The estimation of conditional probability distribution functions (PDFs) in a kernel nonparametric framework has recently received attention. As emphasized by Hall, Racine and Li (2004), these conditional PDFs are extremely useful for a range of tasks including modelling and predicting consumer...