PhD defence B. (Bram) van Os

Below is a brief summary about the dissertation:

The conditional distribution of many macroeconomic and financial variables is subject to considerable temporal variation, which fuels the demand for dynamic models. For example, there are often substantial differences in the distributional characteristics of macroeconomic variables between expansionary periods and recessions. This thesis contains four chapters on time-varying parameter models for macroeconomic and financial data. First, Chapters
2 and 3 have a strong empirical character, examining respectively dating turning points in business cycles and filtering
of large time-varying correlation matrices. Both applications develop problem-specific extensions of popular existing
time-variant parameter models and focus on making real-time predictions. Second, Chapters 4 and 5 focus on the theoretical aspects of dynamic models, in particular relating to filter optimality and stability. Specifically, these chapters introduce new general frameworks for the construction of time-varying parameter models by addressing the problem of updating of conditional density from one period to the next as an optimisation problem.