Reevaluating the prudence of economic forecasts in the EU: The role of instrument persistence
指出Christodoulakis和Mamatzakis（2009）使用的GMM估计量在工具变量高度持久时具有非标准极限分布，并用对持久性稳健的方法和更新至2017年的数据重新检验，发现欧盟委员会损失偏好总体上仍偏向乐观，但变得更对称。
Summary Christodoulakis and Mamatzakis (2009, Journal of Applied Econometrics 24, pp. 583–606) estimate the EU Commission loss preferences for selected economic forecasts of 12 EU Member States. They employ the generalized method of moments (GMM) estimation procedure proposed by Elliott et al. (2005, Review of Economic Studies 72, pp. 1107–1125) and find the forecasts to be somewhat optimistic on average. However, this note shows the GMM estimator to possess nonstandard limiting distributions when some of the instruments are highly persistent, which is the case with one of the instruments employed by Christodoulakis and Mamatzakis. Standard distributions are recovered in some interesting particular cases which are relevant in practice. A reexamination of the EU Commission loss preferences using methods robust to persistence and a dataset extended to 2017 reveals that, while the conclusions of the original study are, by and large, still justified, the EU Commission loss preferences have become more symmetric over the whole studied period.