QQ regression is perhaps one of the latest methods in econometric estimation approaches. In case you have expertise could you please help me by providing useful information as to how to perform QQ regression using R or Stata?
As far as I know, Quantile regression is a linear function, whereas quantile-on-quantile regression is a nonlinear function. Another, major difference is that quantile regression is a regression of the quantile of dependent variable on the independent variables, whereas, QQ regression is a regression of a quantile of dependent variable on the quantile of independent variables.
I attach herewith a research paper based on QQ regression.
You stated that the R package quantreg does a great job with quantile regression. Could you please confirm that it does Quantile-on-Quantile Regression?
@Melain Modeste Senou, I think that's for quantile regression in stata.although, its still the same function that's used for quantile on quantile in R too, but it requires some little tweaking...
Quantile-on-Quantile regression for two variables a and b is carried out in two steps:
Step 1: First, estimate the slope coefficients for q(a) x q(b) quantiles in R, as all the quantile combinations will be considered as separate datasets. Once estimated, store them in a .csv file.
Step 2: In Matlab, assign all those slope values to a single variable, and plot is against q(a) and q(b) quantiles to generate a 3D image of the estimation results
So, basically Sim and Zhou (2015) QQR test is conducted using both R and Matlab. By far, codes of this test is neither available in Stata, nor in Eviews. The codes available in both these statistical packages are of Quantile Regression (QR), not Quantile-on-Quantile regression (QQR).