Note—Analysis of Time Series with Calendar Effects

Most national economic data and many marketing series are compiled monthly according to the Gregorian Calendar, but some of the ancient festivals or holidays, such as Easter, Jewish Passover, and Chinese New Year, are set by lunar calendar. Therefore the date of a holiday may vary between two adjacent months from year to year. Some product marketing and consumer behavior patterns are closely related to such holidays; the amount of monthly sales may change as the date of the holiday changes from one month to the other. This paper shows how calendar intervention influences the sample autocorrelation, both by a theoretical study and an actual example. Model estimation for a time series subject to calendar intervention is also discussed.