MONTE CARLO EVALUATION OF SEMIPARAMETRIC COPULA-BASED ESTIMATORS UNDER JOINT CENSORING, TRUNCATION, AND NONLINEAR DEPENDENCE STRUCTURES
Adamu Jibrilla and Danjuma Ahmad
Volume 5 Issue 2
Abstract
This study conducts a Monte Carlo evaluation of semiparametric copula-based estimators under conditions of joint censoring, truncation, and nonlinear dependence structures, addressing key limitations of classical parametric methods such as the Tobit and Heckman models. By simulating a range of complex data-generating processes, the study examines finite-sample performance in terms of bias, root mean squared error (RMSE), coverage probability, and robustness to misspecification of marginal distributions and dependence structures. The results demonstrate that semiparametric copula estimators consistently outperform traditional approaches, exhibiting lower bias, more accurate inference, and greater resilience to high levels of censoring and truncation. Findings underscore the importance of flexible dependence modeling and suggest that semiparametric copula methods provide a reliable framework for empirical research in complex data environments. These insights have significant implications for applied econometrics, particularly in labor economics, health studies, and development research, where limited dependent variables and nonlinear relationships are prevalent. It is recommended that researchers should evaluate both semiparametric and classical estimators, taking into account sample size, censoring, truncation, and the underlying dependence structure, to ensure reliable and robust inference. Keywords: Semiparametric Estimation, Copula Models, Censoring and Truncation, Monte Carlo Simulation, Nonlinear Dependence
Download Paper