Panel A in Figure 7 simply contains the 90-10 Log Wage
differential for each year. Panels B through D contain
a decomposition of the CHANGE in the differential over
time. So there are your two groups again (incomes in
year x vs. incomes in year y). The confusing thing in
the Figure is that the decomposition results are
reported as deviations from their overall mean. More
details can be found below in my answer to someone who
had a related question
ben
=======
it actually turns out that -jmp- is not so well suited to replicate the analysis in Juhn et al. (1993). The point is that I personally used the technique in another context when I programmed it. I am now thinking of changing the program so it follows Juhn et al. more closely. Meanwhile, a work around could be:
gen long id = _n
expand 2
bysort id: gen id2 = _n
areg ... if id2==2, absorb(year)
est sto pooled
capt matrix drop result
forv y = 1984/2004 {
reg ... if year == `y' & id2==1
jmp . pooled, stat(d9010) ref(2)
mat result = nullmat(result) \ r(stats1)
}
mat list result
The matrix "result" will contain for each year the d9010 statistic of the variables Y1, Y2 and Y3 as described in Juhn et al., where the coefficients and residuals from the -areg- command over the whole sample are used for the "average" coefficients and the "average" residual distribution.
Basically, Juhn et al. plot the contents of "result" after subtracting overall means. (Note that the output displayed by the single -jmp- calls in the code above should not be interpreted.)
=======
> -----Original Message-----
> From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-
> statalist@hsphsun2.harvard.edu] On Behalf Of kaya tas
> Sent: Tuesday, October 24, 2006 11:28 AM
> To: statalist@hsphsun2.harvard.edu
> Subject: st: JMP: Figure 7 In JMP(93)
>
> Is there a way to get JMP to give me a decomposition of the difference
> between two quantiles of the wage distribution in a single year? I am
> talking about Fig 7. in the JMP(93) paper. I know it can be done if you
> have
> 2 different distributions like male vs. female. In this one, though, there
> is only one distribution. Or am I missing something here?
>
> Great program, by the way.
>
> Thanks,
> Kaya
>
> _________________________________________________________________
> Sadece sohbet ile yetinmeyin - eglneceye de doymak için Messenger'i tercih
> edin! http://messenger.msn.com/?mkt=tr&DI=3490&XAPID=2584
>
> *
> * For searches and help try:
> * http://www.stata.com/support/faqs/res/findit.html
> * http://www.stata.com/support/statalist/faq
> * http://www.ats.ucla.edu/stat/stata/
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/