Correlation does imply causation. Causality is embedded in the notions of cause and effect. If a variable Ut tends to cause Pt or, to put it somewhat differently, if Pt is the effect of Ut then, under such an outcome, the variable Ut would be causal to Pt. Deducing a causal relationship between economic variables is one of the most important and principal task of any econometric model. An elegant intrinsic worth of any study comes from the detection, if any, of causal inference between variables. Such evidence offers some guidance on appropriate policy implications.
The genesis, however, of understanding causality can be traced to natural sciences where the feasibility of executing controlled experiments has the potential to provide unambiguous results on the nature of causal relationship unli
ke economics or other social science domain (Lin, 2008, pp. 1). Causality has drawn immense attention of philosophers, statisticians, economists, scientists alike since the times of Aristotle. To be able to distinguish between “a cause and a concomitant effect” in the direction of the movement of any variable was reiterated in a technical fashion by David Hume (1742). Causality’s elementary interpretation, he imparted, was grounded on three empirical phenomenon – “contiguity”, “succession”, and “constant conjunction”. “Contiguity” meant that cause and effect must be tied together in order. Succession can be noticed because the effect is a by-product of cause. Most of the development of causality, since then, has invoked considerable friction among the economists on the issue of most suitable wa
y to test the causality83 (Hoover, 2006, pp. 2).
The discipline of Economic is grounded on general equilibrium which, in essence, applies that one particular variable has a tendency to depend on other variables concurrently. Hence, deciphering the cause, and effect, of one economic variable to another is central to any econometric analysis. They are, altogether, four possibilities that emerges when conducting a causality test for Ut and Pt – that is,
(a) Ut causes Pt
(b) Pt causes Ut
(c) that there is a bi-directional feedback (which means both the vari
ables seem to cause each
other), and finally
(d) the two variables are independent. The underlying challenge is to find out an appropriate
procedure that will enable us to detect accurate causality (Dimitrios and Hall, 2007).
Comments