1: \begin{abstract}

2: A common assumption in pair-copula constructions is that the copula of the conditional distribution of two random variables given a covariate does not depend on the value of that covariate. Two conflicting intuitions arise about the best possible rate of convergence attainable by nonparametric estimators of that copula. On the one hand, the best possible rates for estimating the marginal conditional distribution functions is slower than the parametric one. On the other hand, the invariance of the conditional copula given the value of the covariate suggests the possibility of parametric convergence rates. The more optimistic intuition is shown to be correct, confirming a conjecture supported by extensive Monte Carlo simulations by I. Hobaek Haff and J. Segers [Computational Statistics and Data Analysis 84:1--13, 2015] and improving upon the nonparametric rate obtained theoretically by I. Gijbels, M. Omelka and N. Veraverbeke [Scandinavian Journal of Statistics 42:1109--1126, 2015]. The novelty of the proposed approach lies in a double smoothing procedure for the estimator of the marginal conditional distribution functions. 

3: % Under mild conditions on the bandwidth sequences, this estimator takes values in a class of smooth functions that has sufficiently small entropy numbers for empirical process arguments to work. 

4: The copula estimator itself is asymptotically equivalent to an oracle empirical copula, as if the marginal conditional distribution functions were known.

5: 

6: \noindent \textbf{Keywords:} Donsker class; Empirical copula process; Local linear estimator; Pair-copula construction; Partial copula; Smoothing; Weak convergence.

7: \end{abstract}

8: