Theoretical and Real Trading-Day Frequencies

• Activity Number: 549
• Type: Invited
• Date/Time: Wednesday, August 1, 2012 : 2:00 PM to 3:50 PM
• Sponsor: Section on Government Statistics
• Author(s): Dominique Ladiray*+
• Companies: INSEE
• Keywords: Trading-day frequencies ; X-12-ARIMA

Abstract

Most economic time series are computed according to the Gregorian calendar, a solar calendar brings rhythm to our lives and has a deep impact on the economy. The most well known and important calendar effect is seasonality. But most economic indicators are also linked to a daily activity which is usually summed up and reported each month or each quarter. In this case, the number of working days, which varies from one month to another, can explain some short-term movements in the series. Apart from the day composition of the month, other calendar effects such as public holidays or religious events may also affect the series. In this presentation, we focus on the trading-day effect caused by the day composition of the month. From the periodograms corresponding to the 7 days of the week, we derive the main theoretical trading-day frequencies. We exhibit the most common real trading-day frequencies from a large set of economic indicators. Finally, we check the accuracy of the visual spectral test implemented in X-12-ARIMA, using simulated series. The results permit us to propose some improvements to the test, but it is also shown the F-test will almost always perform better.