Empirical evidence on Seasonalities – Inspecting the market anomalies

Résumé du document

Anomalies, such as calendar effects, could be defined as phenomena where systematic returns patterns are observed. Partisans to informational market efficiency suggest that calendar effects are prone to disappear, as investors in the market could implement strategies in order to take advantage of these anomalous behaviors. Thus we often evocate the ‘self destruction property of anomalies'. However, according to Marquering, Nisser & Valla study's over 1960 – 2003 on DJIA, some calendar effects seem to persist, despite their publication in academic journals. In particular, “turn of the month effect” has been persistent over the time and does not seem to have disappeared. “Week-end effect” and “January effect” were still persistent in the 1990's, although they seem to have progressively disappeared recently. Consequently, have the main calendar effects disappear over the time? The objective of this paper is to determine, whether or not, main calendar effects were persistent over 1998 - 2001 in several European countries and in the United States. The next section synthesizes main previous studies on calendar effects. Then section 3 describes the methodology used for our study.

Extraits

[...] The next section synthesizes main previous studies on calendar effects. Then section 3 describes the methodology used for our study. Section 4 investigates the results obtained for intra-week, intra- month and intra-year effects for our seven countries over 1998 - 2001. Finally, section 5 concludes this study Calendar Anomalies & Previous Research An efficient capital market is one where security prices adjust rapidly and accurately to new information. Current prices, in such a market reflect all relevant and ascertainable information. [...]

[...] As expected, we do not observe September Effect in Belgium, France, US, Germany and Poland Conclusion According to our results, calendar anomalies seem to have disappeared. In Belgium, the United States and Germany, no intra-week, intra-month or intra- year anomalies are observed over our study period (1998 2001). However, a Thursday & Friday effect (with positive returns) persists in France. This Friday effect (with positive returns) seems also persist in the Netherlands. Furthermore, significant positive returns are observed in Poland in the first week of the month. Finally a September effect (with negative returns) persists in UK and the Netherlands. [...]

[...] Tests are executed under E-Views Empirical Results & Interpretation A. Intra-Week Results According to the F-statistic, daily returns on Mondays, Tuesdays, Wednesday, Thursday and Fridays are significantly different in France at significance level and in the Netherlands at 10% significance level. Estimated Daily Returns Significantly different from 0 at significance level Significantly different from 0 at significance level Actually, for both France and the Netherlands, we observe significant daily returns on Thursdays and on Fridays. More precisely, in France, returns are significantly positive on Thursdays at significance level and significantly positive on Fridays at 10% significance level. [...]

[...] Furthermore, although these returns are not significant, their estimates are positive in the other countries during the first week of the month. For the other weeks of the month and the last days of the month, no significant return's behaviour is observed in our seven countries. Consequently, a first week effect seems to persist in both France and Poland. G. First Week Effect In order to test such an hypothesis we construct the following model: , where R is the daily return, D1 represents the first week and D2 represents the second, the third, the fourth weeks and the last days of the month. [...]

[...] Second, a Student test where , j ( enables us to determine the significantly high or low returns. Similar models and tests are executed to highlight intra-month and intra- year effects. For intra-month effects, we use five dummy variables, where D1 represents the first week (from the 1st to the 7th), D2, the second week (from the 8th to the 14th), D3, the third week (from the 15th to the 21st), D4, the fourth week (from the 22nd to the 28th) and D5 the last days of the month (from the 29th to the 31st). [...]