inequality estimation

[Osman Suliman]

What's an appropriate empirical technique for income inequality estimation( some missing data) using panel data. How can I do that using Stata?

[Carlo Lazzaro]

Osman:
welcome to the list.
Such a broad question hardly gets helpful replies.
Please, see the FAQ on how to post (more) effectively. Thanks.

[Osman Suliman]

Thanks Carlo. You can tell I am new. Here is another question:
I have data on China's national wages. I also have data on level of education (skill). Urban population is far more educated. Can I use the interaction variable of education and
wages as a proxy for urban wages? Thanks.

[Carlo Lazzaro]

Osman:
unfortunately , the previous advice still holds.
Newbies are welcomed (as everyone on this list was a newbye at a given point in time) but, in their own interest first, should learn the rules of the game.
That said, your question sounds ill-posed.
With a bit of guess-work, I would say that you have in mind a (linear) regression and your two predictors are urban wages and the level of education (which is something different from skills, by the way).
The first question would concern your dependent variable (that you do not report).
Please, post and example/excerpt of your dataset via -dataex- (type -search dataex- from within Stata to install it) and, again, please read the FAQ. Thanks.

[Osman Suliman]

Carlo: Thanks. I am a just a beginner. Thanks for your advice.
Now: How do I estimate a production function with exponents like a Le^Bt to get estimates for alpha, beta. Which nonlinear estimation? could not find that in FAQs.

[Carlo Lazzaro]

Osman:
your question is still too vague (by the way, FAQ are not about answers, but about the way questions should be posted).
Anyway, I would take a look at -help nl-.

[Osman Suliman]

Carlo:


I used nl to estimate :nl(ln_realgdp={b0}+(1-{b3})*ln(k-{b1}*education*exp({b2}*(year-1)))+{b3}*unskilled)

I got this:

Source SS df MS
Number of obs = 775
Model 586.89613 2 293.448064 R-squared = 0.4561
Residual 699.84786 772 .906538675 Adj R-squared = 0.4547
Root MSE = .9521232
Total 1286.744 774 1.66245993 Res. dev. = 2120.305


ln_realgdp Coef. Std. Err. t P>t [95% Conf. Interval]

/b0 -.8791832 .7521519 -1.17 0.243 -2.355689 .5973222
/b3 .2436297 .0629964 3.87 0.000 .1199652 .3672942
/b1 -1014971 . . . . .
/b2 -.1004462 .0144396 -6.96 0.000 -.1287918 -.0721006

Parameter b1 taken as constant term in model & ANOVA table

Question: why did use b1 and not b0 as constant term.

Best regards
Osman

[Osman Suliman]

Also Carlo: where is the statistics ( std error etc...) for b1? Thanks
Osman

[Carlo Lazzaro]

Osman:
see: https://www.stata.com/statalist/arch.../msg00308.html

[Osman Suliman]

Thanks Carlo. U really appreciate your mind help. Some of the R squares in my panel data using nl are low ( .17, .23 ). Is this ok? If so, what 's a good explanation for why value of R square in a panel data is not important?
Best
Osman

[Carlo Lazzaro]

Osman:
there's non hard and fast rule about Rsq (by the way, within or between Rsq?) magnitude.
Moste depends on the subject under investigation: the literature in your research field can help you out in this respect.

[Osman Suliman]

Hi Carlo
Can the Real interest rate be used to estimate the real return to Physical capital? Thanks.
Osman

[Carlo Lazzaro]

Osman:
unfortunately, I cannot say.

[Sergiy Radyakin]

Osman,

the following book (available in full text) contains a detailed explanation of various methods, formulas, and tools available for poverty and inequality analysis.
https://openknowledge.worldbank.org/handle/10986/11985

also see here for the ADePT tool (poverty and inequality module):

https://openknowledge.worldbank.org/handle/10986/13731

Best, Sergiy Radyakin

[Osman Suliman]

Thanks, Sergiy.