Causal effect of the degree programs on the work path of the graduates in the multivariate latent Markov model

We develop a model for the evaluation of the casual effect of the degree on the work path, when this path is seen as the manifestation of the human capital (HC) represented by a sequence of latent variables. The path is observed in terms of the three response variables: type of contract, skill level of the job, and earning level. The sequence of HC latent variables gives rise to a latent process that is assumed to follow an hidden Markov chain. In oder to asses an effect and following the recent developments in the causal inference literature, we propose to integrate the inverse probability of treatment weighting in this framework. Background covariates are used to asses balances among the groups of students with different degrees. The resulting multivariate latent Markov model is fitted by a maximum likelihood procedure by using the EM algorithm. Standard errors for the parameter estimates are obtained by a nonparametric bootstrap method. The proposed approach is applied to the analysis of data deriving from administrative archives concerning the labour market. Through this application, we also show the advantages of the proposed approach to study the HC development with respect to other approaches proposed in the literature.