have any suggestions?
Hello doctor, I hope everything is well with you. I'm getting a fail from Eviews 12 when I try to test SVAR. Error says that ' ''Optimization may be unreliable (first or second order conditions not met)''. Can you kindly comment how to solve this error. thank you.
Hey, it could either be that the restrictions are not specified properly, or also you can try manually Increasing the iteration limit and/or decreasing the convergence threshold. You can change in eviews the iteration limit. That may help the system find a solution. If that doesn’t work, then you may have issues with your variables and the Eviews can’t find a solution. Good luck
What is the key difference between impluse responses and accumulated responses?. When I didn't tick the accumulated response, the IRF graph was different.
As the name says, it shows the accumulated response to a shock. It will show basically the dynamics of the way it evolves over time. The simple irf shows the immediate response. Regards
@@JDEconomics but sir if we run data with monthly observation ( n=150) and the same data with annual observation ( n =150) then can we use long-run restriction in both cases as the number of observations is the same in both cases.
Hello, you can, but it will involve many restrictions in the identification matrix. There is no real rule of thumbs for how many variable you include, but I have typically seen 4-5 max. Regards, JD
Hello mister, i am working on twin deficit for Haiti monthly datas 1999 to 2015, i don't know if i will see the way to interpret some tests as normality test, and about impule functions it's totally dark, and i need it to make my paper really good for the presentation, please give me some advises to understand the impulse functions. Thanks in advance
Hello Wail, Thank you for your message. Feel free to subscribe for more content. In terms of research and models, you can contact me at jdeconomics.inquiries@gmail.com Regards, JD.
