Dear all:

I am writing to see if anyone knows a Stata command that helps set the optimal bandwidth for bivariate kernel estimation.

I looked into a few commands - they do not seem to do it.
  • In tddens, bandwidth is 0.2 as the default
  • In bidensity, the "optimal" bandwidth is separately chosen for the two variables - it is optimal in the one-dimensional case.
I wonder if I am asking something that is too demanding for Stata. Thanks!

Best,
Long