Re: st: how to fill some gross wage missings if I have the net wage

["Stephen P. Jenkins" <stephenj@essex.ac.uk>]

> > On Wed, Dec 15, 2010 at 10:10 AM, Laura R.
<laura.roh@googlemail.com>
> wrote:
> >> for some reason, I would like to use the gross wage for my
> estimations. However,
> >> - only 30% of respondents have given both net and gross wage,
while
> >> - 70% of respondents have only given their net wage.
> >> These data are all from the same country and therefore underlying
the
> >> same tax laws. So on the basis those who gave both gross and net
wage
> >> and with the help of the net wage of those where the gross wage
is
> >> missing, can I impute the missing gross wages? If so, how could I
> >> implement this in Stata?
-------------------------------------

Please do not under-estimate the magnitude of the problems involved in
getting gross --> net and net --> gross calculations correct.  You
should be aware that large amounts of resources are devoted to this
task by statistical agencies and research organisations.  

For Europe, read e.g. documentation from Eurostat concerning the
European Community Household Panel and, more recently, the Statistics
on Income and Living Conditions.  There is also the EUROMOD project
which is a multi-country tax-benefit micro-simulation model (Google
will find references). 

There are of course US counterparts, and Austin Nichols was drawing on
those in his response to you. The NBER have a tax-benefit calculator
too (note the Statalist correspondence with Daniel Feenberg in recent
days).  As the correspondence indicates, what you want to do is a
potentially very big task. It depends of course on how much complexity
you want to take account of.

All the standard (non-life cycle) tax-benefit models that I am aware
of work with cross-sectional data -- panels don't provide additional
information for the calculations. So, it's a matter of repeating
calculations for each year in your panel.

Using regression methods to summarize the relationship between gross
and net income using data from the individuals -- or families ... it
depends on the tax unit in the country -- is one way to proceed, and I
think Eurostat have employed a version of that in the past.  But then,
there are also potential issues of selection (who provides both gross
and net and who doesn't?)   Maarten Buis was suggesting multiple
imputation after using a regression approach.  I understand the
rationale for MI, but I would suggest that there are many big issues
to address before getting to that -ice-ing on this particular cake.

All in all, don't expect there to be a quick fix.


Stephen
-------------------------------------------------------------
Professor Stephen P. Jenkins <stephenj@essex.ac.uk>
Institute for Social and Economic Research
University of Essex, Colchester CO4 3SQ, U.K.
Tel: +44 1206 873374.  Fax: +44 1206 873151.
http://www.iser.essex.ac.uk ; 
Survival Analysis using Stata:
http://www.iser.essex.ac.uk/survival-analysis
Downloadable papers and software: http://ideas.repec.org/e/pje7.html 


*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/