Variance decomposition: between, within-between, within

[mathilde le moigne]

Hi all,

I am relatively noob at Stata and I am trying to construct variance decomposition of wages. I understand that if I use the command areg (or xtreg) y, abs(groupvar) I can obtain the within and the between components of my variance.
I wish to push my variance decomposition a little further: I want to decompose the variance of wage according to three components: across firms, within firms across occupations and within occupations.
Assuming I have enough observations at each level, is there a clever way to do this?

Thank you so much for your help.

Mathilde

[Phil Bromiley]

You didn't get a quick answer. You'll increase your chances of a useful answer by following the FAQ on asking questions - provide Stata code in code delimiters, readable Stata output, and sample data using dataex.

You might look at refhdfe.

[mathilde le moigne]

Thanks! Sorry so here is the code I have so far:

gen lwage = log(wage)

** Fixed-effect regressions
* Over the whole sample
    egen lw_var = sd(lwage)
    replace lw_var = lw_var^2

* Within/Between firms
    reghdfe lwage, abs(firmid, savefe)
    predict fwithin if e(sample), res
    predict fbetween if e(sample), xbd
    egen temp=sd(fbetween)
    gen sfbetween = temp*temp
    drop temp
    egen temp=sd(fwithin)
    gen sfwithin = temp*temp
    drop temp
    gen sftotal = sfwithin + sfbetween

sum lw_var sftotal sfwithin sfbetween

* Within/Between firm-occupation, interacted (i.e. underlying assumpyion is that the occupation fixed-effect varies across firms)
    reghdfe lwage, abs(firmid#occ, savefe)
    predict fowithin if e(sample), res
    predict fobetween if e(sample), xbd
    egen temp=sd(fobetween)
    gen sfobetween = temp*temp
    drop temp
    egen temp=sd(fowithin)
    gen sfowithin = temp*temp
    drop temp
    gen sfototal = sfowithin + sfobetween

sum sfototal sfowithin sfobetween

I am using StataSE 14.

My issue is that I don't know how to reconcile the two within/between decompositions. I obtain the same total variance in both case, which is normal, but what I would really like is use the variance obtain in sfwithin, and from there decompose in a within/between component across occupation.
Could I just do something like:

* Within/between occupations in a given firm
   reghdfe sfwithin, absorb(occ, savefe)
   predict fowithin if e(sample), res
   predict fobetween if e(sample), xbd

or would that not make any sense?

Thank you very much.