Advances in Computational Management Science – serie

Optimal Control and Dynamic Games has been edited to honor the outstanding contributions of Professor Suresh Sethi in the fields of Applied Optimal Control. Professor Sethi is internationally one of the foremost experts in this field. He is, among others, co-author of the popular textbook "Sethi and Thompson: Optimal Control Theory: Applications to Management Science and Economics". The book consists of a collection of essays by some of the best known scientists in the field, covering diverse aspects of applications of optimal control and dynamic games to problems in Finance, Management Science, Economics, and Operations Research. In doing so, it provides both a state-of-the-art overview over recent developments in the field, and a reference work covering the wide variety of contemporary questions that can be addressed with optimal control tools, and demonstrates the fruitfulness of the methodology.

Portfolio Management with Heuristic Optimization consist of two parts. The first part (Foundations) deals with the foundations of portfolio optimization, its assumptions, approaches and the limitations when "traditional" optimization techniques are to be applied. In addition, the basic concepts of several heuristic optimization techniques are presented along with examples of how to implement them for financial optimization problems. The second part (Applications and Contributions) consists of five chapters, covering different problems in financial optimization: the effects of (linear, proportional and combined) transaction costs together with integer constraints and limitations on the initital endowment to be invested; the diversification in small portfolios; the effect of cardinality constraints on the Markowitz efficient line; the effects (and hidden risks) of Value-at-Risk when used the relevant risk constraint; the problem factor selection for the Arbitrage Pricing Theory.

Management science is often confronted with optimization problems characterized by weak underlying theoretical models and complex constraints. Among them, one finds data analysis, pattern recognition (classification, multidimensional analysis, discriminant analysis) as well as modelling (forecasting, confirmatory analysis, expert system design). In recent years, bio-mimetic approaches have received growing attention from marketing, finance and human resource researchers and executives as effective tools for practically handling such problems. Bio-mimetic approaches include a variety of heuristic methods - such as neural networks, genetic algorithms, immunitary nets, cellular automata - that simulate nature's way of solving complex problems and, thus, can be considered as numerical transpositions of true life problem solving. This text presents a selection of papers on bio-mimetic approaches and their application to management science. Most of these papers were presented at an ACSEG/CAEMS International Congress (Approches Connexionnistes en Sciences Economiques et de Gestion/Connectionnist Approaches in Economics and Management Science).All papers combine the discussion of conceptual issues with illustrative empirical applications, and contain detailed information on the way heuristics are practically implemented. The advantages and limits of the bio-mimetic approaches are discussed in several of the papers, either by comparing these approaches with more classical methods (logit models, clustering), or by investigating specific issues like overfitting and robustness. Synthesizing overviews are provided, as well as tools for coping with some of the limitations of bio-mimetic methods.

This text analyzes decision problems in the management and engineering sciences. It is divided into five parts. The first part explores methodological issues involved in the optimization of deterministic and stochastic dynamical systems. The second part describes approaches to the model energy and environmental systems and draws policy implications related to the mitigation of pollutants. The third part applies quantitative techniques to problems in finance and economics, such as hedging of options, inflation targeting, and equilibrium asset pricing. The fourth part considers a series of problems in production systems. Optimization methods are put forward to provide optimal policies in areas such as inventory management, transfer-line, flow-shop and other industrial problems. The last part covers game theory. Chapters range from theoretical issues, to applications in politics and interactions in franchising systems.

Optimal Control of Credit Risk presents an alternative methodology to deal with a financial problem that has not been well analyzed yet: the control of credit risk. Credit risk has become recently the center of interest of the financial community, with new instruments (such as Credit Risk Derivatives) and new methodologies (such as Credit Metrics) being developed. The recent literature has focused on the pricing of credit risk. On the other hand, practitioners tend to eliminate credit risk rather than price it. They do so via collateralization. The authors propose here a methodological basis for an optimal collateralization.The monograph is organized as follows: Chapter 1 reviews the main avenues of literature related to our problem; Chapter 2 provides a brief overview of the main optimal control principles; and Chapter 3 presents the models and their setting.In the remaining chapters, the authors propose two sets of programs. One set of programs will apply in cases where the information on the assets=value is readily available (full observation case), while the other applies when costly audits are needed in order to assess this value (partial observation case).In either case, the modeling stage leads to a set of quasi-variational inequalities which the authors attempt to solve numerically in the simpler case of full observations. This is done in Chapter 6. Finally a simulation analysis is carried out in Chapter 7, in which the authors study the influence on the control process of changes in the different model parameters. This precedes a discussion on possible extensions in Chapter 8 and some concluding remarks in Section 9.

This volume contains selected papers that were presented at the International Conference COMPUTATIONAL FINANCE 1997 held at London Business School on December 15-17 1997. Formerly known as Neural Networks in the Capital Markets (NNCM), this series of meetings has emerged as a truly multi-disciplinary international conference and provided an international focus for innovative research on the application of a multiplicity of advanced decision technologies to many areas of financial engineering. It has drawn upon theoretical advances in financial economics and robust methodological developments in the statistical, econometric and computer sciences. To reflect its multi-disciplinary nature, the NNCM conference has adopted the new title COMPUTATIONAL FINANCE. The papers in this volume are organised in six parts. Market Dynamics and Risk, Trading and Arbitrage strategies, Volatility and Options, Term-Structure and Factor models, Corporate Distress Models and Advances on Methodology. This years' acceptance rate (38%) reflects both the increasing interest in the conference and the Programme Committee's efforts to improve the quality of the meeting year-on-year. I would like to thank the members of the programme committee for their efforts in refereeing the papers. I also would like to thank the members of the computational finance group at London Business School and particularly Neil Burgess, Peter Bolland, Yves Bentz, and Nevil Towers for organising the meeting.

This volume contains selected papers that were presented at the International Conference "Computational Finance" held at the London Business School. Formerly known as "Neural Networks in the Capital Markets" (NNCM), this series of meetings has emerged as a multi-disciplinary international conference and provided an international focus for research on the application of a multiplicity of advanced decision technologies to many areas of financial engineering. It has drawn upon theoretical advances in financial economics and robust methodological developments in the statistical, econometric and computer sciences. To reflect its multi-disciplinary nature, the NNCM conference has adopted the new title "Computational Finance". The papers in this volume are organized in six parts: market dynamics and risk; trading and arbitrage strategies; volatility and options; term-structure and factor models; corporate distress models; and advances in methodology.

Optimal control and differential games continue to attract strong interest from researchers interested in dynamical problems and models in management science. This volume explores the application of these methodologies to new as well as to classical decision problems in management sciences and economics. In Part I, optimal control and dynamical systems approaches are used to analyze problems in areas such as monetary policy, pollution control, relationship marketing, drug control, debt financing, and ethical behavior. In Part II differential games are applied to problems such as oligopolistic competition, common resource management, spillovers in foreign direct investments, marketing channels, incentive strategies, and the computation of Markov perfect Nash equilibria. Optimal Control and Differential Games is an excellent reference for researchers and graduate students covering a wide range of emerging and revisited problems in management science.

Since the beginning of the 1980's, many new approaches of biomimetic inspiration have been defined and developed for imitating the brain behavior, for modeling non-linear phenomenon, for providing new hardware architectures, for solving hard problems. These approaches include: neural networks, multilayer perceptrons, genetic algorithms, cellular Automates, and self-organizing maps. They can be summarized by the word "connectionism" and consist of an interdisciplinary domain between neuroscience, cognitive science and engineering. First they were applied in computer sciences, engineering, biological models, pattern recognition, motor control, learning algorithms, etc. However, it rapidly appeared that these methods could be of great interest in the fields of economics and management sciences. The main difficulty was the distance between researchers, the difference in the vocabulary used and their basic background. The main notions used by these new techniques were not familiar to the social and human sciences researchers.The purpose of the book is to put these new techniques at the disposal of researchers coming from different horizons, to assess the state of the art, to identify the capability of these new algorithms, to evidence the contribution of these methods to economics and management sciences. The contributions in this book bring new confirmations of the interest of connectionist approaches for researchers in economics and management sciences. The first part is dedicated to theoretical advances; the second to a wide range of applications. All papers contain interesting results on each subject, which would have been very difficult to show with classical techniques but which has been proven by using these connectionist non-linear methods.

Optimal Control and Dynamic Games has been edited to honor the outstanding contributions of Professor Suresh Sethi in the fields of Applied Optimal Control. Professor Sethi is internationally one of the foremost experts in this field. He is, among others, co-author of the popular textbook "Sethi and Thompson: Optimal Control Theory: Applications to Management Science and Economics". The book consists of a collection of essays by some of the best known scientists in the field, covering diverse aspects of applications of optimal control and dynamic games to problems in Finance, Management Science, Economics, and Operations Research. In doing so, it provides both a state-of-the-art overview over recent developments in the field, and a reference work covering the wide variety of contemporary questions that can be addressed with optimal control tools, and demonstrates the fruitfulness of the methodology.

Portfolio Management with Heuristic Optimization consist of two parts. The first part (Foundations) deals with the foundations of portfolio optimization, its assumptions, approaches and the limitations when "traditional" optimization techniques are to be applied. In addition, the basic concepts of several heuristic optimization techniques are presented along with examples of how to implement them for financial optimization problems. The second part (Applications and Contributions) consists of five chapters, covering different problems in financial optimization: the effects of (linear, proportional and combined) transaction costs together with integer constraints and limitations on the initital endowment to be invested; the diversification in small portfolios; the effect of cardinality constraints on the Markowitz efficient line; the effects (and hidden risks) of Value-at-Risk when used the relevant risk constraint; the problem factor selection for the Arbitrage Pricing Theory.

Management Science is often confronted with optimization problems characterised by weak underlying theoretical models and complex constraints. Among them, one finds data analysis, pattern recognition (classification, multidimensional analysis, discriminant analysis) as well as modelling (forecasting, confirmatory analysis, expert system design). In recent years, biomimetic approaches have received growing attention from Marketing, Finance and Human Resource researchers and executives as effective tools for practically handling such problems. Biomimetic approaches include a variety of heuristic methods - such as neural networks, genetic algorithms, immunitary nets, cellular automata - that simulate nature's way of solving complex problems and, thus, can be considered as numerical transpositions of true life problem solving. Bio-Mimetic Approaches in Management Science presents a selection of recent papers on biomimetic approaches and their application to Management Science. Most of these papers were presented at the last ACSEG/CAEMS International Congresses (Approches Connexionnistes en Sciences Economiques et de Gestion/Connectionnist Approaches in Economics and Management Science). All papers combine the discussion of conceptual issues with illustrative empirical applications, and contain detailed information on the way heuristics are practically implemented. The advantages and limits of the biomimetic approaches are discussed in several of the papers, either by comparing these approaches with more classical methods (logit models, clustering), or by investigating specific issues like overfitting and robustness. Synthesizing overviews are provided, as well as new tools for coping with some of the limitations of biomimetic methods.

Decision & Control in Management Science analyzes emerging decision problems in the management and engineering sciences. It is divided into five parts. The first part explores methodological issues involved in the optimization of deterministic and stochastic dynamical systems. The second part describes approaches to the model energy and environmental systems and draws policy implications related to the mitigation of pollutants. The third part applies quantitative techniques to problems in finance and economics, such as hedging of options, inflation targeting, and equilibrium asset pricing. The fourth part considers a series of problems in production systems. Optimization methods are put forward to provide optimal policies in areas such as inventory management, transfer-line, flow-shop and other industrial problems. The last part covers game theory. Chapters range from theoretical issues to applications in politics and interactions in franchising systems. Decision & Control in Management Science is an excellent reference covering methodological issues and applications in operations research, optimal control, and dynamic games.

Since the beginning of the 1980's, a lot of news approaches of biomimetic inspiration have been defined and developed for imitating the brain behavior, for modeling non linear phenomenon, for providing new hardware architectures, for solving hard problems. They are named Neural Networks, Multilayer Perceptrons, Genetic algorithms, Cellular Automates, Self-Organizing maps, Fuzzy Logic, etc. They can be summarized by the word of Connectionism, and consist of an interdisciplinary domain between neuroscience, cognitive science and engineering. First they were applied in computer sciences, engineering, biological models, pattern recognition, motor control, learning algorithms, etc. But rapidly, it appeared that these methods could be of great interest in the fields of Economics and Management Sciences. The main difficulty was the distance between researchers, the difference in the vocabulary used by the ones and the others, their basic background. The main notions used by these new techniques were not familiar to the Social and Human Sciences researchers. What are they ? Four of them are now very briefly introduced, but the reader will find more information in the following chapters.

Optimal control and differential games continue to attract strong interest from researchers interested in dynamical problems and models in management science. This volume explores the application of these methodologies to new as well as to classical decision problems in management sciences and economics. In Part I, optimal control and dynamical systems approaches are used to analyze problems in areas such as monetary policy, pollution control, relationship marketing, drug control, debt financing, and ethical behavior. In Part II differential games are applied to problems such as oligopolistic competition, common resource management, spillovers in foreign direct investments, marketing channels, incentive strategies, and the computation of Markov perfect Nash equilibria. Optimal Control and Differential Games is an excellent reference for researchers and graduate students covering a wide range of emerging and revisited problems in management science.

Optimal Control of Credit Risk presents an alternative methodology to deal with a financial problem that has not been well analyzed yet: the control of credit risk. Credit risk has become recently the center of interest of the financial community, with new instruments (such as Credit Risk Derivatives) and new methodologies (such as Credit Metrics) being developed. The recent literature has focused on the pricing of credit risk. On the other hand, practitioners tend to eliminate credit risk rather than price it. They do so via collateralization. The authors propose here a methodological basis for an optimal collateralization.The monograph is organized as follows: Chapter 1 reviews the main avenues of literature related to our problem; Chapter 2 provides a brief overview of the main optimal control principles; and Chapter 3 presents the models and their setting.In the remaining chapters, the authors propose two sets of programs. One set of programs will apply in cases where the information on the assets=value is readily available (full observation case), while the other applies when costly audits are needed in order to assess this value (partial observation case).In either case, the modeling stage leads to a set of quasi-variational inequalities which the authors attempt to solve numerically in the simpler case of full observations. This is done in Chapter 6. Finally a simulation analysis is carried out in Chapter 7, in which the authors study the influence on the control process of changes in the different model parameters. This precedes a discussion on possible extensions in Chapter 8 and some concluding remarks in Section 9.

This book presents the latest findings on stochastic dynamic programming models and on solving optimal control problems in networks. It includes the authors’ new findings on determining the optimal solution of discrete optimal control problems in networks and on solving game variants of Markov decision problems in the context of computational networks. First, the book studies the finite state space of Markov processes and reviews the existing methods and algorithms for determining the main characteristics in Markov chains, before proposing new approaches based on dynamic programming and combinatorial methods. Chapter two is dedicated to infinite horizon stochastic discrete optimal control models and Markov decision problems with average and expected total discounted optimization criteria, while Chapter three develops a special game-theoretical approach to Markov decision processes and stochastic discrete optimal control problems. In closing, the book’s final chapter is devoted to finite horizon stochastic control problems and Markov decision processes. The algorithms developed represent a valuable contribution to the important field of computational network theory.

This book presents the latest findings on stochastic dynamic programming models and on solving optimal control problems in networks. It includes the authors’ new findings on determining the optimal solution of discrete optimal control problems in networks and on solving game variants of Markov decision problems in the context of computational networks. First, the book studies the finite state space of Markov processes and reviews the existing methods and algorithms for determining the main characteristics in Markov chains, before proposing new approaches based on dynamic programming and combinatorial methods. Chapter two is dedicated to infinite horizon stochastic discrete optimal control models and Markov decision problems with average and expected total discounted optimization criteria, while Chapter three develops a special game-theoretical approach to Markov decision processes and stochastic discrete optimal control problems. In closing, the book’s final chapter is devoted to finite horizon stochastic control problems and Markov decision processes. The algorithms developed represent a valuable contribution to the important field of computational network theory.

Advanced computational methods are often employed for the solution of modelling and decision-making problems. This book addresses issues associated with the interface of computing, optimisation, econometrics and financial modelling. Emphasis is given to computational optimisation methods and techniques. The first part of the book addresses optimisation problems and decision modelling, with special attention to applications of supply chain and worst-case modelling as well as advances in the methodological aspects of optimisation techniques. The second part of the book is devoted to optimisation heuristics, filtering, signal extraction and various time series models. The chapters in this part cover the application of threshold accepting in econometrics, the structure of threshold autoregressive moving average models, wavelet analysis and signal extraction techniques in time series. The third and final part of the book is about the use of optimisation in portfolio selection and real option modelling.

Advanced computational methods are often employed for the solution of modelling and decision-making problems. This book addresses issues associated with the interface of computing, optimisation, econometrics and financial modelling. Emphasis is given to computational optimisation methods and techniques. The first part of the book addresses optimisation problems and decision modelling, with special attention to applications of supply chain and worst-case modelling as well as advances in the methodological aspects of optimisation techniques. The second part of the book is devoted to optimisation heuristics, filtering, signal extraction and various time series models. The chapters in this part cover the application of threshold accepting in econometrics, the structure of threshold autoregressive moving average models, wavelet analysis and signal extraction techniques in time series. The third and final part of the book is about the use of optimisation in portfolio selection and real option modelling.

Many decision problems in Operations Research are defined on temporal networks, that is, workflows of time-consuming tasks whose processing order is constrained by precedence relations. For example, temporal networks are used to model projects, computer applications, digital circuits and production processes. Optimization problems arise in temporal networks when a decision maker wishes to determine a temporal arrangement of the tasks and/or a resource assignment that optimizes some network characteristic (e.g. the time required to complete all tasks). The parameters of these optimization problems (e.g. the task durations) are typically unknown at the time the decision problem arises. This monograph investigates solution techniques for optimization problems in temporal networks that explicitly account for this parameter uncertainty. We study several formulations, each of which requires different information about the uncertain problem parameters.