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

  • GARCH regressions for panels using foreach


    1. I want to estimate my risk measures using bi-variate diagonal GARCH (1,1) model (DVECH) for each of my 360 panels in the data(from a time period of 20 years). when I simply run my code, the error I get is that "sample may not include multiple panels". Could someone help me devise foreach code to extract coeffs, constants, t stat and p values.

    Also, exactly what I need to do is as below:

    "We begin by outlining a simple Gaussian framework under which risk measure(to be estimated) has a closed-form expression, and then present the estimation results. The Gaussian framework is a special case of the stylized financial system (X variable) with deterministic mean and covariance terms, and jointly normally distributed latent shock processes. In the Gaussian framework, risk measure is thus pinned down by three determinants: the correlation, the volatility of the financial system, and the Gaussian quantile"

    2. My variables data is not normally distributed, what techniques I shall consider to normalize it for GARCH modelling? The assumption is that both the Y and X variable follow a bi-variate normal distribution.
    Last edited by Aamina Khurraa; 04 Nov 2019, 10:52.

  • #2
    Clyde Schechter

    Hi Clyde, Since I didnot hear back anyhting from anyone, would you let me know any solution for the above #1. Also if I use the below code, how to loop it for each panel in my data?

    mgarch dvech ( Y = L.Y ) ( X = L.X ), arch(1) garch(1)

    where Y is the dependant variable and X is the independent variable.