The weak-instruments problem in factor models

Autor:	Kazuhiko Hayakawa
Rok vydání:	2019
Předmět:	Clinical Psychology Variable (computer science) Applied Mathematics Instrumental variable Monte Carlo method Astrophysics::Instrumentation and Methods for Astrophysics Econometrics Experimental and Cognitive Psychology Variance (accounting) Scaling Analysis Mathematics Factor analysis
Zdroj:	Behaviormetrika. 47:123-157
ISSN:	1349-6964 0385-7417
DOI:	10.1007/s41237-019-00097-1
Popis:	This study considers the instrumental variable estimation of factor models. Specifically, we investigate the weak-instruments problem, which is not well investigated in the literature, in detail. We show that the signal-to-noise ratios, which are defined by the variance ratios of the common components to the errors, of the scaling variable and the instruments mainly determine the strength of instruments in a confirmatory factor model, while the structure of the factor loading is closely related to the strength of instruments in an explanatory factor model. By Monte Carlo simulation, we confirm that the simulation results are consistent with the theoretical implications, and Stock and Yogo test is useful in practice in detecting the weak instruments.
Databáze:	OpenAIRE
Externí odkaz:	https://explore.openaire.eu/search/publication?articleId=doi_________::6884f4e4f3479a193a6086b882c1b14f https://doi.org/10.1007/s41237-019-00097-1 Zobrazit plný text záznamu Full text from SpringerLink