Deterministic and stochastic algorithms for stabilization policies with large-size econometric models
Abstract
The ultimate aim in building an econometric model is always, at least implicitly, to develop a tool for optimal policy selection. Yet optimal control methods are not as popular as this remark would seem to imply. This is partly due to widespread dissatisfaction with the theoretical status of large-size models: while the theoretical literature is mainly concerned with the concept of equilibrium, the long-run equilibrium properties of large-size models are not usually investigated, so that their consistency with theory cannot be assessed. A further reason for the limited use of optimal control methods is the formidable computing problem of applying these techniques to large-size nonlinear stochastic systems. This work develops operative procedures to implement several alternative optimal control algorithms and suggests some modifications specifically designed to handle the case of stochastic optimal control. These algorithms are then used for a variety of experiments with the Bank of Italy Quarterly Model. It is argued that the scope for using optimal control algorithms is not limited to policy-making. We use optimal control techniques in order to compute the steady-state growth path of the Bank of Italy model. We show how the model needs to be modified for this purpose, and perform comparative statics experiments to highlight its equilibrium properties, as a step towards full characterization of its theoretical status. Optimal control techniques are then used to stabilize the model in a more realistic set-up, where the decision-maker re-computes the optimal policy every time new information becomes available. and the model is subject to stochastic disturbances. Within this framework we investigate the value of information in the policy-making process, by comparing the results that are obtained on different assumptions about the sources and the timing of new information.
Subject Area
Economics|Statistics
Recommended Citation
Siviero, Stefano, "Deterministic and stochastic algorithms for stabilization policies with large-size econometric models" (1995). Dissertations available from ProQuest. AAI9543144.
https://repository.upenn.edu/dissertations/AAI9543144