The aim of this work is to study – via a Monte Carlo experiment – the small sample behaviour of Full Information Least Orthogonal Distance Estimator and to compare it with other two full information methods, namely Three Stage Least Square and Full Information Maximum Likelihood. The comparison is made under two distributional assumption about error components: Normal and Uniform distributions. FI LODE appear to have for small samples, a smaller bias than FIML and 3SLS even when error component is uniformly distributed.
Naccarato, A., Zurlo, D. (2008). A Monte Carlo Study on Full Information Methods in Simultaneous Equation Models. QUADERNI DI STATISTICA, 10, 115-144.
A Monte Carlo Study on Full Information Methods in Simultaneous Equation Models
NACCARATO, ALESSIA;
2008-01-01
Abstract
The aim of this work is to study – via a Monte Carlo experiment – the small sample behaviour of Full Information Least Orthogonal Distance Estimator and to compare it with other two full information methods, namely Three Stage Least Square and Full Information Maximum Likelihood. The comparison is made under two distributional assumption about error components: Normal and Uniform distributions. FI LODE appear to have for small samples, a smaller bias than FIML and 3SLS even when error component is uniformly distributed.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
