|
Activity Number:
|
101
- Time Series Modeling: Mixed Frequency Data, Seasonality, and Model Identification
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Monday, August 8, 2022 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Business and Economic Statistics Section
|
|
Abstract #323074
|
|
|
Title:
|
A Nonstationary Time Series Model for Fractional Seasonal Periodicity
|
|
Author(s):
|
James Livsey*
|
|
Companies:
|
US Census Bureau
|
|
Keywords:
|
seasonal adjustment;
unit roots;
signal extraction
|
|
Abstract:
|
Many applications warrant modeling seasonal time series with fractional (non-integer) periods. In classic Box-Jenkins time series methodology, a suitable transformation from observed nonstationary series is performed. For example, with monthly time series a typical choice might be first differencing and/or differencing at lag 12. These transforms are characterize by their unit roots, which exactly remove seasonal dynamics associated with frequencies that are multiples of the seasonal period. The current methods for handling fractional seasonal periods involve truncated Taylor series expansions for the seasonal differencing operators. We show this current approach to be inadequate in that it does not exactly remove seasonal dynamics when the period is fractional. This research proposes corrected operators and demonstrates the utility with an empirical study of the U.S. Census Bureau’s Business Formation Statistics.
|
Authors who are presenting talks have a * after their name.
Back to the full JSM 2022 program
|
