Correlation and causation 


Correlation does not imply causation when interpreting statistical relationships between economic variables. This is especially problematic for investment and skilled-labour demand. Both are needed for production to take place, and must precede the goods & services produced and consumed. Yet this future consumption is the driver of the earlier production.

Proving evidence of causality is important in the econometric estimation of relationships. In the absence of expectations, temporal causality (i.e. one set of events occurring systematically after another set of (causal) events) may be sufficient proof. This is the basis of Granger’s Causality Test.

Unless variables are properly defined, errors in causation can be very damaging. Expansionary austerity policies, for example, appear to have been based on errors in methods leading to reverse causality between public-sector deficits and GDP growth.

References:

Breuer, C. (2019) Expansionary Austerity and Reverse Causality: A Critique of the Conventional Approach, INET Working Paper.

View all resources
Go to Top