A bayesian approach to model-based clustering for panel probit models

Arbeitspapier

Consideration of latent heterogeneity is of special importance in non linear models for gauging correctly the effect of explaining variables on the dependent variable. This paper adopts the stratified model-based clustering approach for modeling latent heterogeneity for panel probit models. Within a Bayesian framework an estimation algorithm dealing with the inherent label switching problem is provided. Determination of the number of clusters is based on the marginal likelihood and out-of-sample criteria. The ability to decide on the correct number of clusters is assessed within a simulation study indicating high accuracy for both approaches. Different concepts of marginal effects incorporating latent heterogeneity at different degrees arise within the considered model setup and are directly at hand within Bayesian estimation via MCMC methodology. An empirical illustration of the developed methodology indicates that consideration of latent heterogeneity via latent clusters provides the preferred model specification compared to a pooled and a random coefficient specification.

Object Details

• Language: Englisch

References and Relationships

• Bibliographic citation: Series: Economics Working Paper ; No. 2009-03

Classification and Topics

• Classification: Wirtschaft ; Bayesian Analysis: General ; Single Equation Models; Single Variables: Panel Data Models; Spatio-temporal Models ; Single Equation Models; Single Variables: Discrete Regression and Qualitative Choice Models; Discrete Regressors; Proportions; Probabilities
• Subject: Bayesian Estimation ; MCMC Methods ; Panel Probit Model ; Mixture Modelling ; Probit-Modell ; Panel ; Bayes-Statistik ; Schätztheorie ; Theorie

Contributors, Places and Time

• Event: Geistige Schöpfung
• (who): Aßmann, Christian ; Boysen-Hogrefe, Jens
• Event: Veröffentlichung
• (who): Kiel University, Department of Economics
• (where): Kiel
• (when): 2009

Further information

• Handle: http://hdl.handle.net/10419/27738
• Last update: 10.03.2025, 11:45 AM CET