The Day of Week Effect

The day-of-week effect in Chinese ETF market

Xidan Xiao

Abstract

This paper studies weather the anomalous ‘day of week effect’: the pattern of day seasonality that found in many developed and developing markets around the world also exists in the emerging ETF market in China. We use the practical data in the Shanghai and Shenzhen stock market from 2010 to June 2014. Using the test for heteroskedasticity, we find out some of the ETFs presented heteroskedasticity , while some didn’t. We applied ARIMA model on the ETFs that didn’t’ present heteroskedasticity, while applied ARCH(1), GARCH(1),EGARCH(1) model on the ETFs that present heteroskedasticity. Our result examine the day-of-week effect that Monday and Thursday returns are negative in most of the sample ETFs, while the negative returns on Thursday are more significant. It also shows Wednesday has the highest return in the week.

1. Introduction

Monday effect refers to a phenomenon that in the stock markets, the mean returns for Monday have been significantly different from the returns for other days during a week. A lot of literatures have showed that the day of the week effect exists in most of the developed markets such as USA, UK&Canada and the emerging markets such as Malaysia and Hongkong. During the 1980’s, there were a lot of literatures investigating the Monday effects in the US stock market, see e.g., French(1980), Gibbons and Hess(1981), Rogalski(1984), and Keim and Stambaugh(1984). Recently, there were some literatures stated that the Monday effect in the US markets has gradually disappeared, see e.g. Mehdian and Perry(2001), Coutts and Hayes(1999), Steeley(2001). While most of the literatures have shown the negative Monday returns effect, a few literatures support a positive Monday return, e.g. Glenn(2003) and a few literatures does not support day-of-the-week effect e.g. Santemases(!986), Pena(1995), Gardeazabal and Rogulaz(2002). There are also a few literatures show negative returns on Tuesday in the stock markets of Canada, Australia, Hongkong, Japan, Korea, Singapore, Malaysia, Phillippines.

The beginning of China’s financial market is late, especially the area of ETF. The first ETF in China, SSE 50ETF was launched at the end of 2004. The success of ETFs symbolizes that Chinese investors recognize the benefits of index funds as an alternative investment vehicle with lower management fees. China's stock market has been developing rapidly, so as the ETF market. As an emerging market, China's stock market is not mature. The volatility of stock price is extremely high, especially in the year of 2012, which increases the opportunity for speculative investors. During the 1980’s, there were a lot of literatures investigating the Monday effects in the US stock market, see e.g., French(1980), Gibbons and Hess(1981), Rogalski(1984), and Keim and Stambaugh(1984). Recently, there were some literatures stated that the Monday effect in the US markets has gradually disappeared, see e.g. Mehdian and Perry(2001), Coutts and Hayes(1999), Steeley(2001). As a developing country, China’s market mechanism is different with US’s. The purpose of this study is to extend to prior literature by investigating the day of the week effect in Chinese ETF market during of the period between January2010 to March 2014, which is an area that none of the previous studies have investigated.

1. Data

2.1 Data collection

We use the database gathered by Shanghai Dazhihui Investing Concultation Co. Ltd to identify all ETFs that were in continuous operation for more than five years in both Shanghai stock exchange market and Shenzhen stock exchange market. We created an initial list of 10 ETFs from Shanghai stock exchange market and 4 ETFs from Shenzhen stock exchange market. We calculate the daily ETFs returns using adjusted trading price data from Dazhihui database.

2.2 Descriptive statistics for = Ln()[pic 1][pic 2]

Based on the Phillips-Perron result, the order of integration of the price time series is I(1). I use the returns = Ln() , to obtain a stationary series .[pic 3][pic 4]

1. Methodology

The sample statistics value for the return = Ln() are exhibited in Table 1. For the 14 ETFs, the kurtosis for all these samples are greater than three, which means the return distribution have excess kurtosis. Excess skewness is also observed for all the ETFs, leading to the possibility of un-constant variance.[pic 5][pic 6]

Table 1. Selected ETSs in Shenzhen and Shanghai stock exchange market