A Further Look at Model Evaluation

摘要: In this chapter we compare the specification tests generally used in econometric model building with recently developed p-step Stone-Geisser prediction using jackknife procedures. The comparisons are based on a Monte Carlo study different structural forms. forms partly misspecified. Three sets of simulations presented: A variety single equation models, simple multi-equation macro model, and Klein’s Model I US data. statistics enable more critical evaluation out-of-sample performance estimated models.