Abstract

Diebold and Rudebusch (1991) and Haubrich (1993) argue that, when income follows a fractionally differenced process, the Deaton's excessive smoothness paradox can be resolved. A key to the success of their result relies on a valid test for fractional integration. However, most of the tests in the literature are nested within fractional alternatives. This paper designs a new test for a more general hypothesis that the true data generating process is indeed fractionally integrated. The test is applied to the real disposable income per capita of the U.S. and the real quarterly GDP data of the G7 industrial countries.

Recommended Citation

Melvin J. Hinich and Terence T.L. Chong (2007) "A Class Test for Fractional Integration", Studies in Nonlinear Dynamics & Econometrics: Vol. 11: No. 2, Article 5.
http://www.bepress.com/snde/vol11/iss2/art5

Related Files

hinich_datacode.zip (4 kB)
Data and code