Full-horizon asymptotics beyond the marked-memoryless benchmark

Develop full-horizon asymptotic theory for the least-false Cox estimator with affine terminal-time contamination outside the marked-memoryless competing-exponential benchmark, including the required triangular-array analysis as the analysis horizon diverges.

Background

The paper establishes finite-horizon results and gives a conditional full-horizon central limit theorem for the competing-exponential benchmark under additional tail conditions. It states that extending full-horizon theory to non-memoryless settings requires a triangular-array argument, which is not developed.