Page 1 of 1

Model Solve Static, Dynamic& Fit: producing the same result

Posted: Thu Dec 17, 2015 7:52 am
Hi,

I am trying to check the accuracy of my model. I have tried to produce one-step ahead, dynamic and fit forecasts within the sample, and they all produce the same estimates. I am assuming that I am missing something in the codes, but I have tried to compute them manually and the same happens. Please see the codes below ( I am making a new model for each of the forecasts type because I only discovered how to do it with the command "scenario" later on, but I have again tried creating different scenario and they continue to produce the same estimates no matter what the d= identifity is[code][/code]).

Marta

' MODEL SOLUTION: ACCURACY OF THE MODEL
' a) ONE-STEP FORECAST- ESTIMATION FIT:
_sys_2sls.makemodel(_m_onestep)
_m_onestep.solve (d=s)
_m_onestep.makegraph(a,g=v,c) gr_onestep @endog
gr_onestep.display

' b) MULTI-STEP FORECASTS (OUT-OF-SAMPLE) FORECASTS from a sub-sample
smpl @last-60 @last
_sys_2sls.makemodel(_m_dynamic)
_m_dynamic.solve(d=d)
_m_dynamic.makegraph(a,c,g=v) gr_dynamic @endog
rename *lr_0 *lr_dynamic
gr_dynamic.display

' c) FIT ESTIMATION: values of the current period are used for the estimation. It is used to examine the fie of each of teh equation in the model when considered separately - ignoring interdependencies.
smpl @all
_sys_2sls.makemodel(_m_fit)
_m_fit.solve (d=f) ' d = dynamicforecasts.
_m_fit.makegraph(a,c,g=v) gr_fit @endog
rename *lr_0 *lr_fit
gr_fit.display

Re: Model Solve Static, Dynamic& Fit: producing the same re

Posted: Thu Dec 17, 2015 10:38 am
Are there any dynamics or inter-equation relationships in the original estimation?

Re: Model Solve Static, Dynamic& Fit: producing the same re

Posted: Fri Dec 18, 2015 2:03 am
Not really. Could it be I am mis-specifying how I set the instruments up? Here is the system code:

system _sys_2sls

_sys_2sls.append DE_LR = C(1) + C(2)*DE_GDP + C(3)*DE_INF + C(4)*DE_SR + C(5)*DE_LRS @ DE_GDP JP_GDP JP_INF JP_SR UK_GDP UK_INF UK_SR US_GDP US_INF US_SR DE_SR(-1) DE_INF(-1)

_sys_2sls.append JP_LR = C(6) +C(7)* JP_GDP +C(8)*JP_INF + C(9)* JP_SR+C(10)* JP_LRS @ DE_GDP DE_INF DE_SR UK_GDP UK_INF UK_SR US_GDP US_INF US_SR JP_GDP(-1) JP_INF(-1) JP_SR(-1)

_sys_2sls.append UK_LR =C(11) +C(12)*UK_GDP +C(13)*UK_INF+C(14)*UK_SR +C(15)*UK_LRS @ DE_GDP DE_INF DE_SR JP_GDP JP_INF JP_SR UK_GDP UK_INF US_GDP US_INF US_SR UK_SR(-1)

_sys_2sls.append US_LR = C(16) +C(17)* US_GDP +C(18)*US_INF+C(19)* US_SR +C(20)*US_LRS @ DE_GDP DE_INF DE_SR JP_GDP JP_INF JP_SR UK_GDP UK_INF UK_SR US_GDP US_INF US_SR (-1)

' Solving the system with Two Stage Least Square Method:
_sys_2sls.tsls

Thanks a lot for your help!

Re: Model Solve Static, Dynamic& Fit: producing the same re

Posted: Fri Dec 18, 2015 5:48 am
Without any dynamics, a static and dynamic forecast will be identical.

Without any cross equation relationships, a fit will also be identical