Econometric Analysis of the Sequential Probit Model: Application to Innovation Surveys

We analyze a two stage sequential probit model that can be used to analyze survey in which questions are asked sequentially and unobservable variables of each stage can be correlated. We apply the model to innovation surveys. In the first stage, the survey asks if firms have developed a product or a process innovation. In the second stage, the survey inquires about the degree of novelty o the innovation. We estimate parameters of this model in a classical framework in which multiple integrals that arise in the likelihood function are estimated by simulation and in a Bayesian framework in which we use the latent var iable structure of the model to implement a Metropolis Gibbs sampler. We determine the role of information sources on the innovation process and show that they significantly influence the nature of the innovation process.