Interpretation of a ratio variable in regression output.

[Sita Gupta]

Hello,

I have a ratio variable x1/x2 and I want to interpret its beta, so I assume I can just say if my ratio increases by 0,01 my dependent variable increases by 0,01*beta.


But know I have a quadratic term of my ratio, so I compute the margin, and the margin differs highly with the level of the ratio. Can I still interpret it the way I did before? e.g. if my ratio increases by 0,01 at its mean then the dependent variable increases by 0,01*marginal effect?


Thank you!

[Phil Bromiley]

You'll increase your chances of a useful answer by following the FAQ on asking questions - provide Stata code in code delimiters, readable Stata output, and sample data using dataex.

First, have you used factor variable notation? If you use factor variable notation to do your squares, then you can just use margins. You will get a difference dydx at different values of the ratio, but that is what you should get with a squared term. Think about it - with a squared term, it is quite likely you'll have positive slopes in some regions and negative in others. Often, it is helpful to use predictive margins (instead of dydx) to see what is going on.