Model Selection for Panel Data with Fixed Effects : A Simulation Study
- 주제(키워드) Information criteria , Fixed effects estimation , Panel data , Monte Carlo simulation
- 발행기관 서강대학교 일반대학원
- 지도교수 최인
- 발행년도 2010
- 학위수여년월 2010. 2
- 학위명 석사
- 학과 일반대학원 경제학과
- 실제URI http://www.dcollection.net/handler/sogang/000000045772
- 본문언어 영어
- 저작권 서강대학교의 논문은 저작권에 의해 보호받습니다
초록/요약
This paper considers model selection using the Akaike's Information Criterion (AIC), the corrected Akaike's Information Criterion (AICc), and the Bayesian Information Criterion (BIC) for panel data with fixed effects. Applying these information criteria to fixed effects panel models is not a trivial matter due to the incidental parameters problem that might adversely affect their practical performance, especially when T is small. Through the Monte Carlo experiments it has been found that the information criteria are fairly successful in selecting the true model. In particular, given the existence of the incidental parameters problem in the data generating process, the AICc and the AIC operate successfully unless T is extremely small. This study, therefore, concludes that the information criteria--which are simple to use as well as effective--will likely help empirical researchers determine the proper structure of the error term in fixed effects estimation in an objective manner.
more