Announcement

Collapse
No announcement yet.
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • MGARCH Querry

    Hi Statalist
    I want to estimate conditional variance and covariance between two variables. Can somebody help me in answering the following questions
    1. Can we include moving average(MA) terms in the mgarch command
    2. How we can select the lag structure of arch and garch in the mgarch command?

    Your reply is much appreciated.

  • #2
    Hi,

    1) Moving average (MA) terms can be included in the mean equation using VARMA models, nevertheless this models are not available in STATA and are in some sense difficult to estimate given the need to impose restrictions to perform the estimation. And in the variance equation the MA terms of the variance are already included in a MGARCH model, given that a MGARCH model can be expressed as a VARMA model.

    2) There are some LM and portmanteau test that can be programmed to check if there are evidence of ARCH effects. The tests can be applied before you estimate a MGARCH on the residuals of some equations and after you estimate the MGARCH on the standardized residuals, if you find evidence of ARCH effects in the later case you should increase the ARCH and GARCH terms.

    Best regards,

    Comment


    • #3
      Hi,
      I'm using a multivariate VARMA (GARCH-BEKK) and I know that it can't be programmed by STATA. My question is econometric. I have always in my results an autocorrelation problem in the mean equation. I tried many solutions (superior p order and q in the VARMA terms, I increased the ARCH and GARCH terms) and the problem is always there!!! Note that my series are stationnary.
      Any advices please?

      Comment

      Working...
      X