AEA期刊的IV靠不靠谱？

欢迎投稿(荐稿)计量经济圈，计量相关都行

邮箱：[e-mail]

编辑：计量经济圈

计量经济圈希望圈友继续支持咱们品牌，我们回馈的是，没有任何其他公众号比我们提供给大家学习机会更多，只要你愿意付出机会成本迈过这个门槛。

下面这篇文章是伦敦经济学院Alwyn Young用自助法对发表在AEA期刊上的31篇文章里用到的1400个工具变量回归进行的计量检验，发现了那些关于弱工具变量的检验往往检测不出来，而且用弱工具变量稳健方法做的回归表现得比二阶段最小二乘法要糟糕得多。

下面只列出了摘要和部分引言，且不论计量结果，想看关于IV（工具变量）的paper，不妨从我们后面附上的paper里学习一下这个方法。

Consistency without Inference:   Instrumental Variables in Practical Application*

Alwyn Young

London School of Economics

I use the bootstrap to study a comprehensive sample of 1400 instrumental variables regressions in 32 papers published in the journals of the American Economic Association.  IV estimates are more often found to be falsely significant and more sensitive to outliers than OLS, while having a higher mean squared error around the IV population moment.  There is little evidence that OLS estimates are substantively biased, while IV instruments often appear to be irrelevant.  In addition, I find that established weak instrument pre-tests are largely uninformative and weak instrument robust methods generally perform no better or substantially worse than 2SLS.

I:  Introduction
The economics profession is in the midst of a “credibility revolution”
(Angrist and Pischke 2010) in which careful research design has become firmly established as a necessary characteristic of applied work.  A key element in this revolution has been the use of instruments to identify causal effects free of the potential biases carried by endogenous ordinary least squares regressors. 

The growing emphasis on research design has not gone hand in hand, however, with equal demands on the quality of inference.  Despite the widespread use of Eicker (1963)-Hinkley (1977)-White (1980) robust and clustered covariance estimates, in finite samples heteroskedastic and correlated errors still produce test statistics whose distribution is typically much more dispersed than believed. 

This complicates inference in both ordinary and two stage least squares, but is compounded in the latter, where first stage joint test statistics are used to buttress the credibility of second stage results.  In this paper I show that two stage least squares (hereafter, 2SLS or IV) methods produce estimates that, in practice, rarely identify parameters of interest more accurately or substantively differently than is achieved by biased ordinary least squares (OLS).

I use the bootstrap to study the distribution of test statistics for a
comprehensive sample of 1533 instrumented coefficients in 1400 2SLS
regressions in 32 papers published in the journals of the American Economic Association.  I maintain, throughout, the exact specification used by authors and their identifying assumption that the excluded variables are orthogonal to the second stage residuals.  

且不论计量结果，想看关于IV（工具变量）的paper，不妨从我们后面附上的paper里学习一下这个方法。

点击下图，有惊喜

 写在后面：各位圈友，咱们的计量经济圈社群里面资料和计量咨询都很多，希望大家能够积极加入咱们这个大家庭（戳这里）。之后我们会逐步邀请社群里的圈友再直接建立微信群与圈圈对话，进去之后一定要看“群公告”，不然接收不了群息。