Firma Arista

I think these two estimations should give the same estimated
      coefficients, if I am not missing something
:

<hansl>

open data8-2.gdt

wls pop exptrav const income

wls pop exptrav const income --robust

</hansl>

But the results I obtain are:

? wls pop exptrav const income

Model 1: WLS, using observations 1-51

Dependent variable: exptrav

Variable used as weight: pop

             coefficient   std. error   t-ratio   p-value 

  --------------------------------------------------------

  const      -0.486074     0.870649     -0.5583   0.5792  

  income      0.0599703    0.00277146   21.64     1.00e-26
      ***

Statistics based on the weighted data:
...

? wls pop exptrav const income --robust

Model 2: WLS, using observations 1-51

Dependent variable: exptrav

Heteroskedasticity-robust standard errors, variant HC3

Variable used as weight: pop

             coefficient   std. error   t-ratio    p-value

  --------------------------------------------------------

  const       0.498120     116.357      0.004281   0.9966 

  income      0.0555731      1.48871    0.03733    0.9704 

Statistics based on the weighted data:
...