Value weighted returns are the same as equal weighted returns

[Sander Soet]

Dear everyone,
Im sorry to post something similar as my post yesterday, however it is really important that i can figure this out. I want to calculate weighted returns on the basis of the market value of a firm at week=0 (mv2). I duplicated this value for every firm(companyid) over their whole timeperiod up to week 156.
As a next step i calculated the returns with the _gwtmean package. After this i used the following code in order to determine the valueweighted returns.
Code: bys companyid(week): egen wmean_ret = wtmean(returnPE), weight(mv2)

However, this gave the same mean return as without using the weighted mean code.

bys Companyid(week): egen mean_ret = mean(returnco) . So in the end there doesnt seem to be any weighting applied to the returns. Any ideas on how i could get different returns based on firm prominence by market value at time=0?

I'd really appreciate your help!

My data looks as follows:

* Example generated by -dataex-. To install: ssc install dataex
clear
input int Date double PriceCO byte Companyid int week float(returnco mv2 wmreturnPE mean_ret)
18611   9.9 1  0           . 290.7 .0041245334 .0041245334
18618  9.89 1  1 -.001010101 290.7 .0041245334 .0041245334
18625    10 1  2  .011122346 290.7 .0041245334 .0041245334
18632 10.34 1  3        .034 290.7 .0041245334 .0041245334
18639 10.21 1  4 -.012572533 290.7 .0041245334 .0041245334
18646 10.57 1  5   .03525955 290.7 .0041245334 .0041245334
18653 11.27 1  6   .06622516 290.7 .0041245334 .0041245334
18660 12.35 1  7   .09582964 290.7 .0041245334 .0041245334
18667    12 1  8  -.02834008 290.7 .0041245334 .0041245334
18674 12.51 1  9       .0425 290.7 .0041245334 .0041245334
18681 12.24 1 10 -.021582734 290.7 .0041245334 .0041245334
18688 12.59 1 11   .02859477 290.7 .0041245334 .0041245334
18695 12.49 1 12 -.007942812 290.7 .0041245334 .0041245334
18702 11.86 1 13  -.05044035 290.7 .0041245334 .0041245334
18709 12.55 1 14   .05817875 290.7 .0041245334 .0041245334
18716 12.05 1 15  -.03984064 290.7 .0041245334 .0041245334
18723 11.71 1 16  -.02821577 290.7 .0041245334 .0041245334
18730 11.61 1 17 -.008539709 290.7 .0041245334 .0041245334
18737 11.33 1 18  -.02411714 290.7 .0041245334 .0041245334
18744 11.88 1 19   .04854369 290.7 .0041245334 .0041245334
18751 11.41 1 20  -.03956229 290.7 .0041245334 .0041245334
18758 11.82 1 21   .03593339 290.7 .0041245334 .0041245334
18765 12.92 1 22    .0930626 290.7 .0041245334 .0041245334
18772 13.01 1 23  .006965944 290.7 .0041245334 .0041245334
18779 13.07 1 24  .004611837 290.7 .0041245334 .0041245334
18786 13.03 1 25 -.003060444 290.7 .0041245334 .0041245334
18793  12.1 1 26  -.07137375 290.7 .0041245334 .0041245334
18800 13.14 1 27   .08595041 290.7 .0041245334 .0041245334
18807 13.35 1 28  .015981736 290.7 .0041245334 .0041245334
18814 13.07 1 29 -.020973783 290.7 .0041245334 .0041245334
end
format %tdnn/dd/CCYY Date

[William Lisowski]

In both cases, you are calculating a separate mean for each company. But your weight is constant for all the observations in a company, so the weighted mean will not be different than the unweighted mean. Indeed, the output of help _gwtmean tells us

If the weight expression is a non-zero constant, then, too, the results are identical to egen mean.