Easy Estimation Methods for Discrete‐Time Duration Models
DOI | http://doi.org/10.1111/j.1468-0084.1995.tb00031.x |
Date | 01 February 1995 |
Published date | 01 February 1995 |
OXFORD
BULLETIN
OF
ECONOMICS
AND STATISTICS,
57.1
(
1995)
0305-9049
PRACTITIONERS
CORNER
Easy Estimation Methods
for
Discrete-Time Duration
Models
Stephen
P.
Jenkinst
I.
INTRODUCTION
empirical analysis
of
duration data, discrete-time models have several
over continuous-time models.
For
example, one can straight-
forwardly estimate
-
without writing special programs
-
discrete-time
models combining both time-varying covariates and flexible specifications
of
duration dependence.' The recent introduction
of
duration analysis modules
into
widely-available econometric software packages has reduced these
advantages, but these modules remain limited.
In
particular, no
module allows modification
of
the likelihood function
to
take account of the
Of
sampling scheme used.? This paper
shows
how the "'serious but
occasional"
applied econometrician' (MacKie-Mason,
1992,
p.
165)
Can
estimate discrete-time duration models taking account
of
some empirically-
Important sampling schemes and
do
SO
using readily-available packages.
Although inferences about some underlying population are the usual
goal,
data are often derived, not from a random sample
of
that population, but
from
a sample
of
people flowing into
or
out
of
a
particular state
(flow
'ampling)
Or
from
a
sample
of
those occupying a state at a given time
(Stock
If
we do not control for this
-
e.g. by continuing to use existing
To continue reading
Request your trial