Dear All,
I have seen econometric regression studies using panel data (e.g.
across countries and time) which include variables that are
measured with different time intervals, e.g. assume a wage
regression on quarterly data where one includes as regressors
unemployment and some indicator of labour market flexibility which
is only available at the one year frequency.
Or another, more extreme, example one may have panel data where
one of the regressors is only varying in the cross-section but not
along the time dimension. Again, what are the precise
consequences of that what including such a variable in a regression
where the dependent variable shows some variation over time.
I wonder whether this is statistically acceptable but I do not have any
references. Is there a "Moulton"-type of problem?
Thanks very much for any hints!
Yours
Bernd
*********************************************************
P.S. Interested in keeping up with current
developments in Monetary Economics?
Then why don't you join Nep-Mon, a free e-mail
list providing you with information about new
working papers in Monetary Economics.
Join now!
http://www.mailbase.ac.uk/lists/nep-mon/
*********************************************************
Bernd Hayo
Center for European Integration Studies (ZEI)
University of Bonn
Walter-Flex-Str. 3
53113 Bonn
Germany
Phone: +49 (0)228-73-1878
Fax: +49 (0)228-73-1809
E-mail: [log in to unmask]
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|