Stumped by WLS

[diggetybo]

Hello,

I was reviewing a chapter on weighted least squares, and found a different answer than in the book. I've been trying to figure it out for hours :(

The equation is very simple: nettfa = c + inc
nettfa is net total financial wealth
inc is income

There is also a sample restriction for single individuals: fsize=1

The answer in the book for OLS is: nettfa = -10.57 + .821inc + e
The answer in the book for WLS is: nettfa = -9.58 + .787inc + e

the heteroskedasticity is suggested to follow a multiplicative form: var(u|inc)=sigma2*inc, so to weight by sigma, we weight using: sqr(inc) right?

I got the OLS output right (big woop)
But my result for WLS is: nettfa = -10.21 + .811 + e

I checked: weighted least squares in eviews and used the series sqr(inc) for the weight.

The file is attached. Let me know if someone gets the books answer, I'm not sure what the problem is.

[trubador]

Since you want to "remove" the heteroscedasticity, you should multiply the series by inverse standard deviation. Which means that the correct weight would be: 1/@sqrt(inc)

[diggetybo]

Ohhh yea, haha that makes sense. thank you