|
Abstract:
|
Instrumental variables is an approach used to estimate regression parameters when covariates are correlated with errors. Applications of instrumental variables are common in econometrics among other fields where the method is regarded as helping to uncover causal relationships between covariates and a response. Statistically, the method succeeds in producing a consistent point estimator of the regression coefficients. However, standard interval estimates for the regression coefficients are unreliable. We present a new technique, using inferential models, to provide valid inference in instrumental variables regression.
|
