The statistical models confronting econometricians are complicated in nature so it is no easy task to apply the procedures recommended by classical statisticians to such models. This book presents the reader with mathematical tools drawn from matrix calculus and zero-one matrices and demonstrates how the use of their tools greatly facilitates such applications in a sequence of linear econometric models of increasing statistical complexity. The book differs from others in that the matrix calculus results are derived from a few basic rules which are generalizations of the rules used in ordinary calculus. Moreover the properties of several new zero-one matrices are investigated.

This book presents the reader with new operators and matrices that arise in the area of matrix calculus. The properties of these mathematical concepts are investigated and linked with zero-one matrices such as the commutation matrix. Elimination and duplication matrices are revisited and partitioned into submatrices. Studying the properties of these submatrices facilitates achieving new results for the original matrices themselves. Different concepts of matrix derivatives are presented and transformation principles linking these concepts are obtained. One of these concepts is used to derive new matrix calculus results, some involving the new operators and others the derivatives of the operators themselves. The last chapter contains applications of matrix calculus, including optimization, differentiation of log-likelihood functions, iterative interpretations of maximum likelihood estimators, and a Lagrangian multiplier test for endogeneity.

This book presents the reader with new operators and matrices that arise in the area of matrix calculus. The properties of these mathematical concepts are investigated and linked with zero-one matrices such as the commutation matrix. Elimination and duplication matrices are revisited and partitioned into submatrices. Studying the properties of these submatrices facilitates achieving new results for the original matrices themselves. Different concepts of matrix derivatives are presented and transformation principles linking these concepts are obtained. One of these concepts is used to derive new matrix calculus results, some involving the new operators and others the derivatives of the operators themselves. The last chapter contains applications of matrix calculus, including optimization, differentiation of log-likelihood functions, iterative interpretations of maximum likelihood estimators and a Lagrangian multiplier test for endogeneity.

This 2002 book presents the reader with mathematical tools taken from matrix calculus and zero-one matrices and demonstrates how these tools greatly facilitate the application of classical statistical procedures to econometric models. The matrix calculus results are derived from a few basic rules that are generalizations of the rules of ordinary calculus. These results are summarized in a useful table. Well-known zero-one matrices, together with some newer ones, are defined, their mathematical roles explained, and their useful properties presented. The basic building blocks of classical statistics, namely the score vector, the information matrix, and the Cramer-Rao lower bound, are obtained for a sequence of linear econometric models of increasing statistical complexity. From these are obtained interactive interpretations of maximum likelihood estimators, linking them with efficient econometric estimators. Classical test statistics are also derived and compared for hypotheses of interest.

Recent advances in establishing the nature and scope of estimators in econometrics have shed more light on the importance of instrumental variables. In this book, the authors argue that such methods may be regarded as a strong organizing principle for a wide variety of estimation and hypothesis testing problems in econometrics and statistics. In support of this claim they present and develop the methodology of instrumental variables in its most general and explanatory form. They show, for instance, that techniques commonly used to handle simultaneity and related problems can be reduced to one of two generic variables of instrumental variables estimators, allowing them to explore further the conditions under which different proposed estimators are efficient.