EUVIMEDEUVIMED entdecken
Crossref

BEYOND INSTRUMENTAL VARIABLES: A SAME-QUANTILE TWO-STAGE ESTIMATOR FOR SECOND-ORDER DYNAMIC PANELS WITH EXTREME-TAIL CONTAMINATION WHEN \(N=T\)

Hossameldin Ahmed

Advances and Applications in Statistics · 2026 · 93 (11) · 1393-1419

DOI: 10.17654/0972361726062 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Advances and Applications in Statistics, 2026. Verfasst von Hossameldin Ahmed.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

This paper develops the extended two-stage quantile regression for dynamic panel data (ETSQRDPD) estimator, a new framework for estimating conditional quantile functions in dynamic panels. The structural model incorporates two lags of the dependent variable to capture richer dynamic effects than conventional first-order specifications, alongside a vector of strictly exogenous covariates. Because the lagged dependent variables are correlated with the unobserved individual effect, each is first filtered of this dependence through an auxiliary first-order autoregressive quantile projection, estimated at a quantile level \(\tau\). The resulting fitted values are then substituted into the structural quantile regression equation and estimated via linear programming at the same quantile level, so that the endogeneity correction and the conditional quantile function remain aligned across both stages. We establish the asymptotic normality of the ETSQRDPD estimator as a formal theorem, with proof, under the joint asymptotic regime \(N=T \rightarrow \infty\), and characterize its variance-covariance structure under a set of regularity conditions. An extensive Monte Carlo study spanning seven error distributions (normal, Student- \(t\), centered log-normal, Cauchy, Pareto, Fréchet and Weibull), four balanced-panel dimensions and three quantile levels compares ETSQRDPD against pooled (PQRDPD) and instrumental-variable (IVQRDPD) quantile regression for dynamic panels. The results show that PQRDPD performs best in very small panels, while ETSQRDPD becomes the preferred estimator as the panel dimension grows, offering far greater stability than IVQRDPD across the seven error distributions, particularly under heavy-tailed contamination, where IVQRDPD can miss by several orders of magnitude.

Zitieren

APA

Ahmed, H. (2026). BEYOND INSTRUMENTAL VARIABLES: A SAME-QUANTILE TWO-STAGE ESTIMATOR FOR SECOND-ORDER DYNAMIC PANELS WITH EXTREME-TAIL CONTAMINATION WHEN \(N=T\). Advances and Applications in Statistics, 93(11), 1393-1419. https://doi.org/10.17654/0972361726062

Vancouver

Ahmed H. BEYOND INSTRUMENTAL VARIABLES: A SAME-QUANTILE TWO-STAGE ESTIMATOR FOR SECOND-ORDER DYNAMIC PANELS WITH EXTREME-TAIL CONTAMINATION WHEN \(N=T\). Advances and Applications in Statistics. 2026;93(11):1393-1419. doi:10.17654/0972361726062

BibTeX
@article{Ahmed2026,
  author = {Ahmed, Hossameldin},
  title = {BEYOND INSTRUMENTAL VARIABLES: A SAME-QUANTILE TWO-STAGE ESTIMATOR FOR SECOND-ORDER DYNAMIC PANELS WITH EXTREME-TAIL CONTAMINATION WHEN \(N=T\)},
  journal = {Advances and Applications in Statistics},
  year = {2026},
  volume = {93},
  number = {11},
  pages = {1393--1419},
  doi = {10.17654/0972361726062}
}
Kostenlos starten →