thetaSE_louis

Computes standard errors using the expected complete-data information
matrix <code>B</code> from Louis (1982). <code>B = -Hessian(Q)</code>, where <code>Q(theta)</code>
is the expected complete-data log-likelihood (the EM Q-function), evaluated at the
converged parameter values with smoothed state probabilities held fixed.
The complete-data information <code>B</code> is typically better conditioned than the
observed-data Hessian. Since <code>I_obs = B - M</code> where <code>M &gt;= 0</code> is the
missing information, <code>B &gt;= I_obs</code> (PSD), so <code>B^{-1}</code> slightly
underestimates the true variance. The SEs are therefore slightly optimistic but
numerically more stable, especially for weakly identified models.

internal

Implementation of hypothesis testing procedures described in Hansen (1992) <doi:10.1002/jae.3950070506>, Carrasco, Hu, & Ploberger (2014) <doi:10.3982/ECTA8609>, Dufour & Luger (2017) <doi:10.1080/07474938.2017.1307548>, and Rodriguez-Rondon & Dufour (2026a) <doi:10.34989/swp-2026-23> that can be used to identify the number of regimes in Markov switching models. For a detailed description of the package, see Rodriguez-Rondon & Dufour (2026b) <doi:10.34989/swp-2026-7>.

Gabriel Rodriguez-Rondon

MSTest

Hypothesis Testing for Markov Switching Models

Jean-Marie Dufour

thetaSE_louis function

<dl><dt>mdl</dt>
<dd>List with model properties (output from an MS model constructor).
Must contain <code>theta</code>, <code>St</code> (smoothed probabilities), and model-specific fields.</dd></dl>

Arguments

Louis (1982) standard errors for Markov-switching models — thetaSE_louis

<dl>

<dt>mdl</dt>
<dd>List with model properties (output from an MS model constructor).
Must contain <code>theta</code>, <code>St</code> (smoothed probabilities), and model-specific fields.</dd>

</dl>

thetaSE_louis: Louis (1982) standard errors for Markov-switching models

Description

Usage

Value

Arguments

References