1: \begin{abstract}

2: We revisit the question of characterizing the convergence rate of plug-in estimators of optimal transport costs.

3: It is well known that an empirical measure comprising independent samples from an absolutely continuous distribution on $\bbR^d$ converges to that distribution at the rate $n^{-1/d}$ in Wasserstein distance, which can be used to prove that plug-in estimators of many optimal transport costs converge at this same rate.

4: However, we show that when the cost is smooth, this analysis is loose: plug-in estimators based on empirical measures converge quadratically faster, at the rate $n^{-2/d}$.

5: As a corollary, we show that the Wasserstein distance between two distributions 

6: is significantly easier to estimate when the measures are far apart.

7: We also prove lower bounds, showing not only that our analysis of the plug-in estimator is tight, but also that no other estimator can enjoy significantly faster rates of convergence uniformly over all pairs of measures.

8: Our proofs rely on empirical process theory arguments based on tight 

9: control of $L^2$ covering numbers for locally Lipschitz and semi-concave functions.

10: As a byproduct of our proofs, we derive $L^\infty$ estimates on the displacement induced

11: by the optimal coupling between any two measures satisfying suitable moment conditions,

12: for a wide range of cost functions.

13: \end{abstract}

14: