Exogeneity
The condition where an explanatory variable is entirely independent of the error term in a statistical model.
What is Exogeneity?
Exogeneity implies that a variable's value is determined by factors outside the specific model being analyzed, meaning it is uncorrelated with the unobserved disturbances (the error term) affecting the dependent variable. Strictly exogenous variables allow for unbiased estimation of their causal effects on the outcome.
Why Exogeneity Matters
Establishing exogeneity is the cornerstone of causal inference in observational studies. When predictors are exogenous, researchers can trust that the estimated coefficients accurately reflect the true causal impact rather than spurious correlations driven by unobserved factors.
Example
In an agricultural study evaluating the effect of rainfall on crop yields, the amount of rainfall is typically considered an exogenous variable because it is determined by weather patterns outside the model and is not influenced by the crop yields themselves.
Common Mistakes
- Failing to recognize that true exogeneity is extremely rare in social sciences unless utilizing randomized experiments or natural experiments.
- Confusing exogeneity with a variable simply being fixed or predetermined, without checking if it correlates with unobserved shocks.