Strict concordance monotonicity of Kendall's tau under full support #
For bivariate copulas C ≤ D (pointwise, LowerOrthantLE) the difference of Kendall's taus
splits as (LowerOrthantLE.kendallTau_sub_eq)
τ(D) - τ(C) = 4 (∫ (D - C) dD + ∫ (D - C) dC),
using the symmetry ∫ C dD = ∫ D dC of the concordance integral (integral_cdf_swap; Nelsen
2006, Thm. 5.1.1/Cor. 5.1.2). Both integrands are nonnegative and continuous, so if one of
the two copulas has full support (its measure charges every nonempty open set) and
C ≠ D, then τ(C) < τ(D) (LowerOrthantLE.kendallTau_lt_of_isOpenPosMeasure_right,
..._left). No claim is made for arbitrary comparable copulas.
The difference of Kendall's taus as two concordance integrals of D - C.
Strict monotonicity of Kendall's tau when the larger copula has full support.
Strict monotonicity of Kendall's tau when the smaller copula has full support.
Under full support of the larger copula, equal Kendall's taus force equality.
Under full support of the smaller copula, equal Kendall's taus force equality.