Theorem 1.2: the exact ξ–β region of stochastically increasing copulas #
R^{SI}_{ξ,β} = R^{SI,RS}_{ξ,β} = {(x,y) ∈ [0,1]² : y³ ≤ 2x, 3(1-y)² ≤ 4(1-x)}, and
R^{SD} = R^{SD,RS} is its mirror image. Every SI copula satisfies
ξ(C) ≤ 1 - (3/4)(1-β(C))², with equality iff C = R_{β(C)}.
The region of Theorem 1.2: {(x,y) ∈ [0,1]² : y³ ≤ 2x, 3(1-y)² ≤ 4(1-x)}.
Equations
Instances For
L_b ∈ C_SI ∩ C_RS and R_b ∈ C_SI ∩ C_RS, so both attain the two boundary curves.
Theorem 1.2: R^{SI}_{ξ,β} = R^{SI,RS}_{ξ,β} = {(x,y) ∈ [0,1]² : y³ ≤ 2x, 3(1-y)² ≤ 4(1-x)}.
Theorem 1.2: R^{SD} = R^{SD,RS} is the mirror image {(x, -y) : (x,y) ∈ R^{SI}}.
Theorem 1.2, inequality (1.5): every SI copula satisfies
ξ(C) ≤ 1 - (3/4)(1 - β(C))², with equality if and only if C = R_{β(C)}.
Inequality (1.5) in the equivalent form 3(1 - β)² ≤ 4(1 - ξ).
Inequality (1.5) as a lower bound on β: β(C) ≥ 1 - 2√((1 - ξ(C))/3).