Testing the specification of nonlinear time series models is crucial in applied statistics, macroeconomics and finance to make relevant analysis. It is often not enough to check only conditional moments, while specification of the conditional distribution function, which is equivalent to jointly specifying all conditional features of a process, including the conditional mean, variance, and quantiles, is required. Knowledge of the conditional distribution is necessary in many applications.
For instance, knowing the true conditional distribution is important to apply efficient maximum likelihood (ML) methods. The dissertation considers tests of conditional distribution with an application to financial models in mind and consists of three parts: 1) testing univariate time series models, 2) testing multivariate distribution and 3) testing dynamic choice models.
© 2001-2024 Fundación Dialnet · Todos los derechos reservados