Modelling Seasonality with Fractionally Integrated Processes

We propose in this article the use of a particular version of the tests of Robinson (1994) for testing seasonally fractionally integrated processes. The tests have standard null and local limit distributions and allow us to test unit and fractional seasonal roots even with different amplitudes at different frequencies. A Monte Carlo experiment is conducted to check the power of the tests against different types of fractional alternatives and, an empirical application, using quarterly data for the U.S. total expenditure of several monetary aggregates is also carried out at the end of the article