Robust Specification Test for Semiparametric Models

This paper introduces a heteroskedasticity-robust Lagrange Multiplier–type specification test for semiparametric regression models. The proposed test is computed from estimates of a restricted semiparametric model and can be implemented using regression techniques. Under the null hypothesis, the test statistic is asymptotically standard normal, and it is designed to detect a wide class of departures from the null. Applied to Environmental Engel Curves, the test finds that the income–emissions relationship may be quadratic but that its form varies with other model variables. Simulation studies reported in the paper indicate the test controls size well, has good power, and is fairly robust to tuning choices.

What the study examined

This work develops a heteroskedasticity-robust Lagrange Multiplier–type specification test for semiparametric regression models. The focus is on a test that uses estimates from a restricted semiparametric specification and that can be computed in a regression-based way.

The goal is to provide a procedure that can detect a wide class of deviations from the null hypothesis while remaining valid under heteroskedastic errors. The paper also examines how the procedure performs in practice through an empirical application and simulation experiments.

Key findings

The proposed test statistic is asymptotically standard normal under the null hypothesis, allowing for straightforward interpretation of results in large samples.

• The test is implemented using the restricted semiparametric estimates and can be computed with standard regression tools.
• When applied to Environmental Engel Curves, the results suggest that the link between income and emissions may take a quadratic form, but its shape changes with other variables included in the model.
• Monte Carlo studies reported in the paper show that the procedure controls size well, displays good power against alternatives, and is reasonably robust to the choice of tuning parameters.

Why it matters

Semiparametric models offer flexibility by combining parametric and nonparametric elements, but that flexibility also raises the need for reliable specification tests. A heteroskedasticity-robust, regression-computable test provides a practical tool to check whether a chosen semiparametric form is compatible with the data.

The empirical application to Environmental Engel Curves illustrates how the test can uncover nuanced patterns, such as conditional changes in functional shape, that might be missed by simpler checks. The simulation evidence provides support that the method performs well in controlled settings, making it a candidate for applied work where heteroskedasticity is a concern and semiparametric specifications are considered.