Statistical inference on the parameters of interest in a vast number of econometric models is plagued by its dependence on the settings of secondary nuisance parameters. Examples of such models are wide spread and include:
- Linear instrumental variables regression
- Inference in linear regression models after model selection
- Generalized method of moments
- Dynamic stochastic general equilibrium models
- Inference on parameters of large scale models with sparsity estimated by Lasso
- Dynamic panel data models
- Vector autoregressions with cointegration
These models are widely used by applied researchers. Inference on their parameters can be hampered by the value of underlying/secondary nuisance parameters. It was long thought that such settings of the nuisance parameters are unrealistic so they can basically be ignored for applied purposes. The empirical relevance of so-called weak instruments shows, however, that this is not the case. Weak instruments are a commonality in applied work so it is important to have inference methods that are robust to them. The same argument also applies to the other models stated above.
The conference brings together the leading researchers on inference issues in econometric models. During the two day meeting they present around twenty research papers on a variety of the above stated models.
Organizing committee of the 12th Tinbergen Institute Conference: Inference Issues in Econometrics was: