Researchers have developed new robust inference methods for linear panel data models that incorporate latent group structures. These methods are designed to remain valid even when group separation fails, improving upon conventional asymptotic procedures. The approach uses selective conditional inference based on the conditional distribution of coefficient estimates given the estimated group structure, offering exact validity under Gaussian errors and asymptotic validity under general error distributions. The framework accommodates arbitrary linear restrictions on group-specific coefficients and provides selective confidence sets with valid coverage conditional on the estimated group structure, as demonstrated through simulations and an application to growth convergence clubs. AI
IMPACT Provides advanced statistical tools for analyzing complex datasets, potentially applicable in AI research involving structured data.
RANK_REASON Academic paper published on arXiv. [lever_c_demoted from research: ic=1 ai=0.4]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
