DETECTING LACK OF IDENTIFICATION IN GMM
提出一个检验非线性广义矩估计模型中识别不足的零假设的方法，可作为非线性版的线性工具变量第一阶段F检验，用于警告研究者常规渐近理论可能失效。
This paper proposes a test of the null of underidentification in the nonlinear-in-parameters generalized method of moments model. It can be thought of as a nonlinear analog of the usual linear instrumental variables first-stage F-test. It can be used as a diagnostic to warn a researcher when conventional asymptotic theory is unlikely to work well.I am grateful to Don Andrews, Jon Faust, John Fernald, Jim Stock, and two anonymous referees for their helpful comments on earlier drafts of this manuscript. I am also grateful to George Tauchen for providing me with the code for generating artificial asset price data. All errors are my sole responsibility.