VAR-GARCH-M

[TomDoan]

Add sqrthoil{1} to the DET variables in the VAR.

[economics2012]

That's what I got. So, in this case the coefficients of interest psi1 and psi2 would be at the rows of BVEC(2)(9) and BVEC(2)(10) in the table below. Am I correct?

MAXIMIZE - Estimation by BFGS
Convergence in 84 Iterations. Final criterion was 0.0000017 <= 0.0000100
Daily(5) Data From 1986:01:09 To 2011:01:17
Usable Observations 6528
Function Value -23249.1656

Variable Coeff Std Error T-Stat Signif
************************************************************************************
1. B 0.003630837 0.004597287 0.78978 0.42965732
2. BVEC(1)(1) -0.020919103 0.013770255 -1.51915 0.12872437
3. BVEC(1)(2) -0.017907548 0.012138448 -1.47527 0.14013865
4. BVEC(1)(3) -0.025910185 0.013745882 -1.88494 0.05943772
5. BVEC(1)(4) 0.007095474 0.012098455 0.58648 0.55755455
6. BVEC(1)(5) 0.063694484 0.024966843 2.55116 0.01073641
7. BVEC(1)(6) 0.014516131 0.024590681 0.59031 0.55498267
8. BVEC(1)(7) -0.036974500 0.025254800 -1.46406 0.14317801
9. BVEC(1)(8) 0.003019756 0.023903111 0.12633 0.89946818
10. BVEC(1)(9) 0.000000000 0.000000000 0.00000 0.00000000
11. BVEC(1)(10) 0.029592218 0.038326175 0.77212 0.44004626
12. BVEC(1)(11) 0.020315489 0.088257235 0.23018 0.81794804
13. BVEC(1)(12) -0.100079017 0.078496639 -1.27495 0.20232817
14. BVEC(1)(13) -0.141430524 0.077981798 -1.81364 0.06973393
15. BVEC(1)(14) -0.003669281 0.076028871 -0.04826 0.96150770
16. BVEC(1)(15) 0.038306144 0.075008939 0.51069 0.60956974
17. BVEC(2)(1) 0.006861037 0.004349207 1.57754 0.11467190
18. BVEC(2)(2) 0.007140608 0.004188480 1.70482 0.08822779
19. BVEC(2)(3) -0.004294816 0.004409748 -0.97394 0.33008788
20. BVEC(2)(4) 0.001839274 0.004434995 0.41472 0.67834812
21. BVEC(2)(5) 0.050503683 0.013119201 3.84960 0.00011831
22. BVEC(2)(6) -0.018058224 0.012736882 -1.41779 0.15625208
23. BVEC(2)(7) -0.012341526 0.011994855 -1.02890 0.30352589
24. BVEC(2)(8) -0.029380022 0.009576370 -3.06797 0.00215518
25. BVEC(2)(9) 0.021714205 0.041558544 0.52250 0.60132450
26. BVEC(2)(10) 0.002524704 0.041359670 0.06104 0.95132524
27. BVEC(2)(11) 0.001704973 0.032186997 0.05297 0.95775512
28. BVEC(2)(12) -0.009456320 0.031045255 -0.30460 0.76067241
29. BVEC(2)(13) -0.017574618 0.030996118 -0.56699 0.57071814
30. BVEC(2)(14) 0.049693056 0.029394079 1.69058 0.09091696
31. BVEC(2)(15) -0.014526644 0.031511180 -0.46100 0.64479883
32. GARCHP(1)(1) 0.051517641 0.133695987 0.38533 0.69998980
33. GARCHP(1)(2) 0.387974507 0.202605512 1.91493 0.05550198
34. GARCHP(1)(3) -0.243766062 0.201661005 -1.20879 0.22674305
35. GARCHP(1)(4) 0.161764154 0.218537300 0.74021 0.45917075
36. GARCHP(1)(5) 0.015478597 0.236621772 0.06541 0.94784364
37. GARCHP(1)(6) 0.111068501 0.009412590 11.79999 0.00000000
38. GARCHP(1)(7) 0.878391771 0.009982869 87.98991 0.00000000
39. GARCHP(2)(1) 0.024217630 0.020492641 1.18177 0.23729617
40. GARCHP(2)(2) -0.052225550 0.029919436 -1.74554 0.08089105
41. GARCHP(2)(3) -0.010998038 0.035513138 -0.30969 0.75679727
42. GARCHP(2)(4) -0.066363255 0.025568311 -2.59553 0.00944458
43. GARCHP(2)(5) 0.088805125 0.035853755 2.47687 0.01325398
44. GARCHP(2)(6) 0.097831238 0.007462646 13.10946 0.00000000
45. GARCHP(2)(7) 0.890261638 0.008112625 109.73780 0.00000000

should I also set sqrthoil{1}=0 ?

Thanks a lot for all the help.

[TomDoan]

That's correct. To peg both sqrthoil parameters in the oil equation you would do:

nonlin b bvec garchp bvec(1)(9)=0.0 bvec(1)(10)=0.0

[economics2012]

I think I had an error in the previous code I sent.

Code: Select all

compute %%garchh=%zeros(%nvar,%nvar)
do i=1,%nvar
   compute %%garchh(i,i)=(garchp(i)(1)+garchp(i)(2)*M(time)+garchp(i)(3)*Tu(time)+garchp(i)(4)*W(time)+garchp(i)(5)*TH(time))+$
garchp(i)(6)*uu(time-1)(i,i)+garchp(i)(7)*hh(time-1)(i,i)
end do i
end

In the previous one I didn't multiply by (1-garchp(i)(6)-garchp(i)(7)). So, the below one is the correct one. Am I right?

Code: Select all

compute %%garchh=%zeros(%nvar,%nvar)
do i=1,%nvar
   compute %%garchh(i,i)=(1-garchp(i)(6)-garchp(i)(7))*(garchp(i)(1)+garchp(i)(2)*M(time)+garchp(i)(3)*Tu(time)+garchp(i)(4)*W(time)+garchp(i)(5)*TH(time))+$
garchp(i)(6)*uu(time-1)(i,i)+garchp(i)(7)*hh(time-1)(i,i)
end do i
end

Many thanks

[economics2012]

I mean that I need to multiply by (1-garchp(i)(6)-garchp(i)(7)) for the day-of-the-week effect to be parameterized in terms of unconditional variances rather than variance intercepts as discussed in the Baillie-Bollerslev article.

Correct?

[TomDoan]

I mean that I need to multiply by (1-garchp(i)(6)-garchp(i)(7)) for the day-of-the-week effect to be parameterized in terms of unconditional variances rather than variance intercepts as discussed in the Baillie-Bollerslev article.

Correct?

That's correct.

[economics2012]

Hi Tom,

I need to test using any information criteria for the number of lags of sqrthoil I need to add to the model. I added five lags but I need to test it.

Can you please help me doing it using RATS?

I can't thank you enough for all your help.

[TomDoan]

Hi Tom,

I need to test using any information criteria for the number of lags of sqrthoil I need to add to the model. I added five lags but I need to test it.

Can you please help me doing it using RATS?

I can't thank you enough for all your help.

I wouldn't really recommend using an IC to choose that. Just run the 5 lag and do sequential exclusion tests on the longer lags (with TEST).