Announcement

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

  • Lagged Dependent Variable

    Dear Forum Users,

    I am currently a student at a University in which we are practicing the use of Stata. Now I have gotten an assignment from my teacher, and I am at a loss. Unfortunately, I cannot post any results regarding the assignment, since the university doesn't allow that. However, it is mostly a theoretical question, but I cannot find the answer on the forum or anywhere else on google. Or at least not an answer that would help me understand the mechanism behind it.

    So, now on to the problem. Basically, we are using a fixed effects panel data analysis where we regress firm performance on several variables (one of which is GDP growth). Now the expected sign of the GDP growth is positive, according to prior research and economic theory. I am sure most of you know. As for the assignment, it tells us to lag our dependent variable, but now my GDP growth turns negative, while still being significant. While, other variables, such as firm size keep the same sign and stay significant.

    Hence, I was wondering why GDP growth (or any other variable) would change signs while still being significant after lagging the dependent variable?

    P.S. Just to give a clear understanding. GDP growth was significant and positive before lagging the dependent variable and becomes negative and significant after lagging.

  • #2
    Nick:
    the main problem here is that, with a lagged dependent variable, -xtreg, fe- gets inconsistent parameter estimates.
    I would consider -xtabond- instead.
    That said, moving from the theoretical to the applied side of the matter, please note that, even if your university would allow sharing data, this forum deprecates posting and replying about home/classworks (see the FAQ). Thanks.
    Kind regards,
    Carlo
    (Stata 19.0)

    Comment

    Working...
    X