THE DAY OF THE WEEK EFFECT

The Day of the Week Effect

Executive Summary

This study examines the daily seasonalities in London Stock Exchange. Results have shown the existence of significant day-of-the -week-effects both in overnight interest rate changes and stock returns. Overnight interest rates significantly fall on Wednesdays and increase on Mondays relative to the previous days. At the stock market, returns are significantly higher in the second part of the w eek and lower in the first two days of the week. Seasonalities in volatility and liquidity across the week confirmed the prior findings. Dynamic trading strategies based on the daily seasonalities in stock returns are able to produce significantly higher returns than market return of a simple "buy and hold" strategy in some periods neglecting transaction costs. The Day of the Week Effect

Introduction

This effect relates to the difference in returns across different days of the week. The main findings have been lowest and usually negative returns on Mondays and exceptionally large returns on Fridays as compared to other days of the week. The variance in stock returns is found to be largest on Mondays and lowest on Fridays. A study by Wang et al. (1997) finds that the negative Monday returns occur in the last two weeks of the month and that mean Monday returns for the first three weeks of the month are not significantly different from zero.

The international evidence of this effect has been somewhat mixed. Dubois and Louvet (1996) find returns to be lower for the beginning of the week (but not necessarily Monday) for European countries, Hong Kong and Canada. They did, however, observe that the anomaly disappeared in the USA for the most recent periods. Agrawal and Tandon (1994), find negative Monday returns in nine countries and negative Tuesday returns in eight countries (out of a total of 19 countries). Also, the Tuesday returns are lower than Monday returns in eight countries. Draper and Paudyal (2001) find that once fortnight, ex-dividend day, account period, news flow, trading activity and bid-ask spread effects are controlled for the Monday returns do not differ significantly from returns of other days.

Studies done on bond markets have also acknowledged the day-of-the-week effect (Kohers and Patel, 1996; Adrangi and Ghazanfari, 1996) but results are significantly different from that in the stock market. Monday returns are found to be positive on the average.

Many theories have been postulated to explain the day-of-the-week effect with the popular ones are as follows.

Settlement period hypothesis: This attributes seasonality across days of the week to the settlement dates with prices being higher on the pay-in days as compared to the pay-out days. This theory has been opposed by some as the anomaly holds across markets that have different settlement periods (varying from one day in France and Hong-Kong to six-fifteen days in the UK - Agrawal and Tandon, 1994).

Calendar time/trading time hypothesis: Calendar time trading implies that Monday returns should be three times as high as that for other days because Monday returns are spread ...