In the January 2005 version of their working paper entitled “Testing the Significance of Calendar Effects”, Peter Reinhard Hansen, Asger Lunde and James Nason test a broad range of possible calendar effects in multiple equity markets. They examine the following effects: day-of-the-week, month-of-the-year, day-of-the-month, week-of-the-month, semi-month, turn-of-the-month, end-of-the-year and holiday. Calendar effects could be a result of data mining (finding anomalies of randomness), an especially plausible explanation when theoretical explanations are suggested only subsequent to empirical “discovery.” Applying robust tests to daily closing prices of stock indices from Denmark, France, Germany, Hong Kong, Italy, Japan, Norway, Sweden, the United Kingdom and the United States through early May 2002, they find that:
- Calendar effects are statistically significant for returns in almost all of the 25 stock indices from the ten countries studied.
- End-of-the-year (especially), week-of-the-month-of-the-year, and week-day-of-the-month effects stand out as the largest anomalies.
- The strongest evidence for calendar effects is for small-cap and mid-cap indices.
- Calendar effects are generally not economically significant because in many cases the last instances of significance occurred in the late 1980s and early 1990s. Subsequent to this period, there is no evidence of significant calendar effects except in small-cap stock indices. There may be a time variation in calendar effects not consistent with systematic seasonal variation in stock returns.
In summary, calendar effects used to be but mostly aren’t any more.
