New Procedures for Multiple Testing of Econometric Models

The nonexperimental nature of economics makes it important econometricians be able to test their models using accurate procedures. There are a large number of diagnostic tests that can be applied. A major problem is how best to control the overall probability of rejecting the model when it is true. We will develop and investigate a new set of computer-simulation-based procedures for multiple testing of econometric models. This will enable econometricians and statisticians to build better models and to better understand the weaknesses of their models.