Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: Please Help: fixed effects and indicator variables

From   Sangho Kim <>
Subject   st: Please Help: fixed effects and indicator variables
Date   Mon, 28 Oct 2002 10:38:34 +0900

Dear Statalist,

I am estimating life-time income by adding annual incomes. Based on the
annual data
(1984-99), I need to estimate annual incomes for the years which are not
covered in
my panal data.

Wit = XitB + Ai + eit.

For the estimation of the annual income W of individual i in year t, I
am using the fixed effect
model with indicator variables (job dummy). For the variables X, I am
using  fourth-order polynomials of age(age1, age2, age3, age4),
second-order polynomials of the number of children, an indicator
variable for female head and job dummies (worker1-worker10).

1) I am wondering if I can use ALL job dummies without omitting one of
them and at
the same time including a CONSTANT term in the fixed-effect model. As
long as I know, I can't use ALL job dummies and a Constant at the same
time because of the perfect multicollinearity.

My test results, however, show that the estimation result using "the
above method" is much better
that the one omitting one job dummy variable and including a CONSTANT.
Not only is the value of  R-squared higher (0.2 compared to 0.16), but
also the comparison of the estimated values with the real values for the
years between 1984 and 1999 shows that the estimated values using  "the
above method" are good estimates for the real values. By the way, a
Hausman test indicates that a fixed effect model is in favor.

I would really appreciate if someone could help me.

Sangho Kim
Kwandong University, Korea

*   For searches and help try:

© Copyright 1996–2017 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index