Dynamic Input-Output Analysis - alternative approaches

The aim of this dissertation is to present extended dynamic input-output analysis. Dynamic input-output analysis involves an extensive range of methods based on the same data background represented in various forms of input-output tables. The methodology starts out with a theoretical discussion, it provides a comparison of the current background for the input-output analysis to its more common alternative, namely general equilibrium models. A significant portion of this thesis also concentrates on regional application and mathematical methods suitable for data actualization, especially the RAS method and the entropy method. The analysis of general equilibrium models enhances the study for the clarification of basic principles and differences between these models and input-output models. A theoretical profound description and an explanation of the link between general equilibrium models and input-output models is further supplemented by application examples. Special attention is dedicated to an estimation of capital matrices. These types of matrices commonly represent a missing data source necessary for the construction of dynamic models. I have added two original approaches to estimate symmetric capital matrices. In addition, the data estimation procedure for the Czech Republic for the year 2013 includes these suggested, unique methods. The ensuing chapters disclose mathematical tools to disaggregate the preceding aggregate estimates obtained from the case study for the Czech Republic for the regional level. This thesis also presents an original modification of the RAS method for multi-dimensional space and compares it with a multi-dimensional entropy method. The outcome of the study proves the convergence of these two methods. This finding can be of great importance to other researchers as it enables one to substitute a minimalization task for iteration methods and thus significantly reduces the computational power to deliver an analysis focusing on the regional level. The study culminates with an estimation of a complex inter-regional model with six methods covering the current, most commonly applied methods as well as my original approaches. Various criteria including the experimental approach within input-output models and general equilibrium models enable a comparison of these estimated models. The findings prove the great sensitivity of the obtained values from data estimation depending on the choice of the selected method. This outcome appears crucial for the construction of interregional models.