Labor market signaling analysis using the probability of misclassification and neural networks

Carlos Narciso Bouza HERRERA[1]

Josefina Martínez BARBEITO[2]

Pasha G. MITRA[3]

§    ABSTRACT: In this paper the de-codification of labour market signals is studied. The structure of the corresponding signalling game is described. The decisions of the employer are considered as based on a classification of the aspirant. The evaluation of the posterior probability of classifying him correctly provides a decision rule. A naïve estimator based on density function estimator is proposed. Neural Networks models are proposed for coping with the classification using two approaches. Monte Carlo experiments are used for evaluating the behaviour of the proposals.

§    KEYWORDS: Signalling games; posterior probabilities; feedforward neural networks; classification tree.

 



[1] Facultad de Matemática y Computación, Universidad de La Habana, San Lázaro y L. CP 10400, Habana, Cuba, E-mail: bouza@matcom.uh.cu

[2] Facultad de Ciencias Económicas y Empresariales, Universidade A Coruña, CP 15001, A Coruña, Galicia, Espanha.

[3] Departament of Computer Sciences, College of Management Sciences and Business Administration, Universidade A Coruña, CP 15001, A Coruña, Galicia, Espanha.