Realism and Instrumentalism in the Development of Econometrics

WHEN, in the early decades of this century, various economists took a critical stance towards the application of formal probabilistic methods to the material of economic time-series, the proponents of these methods, whatever their intentions, apparently conveyed the view that all legitimate concerns were being attended to. In fact, this impression seems to have been so pervasive amongst the econometrics community that a somewhat dismissive attitude came to be adopted towards those economists, including Keynes, who continued to display any reservations about the soundness and relevance of the project (See e.g. Koopmans, 1941; Haavelmo, 1943a; Schumpeter, 1946). As Vining (1949b) observes economists came to regard Keynes' concerns about econometrics as the "conclusions of a sadly misinformed and misguided man" (p. 93). In truth, however, it is not clear that such concerns have ever been adequately met in practice. Indeed, today, as economists find themselves reluctantly still puzzling over the essential nature of, and the claims that can legitimately be made for, econometric analysis, the sort of concerns put forward by Keynes and others are generally accepted as being as relevant and pressing as ever (see e.g. Meeks, 1978; Hendry, 1980; Lawson, 1985; Pesaran and Smith, 1985a, 1985b; Epstein, 1987; Gilbert, 1987; Rowley and Hamouda, 1987; Rowley and Jain, 1987; and Gillies, 1988). A relevant question, then, is what lies behind this state of affairs? Why are the sort of worries expressed by Keynes still being resurrected? And how were these concerns allayed in the first place? In what follows I want to address these sorts of issues from an explicit philosophy of science vantage point. A general neglect of philosophical considerations in this area has often been noted, (e.g. Caldwell, 1982, p. 216) and what is proposed can best be considered as an exploratory first step. Specifically I want to delineate two largely oppositional positions, realism and instrumentalist (defined below), and to indicate their bearing on the development of the subject in general and particularly on the issues referred to above. This philosophical opposition has often been found to provide a leverage to a better understanding of developments in the 'natural sciences' and in what follows I want to suggest that its explanatory potential with regard to econometric analysis may be no less significant.