Skip to content

Forecasting using VECM

4 messages · Bert Gunter, Preetam Pal, Jeff Newmiller

#
Hi,

I have attached the historical dataset (titled data) containing numerical
variables GDP, HPA, FX and Y - I am interested to predict Y given some
future values of GDP, HPA and FX.

   - Some variables are non-statioanry as per adf.test()
   - I wanted to implement a VECM framework for modeling cointegration, so
   I have used *result = VECM(data, lag = 3, r = 1)* , and I get the output
   below showing that cointegration relationship does exist between these 4
   variables:
   - My question is: How do I get predictions of Y given
   externally-generated future values of the other variables (for say,
   upcoming 10 time points), using this result programmatically?

Regards,
Preetam
#############
Model VECM
#############
Full sample size: 25 End sample size: 22
Number of variables: 4 Number of estimated slope parameters 40
AIC 23.84198 BIC 70.75681 SSR 156.5155
Cointegrating vector (estimated by ML):
   GDP      HPA        FX           Y
r1   1 2.171994 -6.823215 -0.07767563


             ECT                 Intercept           GDP -1
Equation GDP 0.0612(0.0436)      0.0141(0.0687)      -0.4268(0.2494)
Equation HPA -0.6368(0.2381)*    0.1858(0.3749)      3.1656(1.3609)*
Equation FX  0.1307(0.0874)      -0.0039(0.1377)     0.1739(0.4997)
Equation Y   -0.0852(0.4261)     0.3219(0.6711)      -5.0248(2.4359).
             HPA -1              FX -1               Y -1
Equation GDP -0.0910(0.0790)     0.1988(0.2261)      0.0413(0.0299)
Equation HPA 0.4891(0.4311)      -2.2140(1.2337).    -0.3206(0.1631).
Equation FX  -0.2108(0.1583)     -0.2536(0.4530)     -0.0303(0.0599)
Equation Y   -0.3686(0.7716)     0.5234(2.2083)      -0.9638(0.2920)**
             GDP -2              HPA -2              FX -2
Equation GDP -0.2892(0.2452)     -0.0622(0.0563)     0.0598(0.1352)
Equation HPA -0.7084(1.3379)     0.1877(0.3069)      -0.2231(0.7377)
Equation FX  -0.1773(0.4913)     -0.0170(0.1127)     -0.2486(0.2709)
Equation Y   -3.8521(2.3948)     -0.4559(0.5494)     1.1239(1.3205)
             Y -2
Equation GDP 0.0411(0.0279)
Equation HPA -0.2447(0.1521)
Equation FX  -0.0102(0.0559)
Equation Y   -0.1696(0.2723)
-------------- next part --------------
GDP	        HPA	        FX      	Y
0.514662421	0.635997077	1.37802145	1.773342598
0.936722	3.127683176	1.391916535	3.709809052
0.101482324	1.270555421	0.831157511	0.226267793
0.017548634	2.456061547	1.003945759	9.510258161
0.236462416	0.988324147	0.223682679	5.026671536
0.372005149	2.177631629	0.904226065	4.219235789
0.153915709	4.620341653	0.033410743	3.17396006
0.524887329	1.050861084	0.518201484	7.950098612
0.776616937	0.503349512	0.666089868	3.320938471
0.760074361	3.635853456	0.470220952	6.380945175
0.802986662	1.260738545	0.452674872	1.036040804
0.375145127	0.20035625	1.837306306	6.486871565
0.002568896	3.532359526	0.556752154	8.536594244
0.754309276	3.952381767	0.247402168	8.559081716
0.585966577	4.01463047	1.184382133	0.148121669
0.39767356	1.553753452	0.983129422	5.378373676
0.859898623	4.73191381	0.828795696	3.367809329
0.741376169	4.993350692	1.758051281	5.516460988
0.329240391	3.465836416	1.701655508	1.249497907
0.078661064	3.298298811	0.04575857	5.132921426
0.270971873	0.46627043	1.739487411	4.94697541
0.731072625	0.940642982	0.728747166	7.583041122
0.385038046	3.51048946	0.021866584	7.361148458
0.530760376	1.204422978	0.415530715	1.163503483
0.555323667	4.777712592	1.844184811	8.596644394
#
Searching on "VECM" on rseek.org brought up:

"VECM" on the Rdocumentation site, which clearly states:

"The predict method contains a newdata argument allowing to compute
rolling forecasts."

If that is not what you want, you'll need to explain why not, I think.
If it is, please do such searching on your own in future.

Cheers,
Bert
Bert Gunter

"The trouble with having an open mind is that people keep coming along
and sticking things into it."
-- Opus (aka Berkeley Breathed in his "Bloom County" comic strip )
On Tue, Feb 14, 2017 at 4:18 AM, Preetam Pal <lordpreetam at gmail.com> wrote:
4 days later
#
Hey Bert,
The predict function in the link you mentioned does not seem to use
independently generated future values of the variables cpiUSA and cpiCAN in
calculating the future values of the variable of interest, i.e. dolCAN. As
I mentioned in the mail, I have the future cpiUSA and cpiCAN values
externally given to me (instead of generated by VECM), which I need to
use.Let me know if this explains what I am trying to get here. Thanks.
Regards,
Preetam
#
It will if you use it properly. Have you read the help for that function? You didn't show your code, and you didn't post your email using plain text, so we can't help much here.