A nondegenerate Vuong test

• 成果类型: Article
• 署名作者: Shi, Xiaoxia
• 署名单位: University of Wisconsin System; University of Wisconsin Madison
• 刊物名称: QUANTITATIVE ECONOMICS
• ISSN/ISSBN: 1759-7323
• DOI: 10.3982/QE382
• 发表日期: 2015
• 页码: 85-121
• 关键词: Asymptotic size; model comparison; nonnested models; voter turnout; Vuong test; C12; C52

摘要

In this paper, I propose a one-step nondegenerate test as an alternative to the classical Vuong (1989) tests. I show that the new test achieves uniform asymptotic size control in both the overlapping and the non-overlapping cases, while the classical Vuong tests do not. Meanwhile, the power of the new test can be substantially better than the two-step classical Vuong test and is not dominated by the one-step classical Vuong test. An extension to moment-based models is also developed. I apply the new test to the voter turnout data set of Coate and Conlin (2004) and find that it can yield model comparison conclusions different from those of the classical tests. The implementation of the new test is straightforward and can be done using the MATLAB and STATA routines that accompany this paper.