A0362
Title: Model selection in panel data model with large number of fixed effects
Authors: Zhaoyuan Li - The Chinese University of Hong Kong, Shenzhen (China) [presenting]
Abstract: Due to the bad control problem, fixed effects are used to replace control variables in panel regression models. However, a large number of fixed effects will lead to a false positive problem. Suppose that an insignificant result is obtained from a panel model, but a significant result may come up by adding more fixed effects. One question is how to decide which one is correct. And then, what can we do if it is a false positive? A new model selection approach is proposed to answer these two questions.
