Academic journal article The Journal of Philosophical Economics

The Unrealistic Realist Philosophy. the Ontology of Econometrics Revisited

Academic journal article The Journal of Philosophical Economics

The Unrealistic Realist Philosophy. the Ontology of Econometrics Revisited

Article excerpt

(ProQuest: ... denotes formulae omitted.)


Considering their target systems, there are two types of models: (1) models of phenomena and (3) statistical models of data (Saatsi 3017; Frigg and Hartmann 3006). Econometrics is an empirical branch of economics that constructs models employing statistical and quantitative methods with the aim of uncovering empirical regularities in the observational, economic data (Tintner 1953). Econometric models are models of data defined as 'a corrected, rectified, regimented, and in many instances idealized version of the data gained from immediate observation, the so-called raw data' (Frigg and Hartmann 3006, pp. 743). Generally speaking, such models are obtained by estimating parameters that minimize the error term s and choosing the functional form of the following equation:

(1) ...

Yt - endogenous variable (output)

Xit - exogenous variables (determinants/inputs)

F(...) - a relation between a set of exogenous variables and endogenous variable

'Scientific realism' is an umbrella term (Leplin 1984, 1) that covers many versions of this stance. This positive approach to science is usually defined as the commitment to the (1) metaphysical/ontological, (2) semantic, and (3) epistemic dimension (Psillos 2005; Mäki 2005). In other words, SR accepts that models (1) describe the discourse-independent reality, (2) are made true or false by how the world is (i.e., true models isolate or idealize certain aspects of reality), and (3) (most successful ones) are (approximately) true. Papineau (1996, 2) defined being a realist about science as accepting the (1) independence thesis and (2) knowledge thesis. The former states that the world is independent of our discourse about it. The latter acknowledges that it is feasible to decide if a model is (at least approximately) right. Saatsi (2017) indicated that what is typical for different versions of realism is the belief that science is, to some degree, successful and grasps something (at least partially) about the world.

In line with Mäki's (2005) argument that various versions of realism are adequate for distinct sciences, the article focuses on analyzing and arguing against the localized version of SR. The realist philosophers interested in econometrics put forth a few slightly differentiated stances grounded in studying most successful (regarding adequacy and acceptance among economists) empirical models. In general, the SR interpretations of econometrics are realist about empirical regularities between variables depicted by models. Such regularities are interpreted either as probabilistic causal laws (Cartwright 1994, pp. 149-150; Hoover 2002) or as functional and correlational relations (Hoover 2010) that idealize or isolate such a relation in the world of economy. In other words, the SR readings of econometrics accept that (1) the empirical models are about the world (the independence thesis) and (2) it is feasible to indicate the successful models that rightly 'resemble' the regularities produced by the world of the economy (the knowledge thesis). Considering that (3) econometric models often contradict each other (cf. Moosa 2019; the case studies discussed below), being realist about econometrics equals to being committed to the trilemma of jointly inconsistent views.

The structure of the article is as follows. First, the scientific realist interpretations of econometrics are reviewed. Second, the following feature of econometrics absent from the hitherto realist interpretation is discussed: econometric models (at least in some areas of application) are shown to depend on minor changes in sample (such as including an additional variable or further observations) or method (e.g., applying weighted or unweighted averaging scheme and linear or exponential regression). Despite being inconsistent, such pairs of contrary models are accepted by the community of econometricians. Third, the hitherto SR interpretations of econometrics are argued to be inconsistent with the realistic reconstruction of the project. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.