Recent developments in econometrics have revealed weak identification in empirical studies. The linear Instrumental Variable (IV) regression with weak instruments is an example that has received sizable attention; when instruments are weak, conventional asymptotics fail to function, and empirical results based on conventional asymptotics are unreliable. The problem of weak identification, however, is not limited to the linear IV regression or the Generalized Method of Moments (GMM) framework. In a broad range of economic models, the quality of inference depends on the treatment of weak identification. In my dissertation, I consider three issues related to weak identification. The first chapter proposes a method to detect whether weak identification exists. The objective of this chapter is to develop an intuitive tool which can be universally used in IV and GMM applications. The second chapter investigates the empirical studies of the Capital Asset Pricing Model (CAPM) and Consumption CAPM, in which various risk factors are suggested to explain the variation in asset returns. I find that irrelevant risk factors may still appear useful in explaining the variation, because of the weak identification problem induced by irrelevant factors. The third chapter targets the ongoing debate on whether technology shocks increase the hours worked. The debate derives from the highly persistent time series of hours, which makes the impact of technology shocks on hours weakly identified. I construct confidence intervals for this impact by adopting an approach robust to weak identification.
Zhan, Zhaoguo,
"Essays on Weak Identification"
(2011).
Economics Theses and Dissertations.
Brown Digital Repository. Brown University Library.
https://doi.org/10.7301/Z0RX99BT