Correlation versus Causation
From Open Risk Manual
Definition
Correlation versus Causation denotes a common debate in the context of Model Development of Statistical Models for Quantitative Risk Management which can be summarized as the difficulty of establishing cause-and-effect relationships between two (or more) variables on the basis of an observed associations or quantitative correlations between them.
The issue is an example of Model Risk.
Implications for Risk Management
In the context of a Risk Model the challenge can arise in two contexts:
- establishing stable causal relationships between the set of explanatory variables ( risk factors)
- establishing stable causal relationships between explanatory variables and risk events or risk outcomes
See Also
- For a general discussion in the context of statistics see wikipedia:Correlation does not imply causation