Dear Statalists,
I am working on a project where I need to
1. estimate the reduced form VAR to predict the residuals ( the shock/ the innovation) and the contemporaneous terms ( details in picture)
2. estimate the long-run (t=15) cumulative return responses to unit shocks of the innovations in structured VAR ( theta_rm, theta_x, theta_r). The short-run restriction is that rm can affect x and r, x can affect r, while r cannot affect rm and x contemporaneously.
This is the reduced VAR This is the structured VAR This is the relationship of innovation between reduced and structured VAR, considering the contemporaneous terms.
After estimating all the parameters in reduced VAR, I need to get the cumulative impulse response function in structure VAR.
Here I have the problem:
1. What I am doing is basically deduct the contemporaneous terms in the variables of reduced form, and estimate the VAR all over again. Is it correct to do such procedure? Or is there other way to acquire the cirf at step =15 of structured VAR?
2. In the irf file, with 3 variables, there are in total 3^2 impulse-response relationships. Yet, I am asked to calculate the response to unit shock of each innovation. Does this mean the theta is a actually a vector? E.g. the response to unit shock of vwretd is the sum of response (vwretd sdvol_1 r_1)?
3. I noticed there is also -svar- in Stata. Is the mechanism of -svar- the same as my procedure from reduced VAR to structured VAR?
4. How can I return the response values in each step stored in irf file? I have checked the reference: chrome-extension://efaidnbmnnnibpcajpcglclefindmkaj/https://www.stata.com/manuals/tsirfcreate.pdf
With the reference, I tried
But get the error
Sorry the questions are a bit messily presented. The main questions I would love to ask are the problem 1-4.
I am working on a project where I need to
1. estimate the reduced form VAR to predict the residuals ( the shock/ the innovation) and the contemporaneous terms ( details in picture)
2. estimate the long-run (t=15) cumulative return responses to unit shocks of the innovations in structured VAR ( theta_rm, theta_x, theta_r). The short-run restriction is that rm can affect x and r, x can affect r, while r cannot affect rm and x contemporaneously.
This is the reduced VAR This is the structured VAR This is the relationship of innovation between reduced and structured VAR, considering the contemporaneous terms.
Code:
var vwretd sdvol RET, lags(1/5) predict resid_rm, residuals equation (vwretd) predict resid_x, residuals equation (sdvol) predict resid_r, residuals equation (RET) reg resid_x resid_rm rename b_resid_rm b_1_0 reg resid_r resid_rm resid_x rename b_resid_rm c_1_0 rename b_resid_x c_2_0
Code:
gen sdvol_1 = sdvol - b_1_0*resid_rm gen r_1 = RET - c_1_0*resid_rm - c_2_0*resid_x var vwretd sdvol_1 r_1, lags(1/5) irf create myirf, set(myirfs, replace) step(15) irf table cirf, impulse(vwretd) response(vwretd sdvol_1 r_1) individual step(15) irf table cirf, impulse(sdvol_1) response(vwretd sdvol_1 r_1) individual step(15) irf table cirf, impulse(r_1) response(vwretd sdvol_1 r_1) individual step(15)
1. What I am doing is basically deduct the contemporaneous terms in the variables of reduced form, and estimate the VAR all over again. Is it correct to do such procedure? Or is there other way to acquire the cirf at step =15 of structured VAR?
2. In the irf file, with 3 variables, there are in total 3^2 impulse-response relationships. Yet, I am asked to calculate the response to unit shock of each innovation. Does this mean the theta is a actually a vector? E.g. the response to unit shock of vwretd is the sum of response (vwretd sdvol_1 r_1)?
3. I noticed there is also -svar- in Stata. Is the mechanism of -svar- the same as my procedure from reduced VAR to structured VAR?
4. How can I return the response values in each step stored in irf file? I have checked the reference: chrome-extension://efaidnbmnnnibpcajpcglclefindmkaj/https://www.stata.com/manuals/tsirfcreate.pdf
With the reference, I tried
Code:
irf describe myirf mat list _dta(step) or mat list r(step)
Code:
matrix _dta(step) not found matrix r(step) not found
Comment