Online Program Home
  My Program

All Times EDT

Abstract Details

Activity Number: 196 - Time Series Methods with Seasonal, Monthly, and Daily Data
Type: Contributed
Date/Time: Tuesday, August 4, 2020 : 10:00 AM to 2:00 PM
Sponsor: Business and Economic Statistics Section
Abstract #313584
Title: Testing for Time-Variation in Trading-Day Effects on Monthly Time Series
Author(s): Thomas Triumbur* and William Bell
Companies: US Census Bureau and U.S. Census Bureau
Keywords: Seasonality; Day-of-Week Effects; Seasonal Adjustment; Time-Varying Coefficient Models; Daily effects for monthly data; Likelihood ratio tests
Abstract:

Trading-day, or day-of-the-week effects, are important in a large number of monthly economic time series, and are regularly estimated and removed as part of seasonal adjustment. These adjustments are generally made via models that use fixed regression effects, and so assume that the trading-day effects are constant over time. Concerns have been raised as to whether the daily effects (the trading-day coefficients) may actually change over time. To this end Harvey (1989) and Bell and Martin (2004) proposed stochastic models for trading-day effects that allow the coefficients to vary as random walks. Likelihood ratio tests with such models can be used to give a quantitative and coherent criterion for answering the question of whether trading-day effects in a given time series are fixed over time or stochastically time-varying. We apply such models to a large number of Census Bureau monthly economic time series to test for the presence of time-varying trading-day effects, and also to estimate their overall practical significance.


Authors who are presenting talks have a * after their name.

Back to the full JSM 2020 program