The Czech Economy Model: Bayesian Quantitative Analysis of Parameters and Control of Strategies

Facilities of adaptive Bayesian identification as well as utilizing a suitably-rebuilt linearized model as a model controlled by an optimal control system will be demonstrated in this paper. Estimation of time-variant parameters was performed by mean of extended Kalman-Bucy filter with smoothing on base of rather short quarterly time series (dated from 1992, quarter I, to 1996, quarter II), that are not purged of neither seasonalities nor quick jumps due to sudden events in the Czech transitive economy during the period in question. After simultaneous verification, the quantified model was rebuilt into the form of non-linear state equations, which is suitable for applying control. ln handling the future trajectories of state and input variables (i.e. control and exogenous variables, respectively), there was the optimal control task solved as a dual problem  oj optimal control of a linearized discrete-time model (with time-variant parameters) by means of a <em>quadratic junctional(with time-variant weights) which is to be minimized. The above procedures were applied to the Small Czech Economy's Model, where we were trying to find out some admissible input and output trajectories, which would represent up-to-date factual alternative strategies of development of the Czech economy.</em>