Robustness

The authors adapt modern control theoretic techniques based on robust control theory to economic modelling and decision making. The main motivation behind the proposed approach is that concern about model misspecification in economics leads to decision strategies that work over the set of nearby models which may generate the specific set of data. The authors propose to measure discrepancies between models of this set (or in other words a quality of an approximating model) by the relative entropy and in this sense the robust control theory is not only adapted but also extended in the book. The main issues discussed in the book can be collected into six important sets of problems: 1. Formulation of discounted problems that preserve the recursive structure of decision problems (the authors seem not to know the monograph by Bertsekas and Rhodes in which such problems were discussed). 2. Representation of worst-case shock based on reformulation of misspecification perturbation to an approximating model. 3. Formulation of the role and use of multiple agents under assumption that although a common approximating model is created the agents have different interests and different concerns about robustness. 4. Interpretation of relationships between stochastic and nonstochastic models of uncertainty and resulting decision strategies. 5. Calibration of the measure of model uncertainty basing on detection of error probabilities. 6. Formulation of some robust filtering and estimation problems both in the case when a peculiar form of commitment to model distortions is chosen and when there is no commitment to the prior distortions. All ideas discussed in the book are illustrated by a variety of problems in dynamic macroeconomics. The authors concentrate on time-domain analysis and apply different techniques from dynamic noncooperative game theory to solve the formulated decision making problems. The book is self-contained and rigorous and may be interesting not only for macroeconomists who seek to improve the robustness of decision making process but also for control engineers interested in different applications of their professional abilities.