Evaluating Econometric Forecasts of Economic and Financial Variables
本书系统介绍如何评估线性与非线性计量经济模型及调查预期的预测效果，涵盖无偏性检验、预测能力比较等理论，并应用于产出增长、失业、通胀和股票收益等实证案例。
The book by Michael Clements is an interesting monograph on the evaluation of forecasts obtained from linear and nonlinear econometric models and survey based expectations. The book is a useful tool for researchers and practitioners alike as it discusses the theory of forecasts evaluation and provides relevant empirical applications involving among others output growth rates, unemployment and inflation rates, and stock returns. After an introductory chapter, in chapter 2, the author introduces some of the more established techniques for evaluating sequences of point forecasts and forecasts errors in terms of the properties of unbiasedness and efficiency, and the reasons for choosing the conditional expectations as point forecasts. Also more recent important developments are discussed. These include tests for unbiasedness, equal predictive ability of rival sets of point forecasts, and forecasts encompassing. The use of these techniques for survey based forecasts is also considered. Chapter 3 is dedicated to models of volatility or conditional variance and the evaluation of the forecasts of conditional variance. It is well known that modelling and forecasting the conditional variance of a series is important for improving our point forecasts, provided that the conditional variance is varying in a systematic fashion and we are able to model it. The most popular parameterisations of ARCH and GARCH models are described and their forecasting abilities are compared to those obtained by averaging past squared returns.