Rangestat: value weighted variance and skewness

[Olena Onishchenko]

Dear Stata users

I need to calculate equially- and value-weighted variance and skewness for each id/month.

I am aware of rangestat equally-weighted code:

rangestat (sd) market_returns (skewness) market_returns, interval ( monthly_date . 0) by(id)
gen variance_market_returns=market_returns_sd^2

Is there any way to calculate value-weighted (by market_cap) variance and skewness using rangestat command?

If not, what is the alternative way out?

I am aware of collapse command which can adjust for weights but it does not have variance or skewness option.

collapse (sd) market_returns [weight=market_cap], by (id month)

Any advice is highly appreciated.

Thank you.

[Nick Cox]

rangestat is from SSC as you are asked to explain (FAQ Advice #12). It doesn't support weights directly. You could write your own extra code to support weights, using that or rangerun (SSC).

Skewness here evidently means the moment-based measure. I wouldn't trust skewness much for subsets of about 30, if that's your data. I fear nevertheless that this is the standard method in your field. I'd flag that (mean - median) / SD works quite well often and is bounded by [-1, 1].

[Olena Onishchenko]

Nick

Thank you.

Can you advice how to write extra code to support weights using rangerun?

[Nick Cox]

I'd advise studying the help for examples. One example concerns skewness and one uses weights....