findrestrict: Ad-hoc method for (weakly) identifying the factor loadings...

In factorstochvol: Bayesian Estimation of (Sparse) Latent Factor Stochastic Volatility Models

Ad-hoc method for (weakly) identifying the factor loadings matrix

Description

In factor SV models, the identification of the factor loadings matrix is often chosen through a preliminary static factor analysis. After a maximum likelihood factor model is fit to the data, variables are ordered as follows: The variable with the lowest loadings on all factors except the first (relative to it) is determined to lead the first factor, the variable with the lowest loadings on all factors except the first two (relative to these) is determined to lead the second factor, etc.

Usage

findrestrict(dat, factors, transload = abs, relto = "all")

Arguments

dat	Matrix containing the data, with n rows (points in time) and m columns (component series).
factors	Number of factors to be used.
transload	Function for transforming the estimated factor loadings before ordering. Defaults to the absolute value function.
relto	Can be 'none', 'current' or 'all'. If 'none', the series with the highest loadings is placed first, the series with the second highest is placed second, and so on. If 'current', the current factor loading is used as a reference, if 'all', all previous loadings are summed up to be the reference.

Value

A m times factors matrix indicating the restrictions.

Note

This function is automatically invoked by fsvsample if restrict is set to 'auto'.

See Also

ledermann

factorstochvol documentation built on Nov. 24, 2023, 5:08 p.m.