MCGRAPHIRF—graphing end for error bands for IRF's

[TomDoan]

@MCGRAPHIRF organizes graphs for the output from various procedures which draw impulse response functions.

Detailed description

Among examples which use this are:

1. BOOTVAR.RPF (boot strapping a VAR)
2. MONTEEXOGVAR.RPF (Monte Carlo integration for a shock to an exogenous variable)
3. MONTESUR.RPF (Gibbs sampling for a near-VAR)
4. MONTESVAR.RPF (Gibbs sampling for an overidentified SVAR)
5. MONTEVAR.RPF (standard Monte Carlo integration for an OLS VAR)

By separating the tasks of generating the impulse responses (done by a variety of methods) and organizing the graphs (done by @MCGRAPHIRF), we can provide greater flexibility in the graphics.

If you want the calculations of the confidence bands without the graphs, use @MCPROCESSIRF.

[PTillmann-436]

Hi Tom,

How can I get the median responses out of the mcgraphirf procedure in order to ACCUMULATE them?

Thank you so much.

[TomDoan]

Use @MCPROCESSIRF for that. However, you really want to accumulate before computing medians—medians don't have any linearity properties.

[zw83189]

Hello Tom,

I am using RATS version 9.0 and used @mcgraphirf to generate impulse response. On my y-axis (dependent variable axis) the scale is showing 7 decimal places already. However these numbers are all zero so I would like to increase the number of decimal places. Is there a way to do that?
It might not have much meaning in interpreting such shock, but I just want to get an axis that has significant figure on the scale instead of all zeros.

Many thanks!!!

[TomDoan]

@MCGRAPHIRF doesn't have an option for changing the representation of the numbers on the axis. You can always use @MCPROCESSIRF to process the information and do the graphs differently. Are you doing some type of restricted model? In a regular VAR, the responses should never be true zero past the impact.

[zw83189]

I used a VARX model and shock the exogenous variables and examine the response of endogenous variables. In this case, it is normal to have zero response value after shock?

[TomDoan]

If the exogenous variable hits all the endogenous variables, no (as long as the lags are non-zero).