Causal explanation as a fine art
This paper that deals with the explanation of the behavior of dynamic economic systems, an explanation characterized by a chain of presumed causes and effects among the phenomena that integrate its spatio-temporal domain, it is developed from the proposals of Suppes, Giere, Morgan and Granger's empiricist definition of stochastic causality and an axiomatic consisting of the biconceptual interpretation of causality, Lewis's "principal principle" and the properties of Bayesian conditioning. In accordance with these principles assumed as starting points and the postulates of the general theory of stochastic processes in the framework of a deterministic paradigm, the following results were obtained: (i) that, as a correction to the treatment usually devoted to it in the literature, the validity of the main principle is limited to an "absolutely dominant" Bayesian conditioning, that is, to a situation of absolute conviction of chance and that its non-fulfillment implies that neither the uniqueness nor the rationality of the belief function can be guaranteed; (ii) that, in the field of subjective Bayesianism, the only condition for the rationality of the belief function is coherence, that is, the satisfaction of the axiomatic of de Finetti's "theory of foresight"; iii) that, given the assimilation of rationality to the principle of strict coherence and the assimilation of this principle to the static concept of the degree of belief, the consequent impossibility of measuring the variations of the probabilities of the hypotheses in the face of variations in the information set, can be solved by substituting this conditioning by a method of inference based on the principle of maximum entropy, which provides an imprecise belief function; iv) that, as an extension of the epistemic theory of probability, it is possible to postulate an epistemic theory of causality, according to which the degrees of belief in the existence of a causal relation admit a calibration with respect to chance, by means of the convex combination of pairs of probability functions representative of chance; v) that, in order to generalize the application of the principal principle to the explanation of dynamic economic systems, it is possible to replace the classical condition that considers the admissible evidence from a frequentist interpretation of chance conditioning by the postulates of the ergodic partition theorem.
Area: Eshet Conference
Keywords: Dynamic economic systems; economic theory of causality; biconceptual probability.
Please Login in order to download this file