Explore how different GMM estimators perform in small samples and what that means for real data.
This book examines three alternative GMM estimators and compares their finite‑sample behavior using Monte Carlo experiments inspired by asset‑pricing models. It also investigates how different ways of constructing confidence regions affect statistical inferences in practice.
Inside, you’ll see how weighting choices, model specification, and data design (annual vs. monthly data, nonlinear moments, and time dynamics) influence estimator performance. The discussion stays close to practical concerns, such as how well confidence intervals and tests behave when sample sizes are small and when the weighting matrix depends on the parameter estimates.
- Clear comparisons of two‑step, iterative, and continuous‑updating GMM estimators.
- Evidence on how different moment conditions and data patterns shape bias, dispersion, and coverage probabilities.
- Insights into constructing and interpreting criterion‑function–based confidence regions versus Wald‑type intervals.
- Guidance on numerical implementation, potential pitfalls, and when each estimator may be preferable.
Ideal for readers of empirical econometrics and finance who want a grounded view of finite‑sample properties and practical inference with GMM methods.