# What's the use of '% to GDP' type of variables?

In my study I will look for the relationship between the Gini coefficient and trade, FDI and other variables. However, as I was regressing it... the result turn out to be insignificant. My data that I used then were 'trade as a percentage to GDP' and 'FDI as a percentage to GDP'.

On the other hand, when I regress it with just the total of trade and FDI... it brings a somewhat significant result.

Can I drop the 'as a perentage of GDP'? And just use the total trade and FDI without that?

Plus I only have 16 years worth of data, though.

Imagine that I am trying to determine whether eating corn has any effect on your height. I see that in the US, total corn consumption is 20 million tons per year (made up number, all others will be made up as well) and adding up the heights of the 300 million citizens we get 500 million meters. Similar statistics for France are 4 million tons per year and 100 million meters.

Total American corn consumption was higher, as well as total American height. Is this because corn has an effect on height? Or is it that both statistics are likely to be proportional to the size of the country (measured here by population)?

Depending on what your exact hypothesis is, you may want to filter out size effects by measuring these statistics of the national accounts as % of GDP.

Complementing @Giskard's answer, when you are using "trade as a perentage of gdp" or "FDI as a percentage of GPD", you are measuring the extent to which an economy is open to trade, or to inflows and outflows of international investment. Thus, "trade as a perentage of gdp" measures a country's trade openness, while increasing trade per se does not necessarily increase trade openness (for example a country may produce more and export more). Thus, the use of "trade" and "trade as a percentage of GDP" measures different economic phenomena.

It is a Ratio variable. And is well known in the literature that the correct interpretation of a correlation between ratio variables is debatable.

See User's guide to ratio variables and A note on using ratio variables in regression analysis for more details.

• Can you give a summary of the arguments against ratio variables? It would make your answer more useful as a stand alone resource. – BKay Mar 6 '20 at 18:33

You can drop the 'as a perentage of GDP'? Just use the total trade and FDI for evaluating their impact on your dependent variable ? You need not worry about whether the effect is significant (or insignificant). trade as a percentage to GDP' and 'FDI as a percentage to GDP' are structured variables and reflect new /other variables. Given the objective of relating trade (and FDI) with Ginni coefficients (indicating inequality), use of new (structured) variables can not be justified.

• This answer could be improved by elaborating on why you think you can safely do this. – BKay Mar 6 '20 at 18:32