Hey, I need some HELP please!?!

[Unikkatel]

Hey guys,

I know this is not the place to ask for stuff like this but i am stuck... if anyone has some inputs it would be greatly appreciated. I am trying to log the growth rate (average annual percentage growth of real gdp) against the log of trade share on STATA. However, a lot of the data on growth is negative percentages so logging it makes no sense. How do you deal with such a thing? Do you ommitt the data? Is there some sort of way i can log negative percentage on STATA?

Thanks and again I know this is not the place but it would be greatly appreaciated.

[Ludwig]

Off the top of my head, maybe you should try a linear transform on the data that makes all the values positive? (then just convert back when you need to). +100% to everything makes sense, since then you're converting from percentage growth to "GDP as % of last year."

I don't like growth percentages in general since in some sense they're incomparable from year to year in the same country, and further incomparable between countries with different GDPs. But I don't know what the standard in the literature is, so that might be an ignorant statement.

[tm_guru]

Uch, I hate these type of posts!!!!!!!!!!!

[asquare]

As is often the case, this isn't a Stata problem; it's a conceptual problem. Taking logarithms is not a logical transformation for your data. There's no trick to doing this in Stata because the log of a negative number is undefined. I'd suggest thinking about why you want to take logs anyways. Often, we use logs because we want to think about percentage changes, but your data are already in percentages. Think about the units in which you want your coefficients to be expressed, and then determine the right units for Y and X.

In other words, engage brain before operating Stata!