Inference for Seemingly Unrelated Varying-Coefficient Nonparametric Regression Models

Area of Science:

• Statistics
• Econometrics
• Biostatistics

Background:

• Seemingly unrelated (SU) regression models are common in econometrics and biostatistics.
• Varying-coefficient nonparametric regression offers flexibility in modeling complex relationships.
• Existing methods for SU models often lack efficiency or flexibility.

Purpose of the Study:

• To develop an efficient estimation method for unknown coefficient functions in SU varying-coefficient nonparametric regression models.
• To extend the generalized likelihood ratio test for model goodness-of-fit to the SU regression setting.
• To assess the performance of the proposed methods via simulations and a real-world environmental epidemiology study.

Main Methods:

• An extension of the two-stage estimation procedure for nonparametric regression.
• Application of a generalized likelihood ratio technique for model testing.
• Wild block bootstrap method for computing p-values.

Main Results:

• The proposed estimators are asymptotically normal and more efficient than those based on individual equations.
• The extended generalized likelihood ratio test is effective for SU regression.
• Simulation studies support the theoretical asymptotic results.

Conclusions:

• The proposed estimation and testing procedures are effective for SU varying-coefficient nonparametric regression.
• The methods offer improved efficiency and flexibility compared to existing approaches.
• The techniques are applicable to real-world problems, such as environmental epidemiology.