Panel Data Models: Latent Class Models

The latent class model framework is of the form:

            Model | class = familiar structure

            Prob[class] = probability model

The class membership is not observed by the analyst. The estimated ‘model’ is therefore a weighted average of the underlying structures. LIMDEP’s implementation of this structure uses a multinomial logit model for the class probabilities, which allows observed covariates to influence the class probabilities. Among the models supported are:

• Linear regression model
• Probit, logit, Gompertz, complementary log log binary choice
• Tobit, truncated regression, categorical data
• Stochastic frontier
• Survival models: exponential, Weibull, lognormal, loglogistic
• Loglinear models: Weibull, gamma, exponential, inverse Gauss
• Bivariate probit, partial observability
• Ordered probit, ordered logit, ordered Gompertz, ordered complementary log log
• Sample selection
• Poisson, negative binomial, zero inflated Poisson
• Conditional logit (multinomial logit - discrete choice)

Results produced include the class specific parameter vectors for the model (you specify the number of classes), and the parameters of the probability model. The estimator is maximum likelihood. (The literature contains a few applications of this model to the linear regression, Poisson and binomial logit and probit models. The preceding extend the model into several new areas.) After estimation, you can compute an estimate of the class from which an observation is drawn (the one with the largest probability), and posterior estimates of the set of class probabilities. A suitably weighted average produces an individual (group) specific parameter vector as well. Results can also include a listing of the class probabilities. Post estimation results can also include predictions, partial effects, etc.