Saturday, 27 March 2010: 14:30
Title: Calendar Effects in Stock Markets: Recent Evidence in European Countries
Objectives:
This paper examines day of the week and month of the year effects in seventeen European stock market indexes in the period 1994-2007. We propose a model specification suitable for detecting all types of calendar effects, and use statically robust estimation methodologies, including bootstrapping and GARCH modeling.
Data/Methods:
The paper makes several contributions to the literature on calendar effects in stock market returns. First, we discuss the shortcomings of previously used models for the detection of calendar effects, and we propose a simpler model specification that overcomes those shortcomings. Second, we recognize non-normality and autocorrelation in stock market returns, and time-dependent variance of the residuals of linear regressions, and apply appropriate statistical methodologies to tackle these problems, including the bootstrap approach and the GARCH model, adding statistical robustness to our results. Third, we examine the time-stability of the most significant calendar effects in the period under study. Fourth, we use observations from a set of seventeen countries of the same economic region, allowing us to conclude if calendar effects are across-the-board effects in that region or only country-specific effects. Five, we use data from recent years, from 1994 to 2007, on West and Central European stock markets, thus adding and updating international evidence on calendar effects.
Results:
We find no strong convincing evidence of an across-the-board calendar effect in West and Central European countries. In particular, there are no statistically significant across-the-board January effects or weekend effects. European countries seem to be mostly immune to day of the week effects, although the group of seventeen countries, taken together, does tend to show higher daily returns on Thursdays and Fridays, and lower in Mondays in Tuesdays. If any, the only across-the-board calendar effect that warrants further investigation is the general tendency for lower returns in the holiday months of August and September. All the calendar effects are basically country-specific. The number of significant coefficients we detect is very similar to the number we would expect to find in random data. Even though there is some concentration on specific months / days of the week, our results are not immune to the critique that the calendar effects we detect are exclusively a result from intensive data mining. Using rolling windows regressions, we find that the stronger country-specific calendar effects are not stable over the whole sample period, casting additional doubt on the economic significance of calendar effects.
Objectives:
This paper examines day of the week and month of the year effects in seventeen European stock market indexes in the period 1994-2007. We propose a model specification suitable for detecting all types of calendar effects, and use statically robust estimation methodologies, including bootstrapping and GARCH modeling.
Data/Methods:
The paper makes several contributions to the literature on calendar effects in stock market returns. First, we discuss the shortcomings of previously used models for the detection of calendar effects, and we propose a simpler model specification that overcomes those shortcomings. Second, we recognize non-normality and autocorrelation in stock market returns, and time-dependent variance of the residuals of linear regressions, and apply appropriate statistical methodologies to tackle these problems, including the bootstrap approach and the GARCH model, adding statistical robustness to our results. Third, we examine the time-stability of the most significant calendar effects in the period under study. Fourth, we use observations from a set of seventeen countries of the same economic region, allowing us to conclude if calendar effects are across-the-board effects in that region or only country-specific effects. Five, we use data from recent years, from 1994 to 2007, on West and Central European stock markets, thus adding and updating international evidence on calendar effects.
Results:
We find no strong convincing evidence of an across-the-board calendar effect in West and Central European countries. In particular, there are no statistically significant across-the-board January effects or weekend effects. European countries seem to be mostly immune to day of the week effects, although the group of seventeen countries, taken together, does tend to show higher daily returns on Thursdays and Fridays, and lower in Mondays in Tuesdays. If any, the only across-the-board calendar effect that warrants further investigation is the general tendency for lower returns in the holiday months of August and September. All the calendar effects are basically country-specific. The number of significant coefficients we detect is very similar to the number we would expect to find in random data. Even though there is some concentration on specific months / days of the week, our results are not immune to the critique that the calendar effects we detect are exclusively a result from intensive data mining. Using rolling windows regressions, we find that the stronger country-specific calendar effects are not stable over the whole sample period, casting additional doubt on the economic significance of calendar effects.