## HDecomp (historical decomposition)

For questions about EViews Add-ins available from the EViews Add-ins webpage. Note each add-in available on our webpage will have its own individual thread.

Moderators: EViews Gareth, EViews Moderator, EViews Esther

Did you use forum search?
Posts: 1518
Joined: Thu Nov 20, 2008 12:04 pm

### Re: HDecomp (historical decomposition)

In the help document it says:
Resulting matrix will have (k*(k+1)) columns, k being the number of variables. Columns k*(i-1)+i will represent the base projections for ith variable and k columns next to these projections will include the impact of each shock to that variable.
So, the order of structural shocks do not change...

wahyoe106
Posts: 2
Joined: Fri Nov 14, 2014 5:12 am

### Re: HDecomp (historical decomposition)

Thanks for your clarifications.

dtideman
Posts: 8
Joined: Tue May 12, 2015 9:34 am

### Re: HDecomp (historical decomposition)

I conducted an SVAR in eviews using three variables: government revenue, government expenditure and gdp using the methodology of Blanchard and Perotti (2002).

In order to measure the government spending shocks would it be correct to take the structural residuals from the historical decomposition addin? Or should one of the variables from the histmat be taken?

I am using eviews 8,

thank you!

Did you use forum search?
Posts: 1518
Joined: Thu Nov 20, 2008 12:04 pm

### Re: HDecomp (historical decomposition)

dtideman wrote:In order to measure the government spending shocks would it be correct to take the structural residuals from the historical decomposition addin?

Yes, you should use histdecomp add-in and check the related box to obtain the structural residuals.

parulb
Posts: 9
Joined: Mon Sep 07, 2015 6:27 am

### Re: HDecomp (historical decomposition)

You have indeed made a very useful add-in for people like me working on VAR and SVAR. I have a question to ask, though. In the first study of my research work, I am estimating a SVAR with 9 variables. One of these variables is a monetary policy variable. Now in my second study, I want to estimate the impact of monetary policy shock (estimated from the first study) on firm level output. Using the HDecomp add-in, I was able to retrieve the structural shocks. But If I specifically want to retrieve the monetary policy structural shock to be used in the second study, how to do it- Right now I am getting 9 monetary policy structural shocks (one for each of the 9 variable used in my model). How to decide, which of these monetary policy shock to use?

Thanks
Parul

Did you use forum search?
Posts: 1518
Joined: Thu Nov 20, 2008 12:04 pm

### Re: HDecomp (historical decomposition)

parulb wrote:How to decide, which of these monetary policy shock to use?

viewtopic.php?f=23&t=5901&start=15#p39189

parulb
Posts: 9
Joined: Mon Sep 07, 2015 6:27 am

### Re: HDecomp (historical decomposition)

I had asked you the below query earlier too. But I still not very clear on what to do. My research is heavily dependent on obtaining the Monetary policy shocks in a proper manner. Will MP shock be just the data present in one column (as per the explanation provided by you in column 68, see the sequence of variables below) or will I would need to add MP shocks present in every variable. I wanted to send you my Eviews workfile to have a look, but I am unable to attach either excel or Eviews workfile.

The sequence of my variables are Oil prices, Global Growth rate, Fed rate, Domestic GDP, WPI, M3, MP indicator, NEER, Stock market variable.
_____________________________________________________________________________________________

You have indeed made a very useful add-in for people like me working on VAR and SVAR. I have a question to ask, though. In the first study of my research work, I am estimating a SVAR with 9 variables. One of these variables is a monetary policy variable. Now in my second study, I want to estimate the impact of monetary policy shock (estimated from the first study) on firm level output. Using the HDecomp add-in, I was able to retrieve the structural shocks. But If I specifically want to retrieve the monetary policy structural shock to be used in the second study, how to do it- Right now I am getting 9 monetary policy structural shocks (one for each of the 9 variable used in my model). How to decide, which of these monetary policy shock to use?

Thanks
Parul

parulb
Posts: 9
Joined: Mon Sep 07, 2015 6:27 am

### Re: HDecomp (historical decomposition)

I was able to attach the excel workbook. Pls see the highlighted column...under C68. Will I take the data under that column only as Monetary policy shock?

Thanks
PB
Attachments
historical decomposition_eviews.xlsx

parulb
Posts: 9
Joined: Mon Sep 07, 2015 6:27 am

### Re: HDecomp (historical decomposition)

Appreciate any help please.

Many thanks!!

Did you use forum search?
Posts: 1518
Joined: Thu Nov 20, 2008 12:04 pm

### Re: HDecomp (historical decomposition)

How do you define the monetary shock from the theoretical standpoint? Fed rate, M3 or MP indicator? And you also need to decide which variable are you particulary interested in analyzing the impact of a monetary shock. GDP, NEER or stock market?

marto91
Posts: 2
Joined: Sun Apr 10, 2016 10:02 am

### Re: HDecomp (historical decomposition)

I have a quick question. Is it possible to use hdecomp on a Var that is not structural? And would that make sense ?

Did you use forum search?
Posts: 1518
Joined: Thu Nov 20, 2008 12:04 pm

### Re: HDecomp (historical decomposition)

What do you think historical decomposition does? Yes, it is possible and would make perfect sense.

valleire
Posts: 1
Joined: Mon Jul 11, 2016 10:26 am

### Re: HDecomp (historical decomposition)

When using the historical decomposition, I am looking to identify the SVAR using the Choleski Decomposition which you have provided

However, there does not appear to be any place to provide the casual ordering necessary to identify the SVAR using the Choleski Decomposition approach. (This is present on the other hand on the impulse analysis and variance decomposition pre-programmed eviews options)

How do I set my recursive ordering, similar to the eviews options, with your add in using the Choleski Decomposition?

Many Thanks

Did you use forum search?
Posts: 1518
Joined: Thu Nov 20, 2008 12:04 pm

### Re: HDecomp (historical decomposition)

When using the historical decomposition, I am looking to identify the SVAR using the Choleski Decomposition which you have provided

However, there does not appear to be any place to provide the casual ordering necessary to identify the SVAR using the Choleski Decomposition approach. (This is present on the other hand on the impulse analysis and variance decomposition pre-programmed eviews options)

How do I set my recursive ordering, similar to the eviews options, with your add in using the Choleski Decomposition?

Many Thanks

There is no need to use SVAR if you'll apply recursive ordering as it is the default factorization of regular VAR. However, you can obtain cholesky decomposition within SVAR framework as follows: viewtopic.php?f=5&t=13431#p46638

kartikanindya
Posts: 3
Joined: Mon Jun 27, 2016 12:45 pm

### Re: HDecomp (historical decomposition)

Hi everyone,
I also want to use this Historical decomposition add ins on the structual shocks of oil price.
So, I have 4 structural shocks : oil supply shock, aggregate demand shock, oil-market specific demand shock, and other shock.
As discussed before, the result of the add-ins is histmat matrix shows 20C. However, I am confused on how translate this result to graph.
I read this thread and find that C1-C5 represent the first shock, and so on.
Do I have to create graph manually by adding C1-C5 into Shock 1, etc?