Encompassing: formulation, properties and testing (Q1352563)

In many scientific investigations, a new theory is acceptable for explaining results if it is able to explain the results obtained from other competing theories known earlier besides possibly explaining some new results from the proposed new theory. This principle is named by econometricians as the encompassing principle [cf. \textit{G. E. Mizon} and \textit{J.-F. Richard}, Econometrica 54, 657-678 (1986; Zbl 0625.62105)]. A stochastic model \({\mathcal F}\) is said to encompass another model \(\zeta\) if it can account for the results, obtained under the model \(\zeta\). This monograph discusses the theory of encompassing. It describes a framework for a formal comparison of a pair of parametric models using the Kullback-Leibler information criterion and develops a methodology for testing hypotheses in the context of data generating processes with independent and identically distributed observations. The contents are as follows: Introduction; Chapter 1: Pseudo-tree values; Chapter 2: Encompassing; Chapter 3: Testing the encompassing hypothesis; Chapter 4: Encompassing tests for the linear model; Conclusion. The monograph is well-written and should be of interest to econometricians and statisticians interested in stochastic modelling and statistical inference.