Traditional cointegration analysis asserts that the observed series are unit root processes, but a linear combination among these series is a short-memory stationary process. By allowing deviations from equilibrium to follow a fractionally integrated process, fractional cointegration analysis captures a wider range of mean-reverting behavior than traditional cointegration. Under the assumption that the integration orders of the underlying series are equal, based on the memory estimates from observed series and the regression residual, we propose a residual-based test for the null hypothesis of no fractional cointegration against the alternative of fractional cointegration. Under some regular conditions, the test statistic has an asymptotic standard normal distribution under the null and diverges under the alternative. The test is easy to implement and performs well in Monte Carlo experiments. A necessary condition for fractional cointegration is that the integrated orders of underlying series are equal. We present a procedure to test for the equality of integration orders of the underlying series. / Wang, Bin. / Adviser: Ngai Hang Chan. / Source: Dissertation Abstracts International, Volume: 70-06, Section: B, page: 3589. / Thesis (Ph.D.)--Chinese University of Hong Kong, 2008. / Includes bibliographical references (leaves 50-54). / Electronic reproduction. Hong Kong : Chinese University of Hong Kong, [2012] System requirements: Adobe Acrobat Reader. Available via World Wide Web. / Electronic reproduction. [Ann Arbor, MI] : ProQuest Information and Learning, [200-] System requirements: Adobe Acrobat Reader. Available via World Wide Web. / Abstracts in English and Chinese. / School code: 1307.