Variance decomposition with BVAR model

[TomDoan]

Thank you for reading the paper,

Just to explain the different questions that I ask, causing your confusion, let me explain to you what I want to do.

My application is a mix of different applications. I would like to determine the different sources of shocks for the Egyptian economic growth, so I'm estimating a BVAR that includes:

Foreign block (US GDP, Oil Price, Terms of Trade, US short run interest rate, Foreign direct investment to GDP) and domestic one (Domestic GDP, trade balance to GDP, and Inflation rate).

I use informative Minnesota independent normal wishart prior, using surgibbs procedure. The "block exogeniety" assumption is imposed via the prior, giving their coefficients a tight variance. So the model is a full VAR.

My first question was: should I estimate this model in level or in first difference. and your answer is that the BVAR's are generally done in levels because there is no advantage to doing them in differences or mixed levels and differences. And in this case the problem of cointegration is not relevant.

The problem is that you're mixing applications which typically are done using different forms for the data. A forecasting BVAR generally leaves data in levels for reasons I've explained. But in addition to that, you want to identify shocks using long-run restrictions. I'm not sure what exactly you expect those shocks to mean, but the calculation of LR restrictions generally requires estimation in differences.

The paper that I give you the reference, carries out the same economic analysis as mine, even if it's more simple. They use a mix of variables in level and in 1st difference. That's why I wanted to know if this could cause any estimation problems. And I didn't find your answer to this "may be there is something that I didn't get".

They needed everything to be at least somewhat stationary because they were using an informative prior on the mean of each variable, which doesn't exist for non-stationary series.

I use the Gibbs sampler for estimation, and I asked you in another post, if I can use the resulting new model for forecasting and comparing its performance to a classic VAR and I try to do this now.

Then, in order to identify this model I use short and long restrictions. Short for the foreign bloc and long for the domestic one. I get the IRF and FEVD.

I see that in some applications, you use simulate, but I don't know what is the real difference between simulate and forecast. Would you kindly explain me the difference.

SIMULATE is needed if you want to get the model's best guess as to the distribution of the future course of the series. FORECAST just gives the mean of that distribution (thus the point estimate for forecasts), and, in the case of a model estimated with Gibbs sampling, it can only give you the mean of the forecasts generated by one draw for the coefficients.

[AhmedSahlool]

Dear Tom,

I hope that finds you well,

Thank you for the explication, I think that I mixed different applications.

I wanted just to ask, if the main objective of my BVAR is to analyze the inter-relationships among variables; IRF and FEVD, with same variables and same prior, then should I keep variables in level? and Which identification scheme do you advise me.

Thank you in advance.

Ahmed

[TomDoan]

Is there anything wrong with using a standard Cholesky factor? As I said, I'm not sure how you would even interpret LR restrictions given the set of data that you have.

[AhmedSahlool]

Hi Tom,

I hope this finds you well,

Just an economic question:

If I use the Cholesky factor, is the order of the following variables okay:

US GDP, Oil Price, Terms of Trade, US short run interest rate, Foreign direct investment to GDP, Domestic GDP, trade balance to GDP, and Inflation rate.

I'm confused about the trade balance to GDP and domestic GDP variables.

Thank you for your time and help.

Ahmed

[TomDoan]

I'm not sure what you mean by being "confused". You don't know which is "correct"? Are their residuals highly correlated? If not, it doesn't matter. If they are, which seems to make more sense to you? You can also do both and see whether one gives results that make more sense than the other.