From peter.summers@ttu.edu Sat Mar 12 20:08:25 2011
From: Summers, Peter <peter.summers@ttu.edu>
To: gretl-users@gretlml.univpm.it
Subject: Re: [Gretl-users] Threshold VAR: Coding Help
Date: Sat, 12 Mar 2011 19:04:27 -0600
Message-ID: <8885DA157278484489B1EBC8B8C043E3FDC8A13364@COTTUS.ttu.edu>
In-Reply-To: AANLkTimjS4BFqw6p1vokbKXqpLkmUaYXCQZcOiJtnNLm@mail.gmail.com
MIME-Version: 1.0
Content-Type: multipart/mixed; boundary="===============8087543376277713066=="

--===============8087543376277713066==
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: quoted-printable

FWIW, in a univariate TAR model the data gets re-ordered by (say) ascending v=
alues of the threshold variable, then separate regressions get run on the 2 s=
ub-samples. Could this be done with an "msortby" command followed by smpl com=
mands?

PS

________________________________________
From: gretl-users-bounces(a)lists.wfu.edu [gretl-users-bounces(a)lists.wfu.ed=
u] On Behalf Of Henrique Andrade [henrique.coelho(a)gmail.com]
Sent: Saturday, March 12, 2011 6:48 PM
To: Gretl list
Subject: Re: [Gretl-users] Threshold VAR: Coding Help

Em 12 de mar=C3=A7o de 2011 Allin escreveu:

On Sat, 12 Mar 2011, Henrique Andrade wrote:

> *Y(t) =3D X(t-1) + e1(t); if z(t-d)>=3Dtau*
> *Y(t) =3D X(t-1) + e2(t); if z(t-d)<tau*

If it's a TVAR I _think_ it should look more like

Y_t =3D D1 + B1(L)Y_t + I_t*(D2 + B2(L)Y_t) + U_t

where I_t =3D 1 if z(t-d) >=3D tau, otherwise 0.

Dear Allin, I need that the errors (and variances) differ in each regime:

Y_t =3D D1 + B1(L)Y_t + U1_t,  if z(t-d) >=3D tau
Y_t =3D D2 + B2(L)Y_t + U2_t,  otherwise

So I think I need to use the threshold values to build two data segments. To =
do this I'm doing something close to your script:

<script>
series z =3D normal()
scalar tau =3D 1.5
series r1 =3D z(-1) >=3D tau
series r2 =3D 1-r1
loop foreach i Y
    series r1$i =3D zeromiss(r1 * $i)
    series r2$i =3D zeromiss(r2 * $i)
endloop
list regime1 =3D r1PAU r1IAU r1E
list regime2 =3D r2PAU r2IAU r2E
</scrip>

Now I just need to estimate two separate VARs:

<script>
 "VAR regime 1" <- var 1 regime1
 "VAR regime 2" <- var 1 regime2
</scrip>

But we have missing observations... How can we handle this?

Finally, I would like to make one more question. Looking at your original cod=
e, we have, in the last line, this expression: var 1 Y ; r2*. What the symbol=
 "*" (in r2*) means?

Best regards,
--
Henrique C. de Andrade
Doutorando em Economia Aplicada
Universidade Federal do Rio Grande do Sul
www.ufrgs.br/ppge<http://www.ufrgs.br/ppge>


--===============8087543376277713066==--