Econometric Model Specification reviews and extends the author's papers on consistent model specification testing and semi-nonparametric modeling and inference. This book consists of two parts. The first part discusses consistent tests of functional form of regression and conditional distribution models, including a consistent test of the martingale difference hypothesis for time series regression errors. In the second part, semi-nonparametric modeling and inference for duration and auction models are considered, as well as a general theory of the consistency and asymptotic normality of semi-nonparametric sieve maximum likelihood estimators. Moreover, this volume also contains addendums and appendices that provide detailed proofs and extensions of all the results. It is uniquely self-contained and is a useful source for students and researchers interested in model specification issues.