Encompassing

Abstract

In many scientific fields, it is a widespread idea that a theory should be able to explain the results obtained by other, competing theories. This idea, although implicitly adhered to by many authors, has only recently been in­troduced formally for the purpose of evaluating empirical statistical models. In the econometrics literature, it became known as the encompassing prin­ciple through the work of Hendry, Mizon and Richard, who advocated and formalized it in a number of papers (Hendry-Richard [1982, 1983, 1990], Mizon [1984] and Mizon-Richard [1986]). See also Govaerts [1987], Smith [1993] and White [1994]. A recent comprehensive treatment of encompassing is given by Gouriéroux-Monfort [1995]. Loosely speaking, a model F is said to encompass a model g if it can account for the results obtained within the context of g.