Are the month-of-the-year effects in China caused by government interventions?

(1)

Are the month-of-the-year effects in China caused by

government interventions?

Zibo Yang

Faculty of Economics and Business Rijksuniversiteit University

(2)

Abstract

This paper examines the month-of-the-year effects in Chinese stock market which has not been a subject of considerable amount in previously academic research. Compared to those mature markets, the calendar anomalies in emerging markets can be attributed to very various reasons since many differences between these two kinds of markets. In the contest of China, the stock market is strongly manipulated by government policies. It’s evidenced that in several specific months, the interventions of government on the Chinese stock market are more profound than that in other months. This suggests there are possibilities that the political window dressing can explain the month of the year effects in Chinese cases, which indicate that interventions by government play the dominant role. By using GARCH-in-Mean model, this paper observes the March effect in Chinese stock market and argues that the causes for this March effect are from government interventions.

(4)

Ⅰ. Introduction

The topics of financial anomalies such as seasonalities and calendar anomalies have been investigated extensively in the literature of finance. Many existences of financial anomalies have been disclosed by previous researches. Researchers also have collected a great deal of evidence against the main factors causing these financial anomalies. However, majority of researches concentrate on the mature markets either in North America or Europe. The fields of emerging markets haven’t been given much attention.

Chinese stock market as an important market with long history in Asia is entering the fast lane these years. Up to 2008, the Shanghai and Shenzhen stock exchanges had more than 1,500 companies listed. The combined market capitalization of both exchanges had reached US$2,658.2 billion which is approximately equal to the Hong Kong Stock Exchange (US$2,121.8bn), ranking as Asia's second-largest stock market behind the Tokyo Stock Exchange (US$3,925.6bn). It’s reported that Chinese stock market will overtake Japanese stock market to be the largest by 2010.

Not only in Asia, has Chinese stock market broken 2.8 trillion RMB to be the biggest emerging stock market up to 2007. According to the report to the 17th CPC National Congress, China will improve its financial system and work out a complete, efficient and secure modern financial market. Experts believe that the Chinese market will rank itself among emerging financial centers in the next few years. The important position of Chinese stock market makes it worth being studied.

(5)

and Li (2006), Jiang et al (2008), etc., report the month-of-the-year effects pattern in their research.

Due to the gap in development, there are many differences between emerging and mature markets. As a matter of fact, most of the emerging markets are still struggling with some general problems which are inevitable during their initiative periods. One of the IMF Working Paper by Yartey (2008) about the macroeconomic environment of 42 emerging markets, the political factor, implementation of law and order, and bureaucratic quality are the most crucial determinants of the development of emerging stock markets. All these problems mentioned above will probably result in different month of the year effect in emerging market, compared to well-developed market.

Chinese stock market encounters various problems because several reforms of systems or institutions during the transition from plan-base economy to market-base economy are not completed. Take the credit systems for instance, the credit systems under planned economy are vanished, while under market economy these systems are still under construction. In this way, Chinese stock market is in the midst of the problems like non-transparent operations, inadequate institution, weak rule of law and the lack of strong supervision for financial practitioner. This makes China a good example to analyze month of the year effect from the perspective of particular institutional details typical in emerging market.

(6)

economic issues are very highly politicized in China. The core reasons of this kind of policy-driven economy comes from the consolidation of the leaderships of authorities. In China, the economic policy issues will affect the leadership succession in numerous ways. In one way, social stability is the overwhelming concern for the top leaders of China because it plays a very important role in maintaining the incumbent in power. Since the dramatic decline or fluctuation leads to social chaos or resentment, the pursuit of stock market stability is the prerequisite for social stability. Consequently, Chinese political authorities impose impelling policies on stock market to ensure its stability. In another way, economic growth is also a dominant factor of sustaining the positions of leaders. Thus, the priorities of top leaders are the fast development, emphasizing either successful or credible stock markets.

Based on all these characteristics mentioned above, Chinese stock market is a weighty case for analyzing the month-of-the-year effect in emerging markets. Most importantly, the causes for month-of-the-year effect probably are apart from those widely accepted hypothesis for developed markets, owing to the close relationship between politics and economy. Thus, it’s argued that it’s meaningful to do research on month-of-the-year effect in China from a new perspective taking political factor into consideration.

(7)

interventions carried out by the Chinese government.

My main contribution in this paper is fourfold. Firstly, my study enriches the research on China’s month-of-the-year effects by analyzing it from a political view. The evidence on government interventions as the main attribution of China’s month-of-the-year effect is presented. Secondly, this paper applies GARCH-M model to exercise the main tests relating the risk factor to the results of return. By taking into the risk factor into consideration, the statistic analysis is more acceptable than those without such concerns. Thirdly, I extend the month-of-the-year effect research in China to Hang Seng China Enterprises Index (HSCEI) and Chinese Companies listed in NSDAQ (known as Concept Shares) which are seldom examined in the past. Lastly, in this paper, both Shanghai and Shenzhen indexes are classified as A Shares and B Shares by the different regulations or requirement upon them. Such classification brings the disguisable interventions behavior of authorities more visible and easy to detect.

This paper is structured as follows. The next part, Section Ⅱ reproduces the previously studies on month-of-the-year effects in several markets and especially in Chinese stock market. Section Ⅲ indicates the distinctive characteristics of Chinese stock market between politics and economy and presents an argument that the month-of-the-year effects in China can be caused by political interventions. The next section explains the econometric formula and model used in this paper. Section V provides the main tests and the results and the possible explanations of the results is considered. A brief conclusion is presented in the last part.

Ⅱ. Previous Reviews

(8)

which has been widely detected in several stock markets.

As one kind of calendar anomalies, month-of-the-year effects are documented in many previous researches. Not only the existences of month-of-the-year effects are widely disclosed in stock markets, but also the causes for them are proven by researches. However, majorities of reasons for month-of-the-year effects are based on mature markets, the investigations about month-of-the-year effects in emerging markets are very limited. These attributions for developed markets involve tax-loss selling hypothesis (to create the tax losses by year-end selling), information hypothesis (selling securities due to the disclosure of information), holiday season hypothesis (to derive more cash-flow during the main holidays for shopping reason), modeling hypothesis (errors on the data collection or statistical methodology).

ⅰ Previous research on month-of-the-year effects in mature markets

Among all forms of month-of-the-year effects, January effect is definitely a better-recognized one. The first observation on January effect was conducted by Donald Keim (1983). He examines month-by-month, the empirical relation between abnormal returns and market value of NYSE and AMEX common stocks over the period 1963 to 1979. There is evidence that daily abnormal return distributions in January have large means relative to the remaining eleven months, and that the relation between abnormal returns and size is always negative and more pronounced in January than in any other month. Keim shows that the small firms’ premium is always positive in January during this period. He reports that nearly fifty percent of the average annual size effect can be attributed to the month of January and more than half of the January effect occurs during the first week of trading. Large firms earn even larger risk-adjusted returns compared to small firms. In this paper, Keim also notes that these high January returns accrue disproportionately to small firms and especially during the early days of January.

(9)

in the UK (Clare et al, 1995), Australia (Brown et al, 1982), New Zealand (Raj and Thurston, 1994) and Canada (Berges et al, 1984). Clare et al (1995) analyze seasonal fluctuations in the UK equity market during the period 1955 to 1990. Their results suggest that seasonal variation is similar across size groups and not confined to small firms. However, the market tends to rise in both January, April and to a lesser extent in December and fall in September. Brown, Keim, Kleidon and Marsh (1982) examines the month-to-month small firm return premium for a sample of Australian stocks for the period 1958 to 1981. They find that the raw returns for most Australian stocks exhibit pronounced December-January and July-August seasonality, with the largest effects in January and July. Australian returns show an average premium of at least 4%; per month for the smallest 10% of the firms relative to the other deciles, and this premium appears to be fairly constant across months in contrast to U.S. data. Raj and Thurston (1994) observes no month-of-the-year effects in New Zealand. They believe both the small size and poor liquidity of the market may attribute to this result. Berges et al. (1984) analyze the Canadian data for the period from 1951 to 1980 and find that January effect is present in Canadian market.

Most of the month-of-the-year effects found above are consistent with the “tax-loss selling” hypothesis. As the tax laws are various in each developed markets, the month-of-the-year effects is different accordingly.

(10)

coefficient for both August and September and they attribute these findings to “information hypothesis”.

Alagidede and Panagiotidis (2006) research both the day of the week and month of the year in the stock returns in Ghana. Different from a January return pattern in most markets, an April effect for Ghana’s stock market is found. Al-Khazali (2003) used stochastic dominance and parametric analyses to examine the turn-of-the-year and week effects of Jordan’s market from 1978 to 2001. Their results indicate that returns of Amman financial market exhibit substantial deviation from normality. The parametric analysis tests demonstrate there are January and week effects. However, stochastic dominance results show that January and week effects don’t exist in the AFM. Fountas and Segredakis (2002) test for the January effect and the tax-loss selling hypothesis using monthly stock returns in 18 emerging markets for period 1987-1995. Little evidence in favour of the January effect is shown.

ⅲ

Previous research on month-of-the-year effects in China

(11)

New Year. According to their researches, the holiday effect around the Spring Festival period is stronger and more persistent compared to other public holidays in the rest of the year. Fan and Dong (2007) evidence the stock return of A Shares in March exceeds the rest or the others in a year, while December has the lowest monthly return. After they keep the B/M effects, β risk premium and size effects constant, the month-of-the-year effects are still present. Thus, they conclude the causes for the month-of-the-year effects come from the whole market not only some effects influence some parts of the markets. Jiang et al (2008) detects relatively high stock return in both March and April.

(12)

Brown, Keim, Kleidon and Marsh (1982) extend the discussion of the tax-loss selling hypothesis based on Australia market since Australia has similar tax laws but a July-June tax year. In Australia, the tax year-end is June 30 and the tax treatment of capital gains or losses is such that the tax-loss selling hypothesis predicts a July seasonal in returns for small stocks. Results indicate that Australian returns show pronounced December-January and July-August seasonality, and a premium for the smallest-firm deciles of about four percent per month across all months.

However, in China, it has a counter example in Chinese tax law because the capital gains are free of taxes. Hence, the tax motivated selling in the tax-year end to reduce the taxable speculation loss by a capital loss is the case to be observed in Chinese stock market. Obviously, tax-loss selling hypothesis is not suitable for Chinese case.

Besides tax-loss hypothesis, Rozeff and Kinney (1976) also provide another explanation known as information hypothesis. This hypothesis is related to the information released by firms at certain times in fiscal year. The information dissemination may vary with different categories of investors during the disclosure period and also the level and speed of information dissemination is different. This kind of variety probably lead to insider control popping up the returns in the information release time. Miller (1990) also show a phenomenon that people postpone some decisions like investment decision so as to buy gifts during Christmas, which result in the reduction in the speed of their reaction to information. For Chinese case, there is no common rule for company listed in exchange to publish information in a certain time.

(13)

biases in data measurement. Thus, this hypothesis is not proper to Chinese stock market.

The holiday and consumption habit hypothesis is capable to explain some month-of-the-year effects as well. This explanation is based on both the specific consumption habit of investor during some important holidays. Fan and Dong (2007) finds the March and December effects in China and they present the consumption habit hypothesis to explain these effects. According to their research, the excessive returns in March /December are caused by the consumption habit of Chinese investors during Chinese Lunar New Year / New Year. It’s detected that a large spending boom during Chinese Lunar New Year and New Year.

The window dressing theory is an additional explanation for month-of-the-year effects which is result from the efforts of institutions or regulators to affect investor’s perceptions of portfolio performance. Several researches on Chinese stock market evidence the window dressing theory can be the causes for month-of-the-year effects in China.

Liu and Chen (2004) find that the window dressing explanation of January effect. They notice that in January, the fund managers tend to increase investment in the fund and accordingly the returns in January increase. However, they don’t give the exact reasons why fund managers act like this. Liu et al (2003) reveals that the existence of manipulation in the Chinese stock market, such as wash sales and spread of false information in the market. They also believe that these kinds of manipulation have big influence in the stock markets.

(14)

window dressing, like fund managers are not really the significant actors in shaping the stock market. Heilmann finds these market participants are under the more powerful force which is the Chinese government. Therefore, Heilmann shows that the Chinese government is a key reason to shape the market. Most importantly, Heilmann demonstrate that this kind of interventions tend to happen on specific months which may lead to month-of-the-year effects.

If the window dressing activities are relevant for explain the month-of-the-year effects in China, then this activities must be carried out year by year in a certain month to form the month-of-the year effects. But why do these activities happen only on some certain periods every year? What is in the control of these behaviors behind? To solve all problems, we have to take a look at the nature of Chinese stock market

Ⅲ.The policy-driven Chinese stock market

The Chinese stock market can be characterized as a policy-driven market which means that the government has the overwhelming influence in the stock market. Since the initiation of Chinese two exchanges, SHSE and SZSE, Chinese government has created stock market institutions that allow the state to maintain control over listed companies, and over the whole market (Copper, 2003).

(15)

ⅰThe reasons of Chinese government’s intervention

There are three main reasons that Chinese government frequently intervene the stork and security market. Most importantly, the stability of security market is extremely important in maintaining the incumbent in power since the influence of Leninist political habits. Secondly, it’s a necessity of supporting state owned companies. Thirdly, it is to show the public a “successful” and “creditable” market with quick economic growth as an achievement of political top-leader in the government.

The Chinese government facing the biggest challenge for financial regulation is how to deal with the carry of heavy socialist tradition. In the past planned economy, national financial system is a centralized passive/inflexible instrument. State owned banks had to allocate financial resources into those sectors where China selected to support and develop according to political criteria instead of market demands. This is the main reason that the establishment of a capital market and securities trading in China was incurred intensive controversies in terms of political and ideological aspect.

(16)

In addition, owing to the deep influence of Leninist political and economic habits, economic growth and stock market stability is the dominant factor in maintaining the incumbent in power. From the Chinese Communist Party’s perspective, a “successful” and “credible” stock market is overwhelming concern. Thus, the stock market is strongly policy-driven by the Chinese government.

ⅱStock market institutions under Chinese policy-driven market

In the context of the political and economic stock market, the regulations on the stock markets not only are designed to function as supervision standards, most importantly, but also to execute the political functions of top leaders. Many reports provide the clues about the main mission of the policy missions of Chinese stock market. These policy missions are laid down by Chinese political top leaders in order to make sure their political power. The policy missions imposed on Chinese stock market has two main streams. One is supporting state-owned enterprises; the other one comes to ensuring quick economic growth (Cooper, 2003).

To carry out the policy missions and implement the political supervision on the stock market, the specific stock market and supervision institutions are designed by the Chinese Communist Party. Chart 1 (see below) gives the structure of these specific institutions. The creation of the Central Financial Work Committee is to realize the centralization of financial supervision and manipulation by political top leaders. As one of the most important institutions in financial stock market, the main tasks for Central Financial Work Committee are recruitment and control of leading personnel both in crucial regulatory organs and financial firms. With the help of this Committee, the political authorities in the top are capable of carrying out their interventions on the stock market.

(17)

administrative operations. The tasks of monitoring stock markets have been given from CSRC to the financial practitioner on the basis of Communist Party principal. The roles that CFWC and CSRC play are very important for putting the policy mission of Chinese government into practice.

Chart 1. The security market institutions

Chinese Communist Party

To maintain a successful and credible stock market with a quick economic growth

China Securities Regulatory Commission

To carry out the administrative monitoring

Central Financial Work Committee

To implement political market

Shenzhen Stock Exchange Shanghai Stock Exchange

The leading elites of Central Financial Work Committee, the China Securities Regulatory Commission, Shanghai Stock Exchange and Shenzhen Stock Exchange is a relatively small group. According to Heilmann’s research（2002）, the number of these core members is around 180. Most importantly, these core people meet on the occasion of high-level conferences. The constitution of this core group is very special. Heilmann believes that the majority people in this group have common roots in the formative period of Chinese stock market in the late eighties and early nineties which make them very loyal and trusty.

(18)

said that: “We need to use the visible and invisible hand simultaneously.”

ⅲ The forms of intervention

The Chinese government interventions on stock markets have two main streams. One is occasionally interferences by cooling down the overheated market. The other one is primarily involvement in popping up the slumped market. There are various forms of government interventions, consisting of talking up the market, disclosing selective information in the state media, introducing new rules which work as a market-supporting measure or market-slumping measure and secretly political directed buy and sell transaction.

The state-owned enterprises are one of the most important tools by Chinese government to realize the stock market manipulations. In terms of the shares in the stock market, a number of them have been controlled by Chinese government because the concentration of ownership by state. As of the end of July 2004, there are about 64% of Chinese shares are held by state-owned firms. Obviously, about two-thirds of shares outstanding are controlled by the state or state-related entities and only one third left are not state-owned. (Zhang et al, 2008). Due to the nature of Chinese political regime, these state-owned shares have special function as carrying out the social and political mission designed by the top leaders to support their political power. Because of the existence of these state-owned shares, the secretly political directed buy and sell transactions by the high-level authorities can be implemented. Thus, the special constitution of Chinese stock shares allows the state to interference the stock market.

(19)

on the People’s Daily (the first most important state newspaper) on 16th Dec, 1996. Just on the same day of the publication, the Shanghai Index decreased by 9.91% from 1110 point to 1000 point. The turnover is -86.35% times compared to the preceding day.

The cases of supporting the market when the dramatic declines are more common than restraining overheated market. One of the rescue operations was implemented in late 1992 since the Chinese stock market experience a large market declines from 1429 point to around 400 point within 5 months. The government managed to pop up the slumped market by talking up the market in the media and bringing into several market-supporting measures. With the help of the government interventions, the market rise to 1172 point. In August 1994, a series of interventions is carried out to boost the market from 325 to 1052 point, rising by 120%. In terms of Lin (2005), the government intervention to rescue the market is very frequently and every time the market increase blustering from the intervention.

ⅳIntensified administrative interference in March

(20)

way, the social stability is much more important in March than in other months.

By the nature of stock market and the setting of investors, Chinese stock market is not a barometer of national economy but barometer of social problems. The government intervention is applied by the Chinese government to solve various social problems due to the nature of the close relationship between politics and economy, the Chinese stock market is particularly prone to be a hot bed of raising all kinds of social unrests due to the resentment against political establishment. More importantly, the stock market is a trigger for the explosion of conflict between various interest groups. The majority of investors in Chinese stock market are individual with less knowledge and low income; hence any severe price fluctuation will disappoint these investors. Worse more, since the widely speculative activities are common in Chinese stock market, these investors with limited experience in speculation and less awareness of risks often gamble with their whole saving in the pursuit of high return. Following these structures and features of investors, a small ripple of price movement will lead to panic from a huge number of small investors. This kind of large-scale panics will easily result in social instability and even create political crisis during political sensitive periods. Therefore, the influence of social instability will have a magnifying effect in March.

Based on this background, the government intervention is a considerable factor in explaining the month-of-year effects in China. Especially in March, the government or top regulators make their effort on popping up the market which accordingly creates an increase in the market. Thus, it’s very interesting to test whether the month-of-year effects existing in Chinese stock market can be contributed to government interventions behavior.

Ⅳ.Data and descriptive statistics

(21)

used data in examining the month-of-the-year effects of China. Being the only two exchanges in China’s mainland, the stock exchange in Shanghai (SHSE) was established in Dec 19th, 1990, while the Shenzhen stock exchange (SZSE) was founded one year later, in June 3rd, 1991. China’s stock exchanges are quite young compared to others in developed countries like NASDAQ (founded in 1971), NYSE (founded in 1792).

In addition, unlike other stock indexes, both SHSE and SZSE are divided into two major aspects in terms of the categories of investors from different markets and the currency traded. In principal, companies both in A and B Shares are all listed in Chinese domestic markets. Nevertheless, the B Shares are traded in USD instead of RMB as A Shares does. Furthermore, A Shares are solely open to investors in the domestic markets while B Shares is open to the investors in foreign markets. Xu (2000) proclaims the different characteristics of B Shares from A Shares in four parts. As B Shares aims at foreigners only, it’s natural that foreign investors is much easy to obtain the latest information and updates on the B Shares than A Shares from media abroad. So B Shares is more open to the world and to the transmission of international price compared to A Shares. For the same reason, those foreign investors towards B Shares are provided with alternative financial products that are not available to domestic Chinese investors. Thirdly, since the more supervision from media abroad and more demands on information revelation by the foreigner investors, Chinese companies that are listed as B Shares are subject to more strict disclosure requirements and they are generally more financially stable than A Share companies. Finally the B Share market is much smaller and may also be less liquid. Because of all the differences mentioned above, the rules and control exposed by top regulators on A and B Shares vary largely. In a word, the intervention by Chinese government can only influence A Shares but hardly any influence on B Shares.

(22)

is no doubt that the A and B Shares classification is essential in testing the month-of-the-year effect in policy-driven market.

To carry out the robust test for my hypothesis, I supplemented the range of my data with Hang Seng China Enterprises Index (HSCEI) and those Chinese companies listed in NSDAQ (known as Concept Shares) in favour of the analysis. Hang Seng China Enterprises Index short for H shares is also employed in this paper. Similar to B Shares, H Shares is also traded in foreign currency and open to foreign investors only. As regards to Concept Shares, these shares obey the same regulations and standards as the normal shares listed in NSDAQ. The only difference is those shares are Chinese companies listed not American. The main cause for the differences between A, B, H and Concept Shares is the tight capital control exercised by the Chinese government authorities.

It’s easy to believe that the A Shares get more intervention and manipulation by the regulators than B Shares and H Shares because they are less open to the world and under the control of Chinese government. With mass inspection by others countries’ media and more strict disclosure requirement, B Shares and H Shares that are entirely open to foreign investors are hardly controlled by Chinese government authorities.

In this paper, all the data are collected from the Chinese Stock Markets and Accounting Research (CSMAR). Six market indices are collected in total: SHSE-A, SHSE-B, SZSE-A, SZSE-B, HSCEI and Concept Shares. The sample period is ten years from January 1 1999 to December 31 2008. The listed companies are given certain weights as regards to their percentage on the aggregate market values in the market. In others words, all these stock indexes are value-weighted.

(23)

(see below) gives a clear show on the descriptive statistics on March for five indexes. Statistically, the average monthly return in March varies between -0.02% and 0.48% for the various stock indexes. The standard deviation of March differs from 0.40% to 1.34%.

Table 2. Descriptive statistics for March in five indexes

SHSE-A SHSE_B SZSE-A SZSE-B HSCEI

Mean _0.22 _0.22 _0.25 _0.48 _-0.02 Median _0.40 _0.01 _0.43 _0.15 _0.04 Upper _0.62 _2.63 _0.51 _4.16 _0.65 Lower _-1.21 _-1.12 _-0.99 _-0.81 _-0.88 Std. Dev _0.53 _0.95 _0.46 _1.34 _0.40

All the values are in percentage form.

Moreover, the critical value acquired from t-table is approximate 2. Thus, the confidence intervals of monthly return are formed by plus or minus two times the standard deviation from the average monthly return. Judging by the confidence interval, the distribution of the average monthly returns in all cases includes the zero return. Accordingly, it’s no clear clue for the positive or negative effect. Further tests are supposed to give more useful information.

Both the Ordinary Least Squares model (OLS) and ARCH family models have been widely used to test the month-of-the-year effects. To figure out whether ARCH family model is fitting to be employed, the Jarque-Bera test is applied to examine the normality of the series of China’s stock market. Denoting the error terms by u and their variance byσ2, the JB test statistic is given by

] 24 ) 3 ( 6 [ 2 2 2 1 + − =T b b W (1) Where 2 / 3 2 3 1 ) ( ] [ σ u E b = and 2 2 4 ) ( ] [ σ u E b=

(24)

appropriate to use ARCH family model for the regression analysis.

The results of normality test are shown in Table 3 (see appendix). Except HSCEI, the p-value of all the indexes is statistically significant, which is a confirmation of the use of ARCH family model. It also can be concluded from those figures that all the mean value tends to be zero when it’s rounded up to four decimals. However, the mean of SZSE-B is negative and the rest of mean are positive.

The non-stationary issue is another problem that has to be considered when evaluating the month-of-the-effect. Due to the property of financial data, the time series are likely to have a unit root which implies that these series are non-stationary. The non-stationary series can be converted into stationary ones by differential approach. To check the unit root, the Augmented Dickey-Fuller (ADF) test is employed. The relevant figures are shown in Table 4 (see appendix). ADF test statistics show that all p-value are zero, which is significant without any exception. It concludes that these return series are stationary. Thus, no further bias adjustment is required.

Ⅴ. Methodology

Since the time-series are non-normal distributed and non-linear according to the tests beforehand, I use the ARCH family model for estimation. The ARCH family model exceeds OLS in many aspects when analyzing the financial series. According to Engle’s research (1982), the ARCH family models facilitate in capturing stylized facts of financial series such as fat tails, volatility clustering, etc. Not only the deficiencies of OLS corrected, but also a prediction can be computed for the variance of each error term by modeling the conditional variance as a measure of risk.

(25)

t t t x u y =β₁+β₂ ₂ + ~ (0, ) (2) 2 t t N u σ 2 1 1 0 2 − + = _t t α α u σ (3)

The ARCH model effectively model the ‘volatility clustering’ by setting the conditional variance of the error term to be dependent on the previous term of square error . In the conditional mean equation (2),

2 t σ 2 1 − t u 1 β is the intercept; β₂ represents the coefficient of independent variable . In the conditional variance equation (3),

t

x2

0

α is the intercept while α₁ is the coefficient of square error .

2 1 −

t u

To make up the nonnegativity constraints of ARCH model, the GARCH model is developed independently by Bollerslev (1986) and Taylor (1986). GARCH model extends the conditional variance equation (3) to:

2 1 2 1 1 0 2 − − + + = t t t α α u βσ σ ₍₄₎

Compared to ARCH model in equation (3), GARCH model allows the conditional variance to be dependent on its own previous lags .

2 t σ 2 1 − t σ

The problems existing in analyzing month-of-the-year effect by GARCH family models are mainly concerned with the lack of considering the conditional variance or the conditional standard deviation

2

t

σ

t

σ _{in the conditional mean equation. Obviously,} the return is related to the risk and should be determined by conditional variance or the square root. Because GARCH-in-Mean (GARCH-M) take both risk and return into consideration, it exceeds the typical GARCH model by getting involved the concept that the investors should be rewarded with extra returns for bearing additional risk (known as risk premium,δ ).

(26)

month in a year, t it i t t

x

u

R

=

μ

+

δσ

₋1

+

φ

+

₍₅₎

Where Rt represents the first log difference of stock price indices. 1

ln − = t t t P P R 1 − t

σ _{is the square root of conditional variance,} _{is a vector of monthly dummy} variables. From January through November, =1 for the i-th month and =0 otherwise. For December, the intercept

it

x

it

x x_it

μ_{suggests the December dummy.} _{is the} error term. When there is an existing for i-th month effect, the p-value of the i-th month is supposed to be statistically significant. Therefore I expect the March effect is likely to be found in the light of the previous research on month-of-the-year effect in Chinese stock market plus the feature of China’s policy-driven market with strong political intervention.

t

u

Ⅵ. Empirical Results

This part contains all the results on the main concerns in this paper. Before the main test, the normality test is employed to confirm that if GARCH-in-Mean is suitable for the data. What’s more, after the main tests, a robust test is carried out to consolidate the argument issued in the discussion.

ⅰResults of the main tests

The results of the main regression are on Table 5 (see below). There are ARCH effects in both indexes because the p-value for both α1_{is significant statistically. Namely,}

(27)

the increase in risk will not result in higher return accordingly.

For my main concern, the coefficient of the monthly dummies from January to November and the intercept representing the December dummy suggest the financial anomalies results. From Table 5, we can tell the coefficient of March is very significant at a 5% level for SHSE-A. But it’s not the case for SHSE-B: the p-value of January, February, May, July, September and December is all significant statistically on the same 5% significant level. In SHSE-B shares, no monthly return stands out the rest of other months specifically in a year as there are six months namely half year reveals that significant figures. It makes no sense to analyze and attribute these excessive returns in six different months as it’s hardly to indicate the dominant reasons for this phenomenon. Thus, here I consider there are no month-of-the-year effects in SHSE-B.

Table 5. GARCH-in-Mean test for SHSE: 1998(1)-2008(12)

(28)

The results for another exchange SZSE is shown in Table 6 (see below). Unlike other published researches, there is no January effect or Chinese New Year effect found in this research. Those figures tell that the A shares in SZSE shows a March effect since the March dummies are significant whereas there is no month-of-the-year effect displayed in SZSE-B index. As the same to SHSE, the January effect is also absent in Shenzhen Exchange. It’s also founded that the A shares in Shenzhen Exchange shows no ARCH effect while B shares has ARCH effect by judging from α1_{. Contrary to}

SHSE, the coefficient for δ is all negative meaning that there is no relationship between the return and risk.

Table 6. GARCH-in-Mean test for SZSE: 1998(1)-2008(12)

Coefficient Std. E P-value Coefficient Std. E P-value SZSE-A SZSE-B Jan 0.0023 0.0014 0.0873 Jan 0.0017 0.0018 0.3552 Feb 0.0019 0.0013 0.1576 Feb 0.0014 0.0030 0.6497 March 0.0042 0.0020 0.0352** March 0.0011 0.0025 0.6506 April 0.0010 0.0013 0.4678 April 0.0011 0.0020 0.5930 May 0.0021 0.0012 0.0893 May 0.0010 0.0019 0.5910 June 0.0015 0.0013 0.2455 June -0.0017 0.0020 0.3983 July 0.0008 0.0013 0.5663 July -0.0005 0.0020 0.8088 Aug 0.0013 0.0019 0.4999 Aug -0.0016 0.0028 0.5758 Sept -0.0001 0.0016 0.9597 Sept -0.0006 0.0020 0.7743 Oct -0.0007 0.0018 0.7079 Oct -0.0016 0.0018 0.3648 Nov 0.0010 0.0018 0.5682 Nov 0.0007 0.0023 0.7579 Dec 0.0015 0.0017 0.3704 Dec 0.0024 0.0018 0.1709 δ -0.7608 0.4141 0.0661 δ -0.5768 0.2293 0.0119 0 α 0 0 0.8693 α0 _{0 0}_0.3413 1 α 0.0724 0.0505 0.1516 α1 _{-0.0298 0.0007 0**} β 0.9427 0.0852 0 β 1.0361 0.0005 0 ** Significant at a 5% level

(29)

the previous researches on Chinese stock markets, the January effect or Chinese Lunar New Year effect is absent from the results. However, we can tell that both A Shares in two exchanges have March effect, while no month-of-the-year effects in both B Shares in two exchanges.

As discussed before, March is the intensified political high season as explained before. To get rid of social chaos, the Chinese government makes their efforts to make the stock market stable and successful at least no large slump in the market. During this sensitive season, disclosure the selective information on the state media or other channels, Chinese government always succeeds in creating a happy atmosphere during March. Therefore, the influence of the political force by the Chinese government can be the dominant reason of this March effect.

Most importantly, the key differences between A Shares and B Shares come to the magnitude of Chinese government interventions. A Shares is totally under the control of Chinese government, but B Shares is not. Since the March effect is solely present in A Shares but not in B Shares, we have the reasons to argue that this March effect is caused by Chinese government interventions. More specifically, it’s argued here that the government interventions can explain the March effect in China.

Table 7. Results of the main tests

Index Outcome of month-of-the-year effect

SHSE-A _{March effect}

SHSE-B No clues for specific month-of-the-year effect

SZSE-A March effect

SZSE-B No month-of-the-year effect

ⅱ

Results of the robust tests

(30)

observation period for Concept Shares is from Jan, 2000 to Dec, 2008.

However, there is no existing composite index for the Concept Shares in the CSMAR DataStream. Therefore, I compute the index for those companies by the method in which the index is derived.

Index =

N

y

x

N i i i

∑

=1 (7)

Where denotes the latest price of stock i, stands for the market capitalization of stock i, N represents the overall market capitalization of market,

is the weight of stock i in the whole index. By using this method to compute, the index for Concept Shares is value-weighted which is consistent with other five indexes mentioned in this paper.

i

x yi

i iy

x

Table 8. GARCH-in-Mean test for HSCEI: 1998(1)-2008(12)

(31)

As explained in Part Ⅲ,neither H shares (short for HSCEI) nor Concept Shares are listed in the mainland of China. Accordingly, H Shares and Concept Shares follow others regulations of exchanges which are not laid down by Chinese government. So obviously, unlike SHSE and SZSE, Chinese government interventions are not capable to influence these shares. Based on this fact and the finding from the main tests, the regressions on both H Shares and Concept Shares are supposed to have no March effects.

Table 9. GARCH-in-Mean test for Concept Shares: 2000(1)-2008(12)

Coefficient Std. Error P-value

Concept Shares Jan -0.0270 0.0896 0.7632 Feb -0.1034 0.1258 0.4113 March -0.0222 0.0958 0.8171 April -0.0165 0.0951 0.8622 May 0.0251 0.0994 0.8008 June -0.0289 0.1053 0.7838 July -0.1006 0.1021 0.3246 Aug -0.0554 0.0991 0.5765 Sept -0.0125 0.0935 0.8936 Oct 0.0155 0.0946 0.8701 Nov 0.0013 0.0931 0.9892 Dec 0.1052 0.1071 0.3263 δ -0.5751 0.3766 0.1268 0 α 0.0011 0.0013 0.3950 1 α 0.1343 0.1021 0.1885 β 0.8101 0.1157 0

(32)

The results of the robust tests on HSCEI and Concept Shares are shown in Table 8 and Table 9 (see above). From Table 8 and 9, we can conclude that there are no specific monthly dummies statistically significant. These results are consistent with my assumption and also with the finding from the main tests. Consequently, the robust tests support the result that there is only March effect in China stock market, which is acquired from the main tests.

Ⅶ Conclusion

Month-of-the-year effects in developed market have been widely researched and various explanations have been presented. When it comes to the emerging markets, the result of month-of-the-year effects and the cause of them can be very different. This paper argues that the influence of political power can be one explanation of month-of-the-year effects in China’s strongly policy-driven economy,. Therefore, the window dressing theory is also relevant to Chinese cases.

More specifically, it’s politically sensitive period in March because the social stability is overwhelming concerned by the China state. The government tends to increase the stock prices to window dress the security market performance. These interferences succeed in making people feel good to prevent possible social restless and chaos. The main forms of these window dressing actions comprise talking up the market, disclosure selective information in state media and direct purchase or sell via state-owned shares. In this way, the return in March is likely to exceed that of other months.

(33)

effect is only present in A Shares but absent in B Shares. Because of the fact that only A Shares are under the close eyes of Chinese government while others shares are not, this paper argues that this March effect detected is attributed to the Chinese interventions. In other words, government interventions can be the cause of month-of-the-year effects in China. To confirm the finding, a robust check using H Shares and Concept Shares are conducted. These shares are listed in exchanges not in the control of Chinese government. Accordingly, if the March effects are caused by the Chinese interventions, there is supposed to have March effects neither in H nor Concept Shares. As it’s expected, no effect exists in this index.

(34)

References

Alagidede, P. and Panagiotidis, T. (2006) Calendar anomalies in an emerging African market: evidence from the Ghana stock exchange, Journal of Emerging Market Finance, 8, 13-23 Al-Khazali, O (2003) Distributional characteristics and stochastic dominance versus parametric

analysis in testing anomalies in the emerging market: Evidence from Jordan, Journal of Business & Economics Research, 1, 765-777

Barry N. (2002) The politics of the stock market, China Leadership Monitor, No.3

Bepari, K. and Mollik, A.T. (2009) Seasonalities in the monthly stock returns: Evidence from Bangladesh Dhaka Stock Exchange (DSE), International Research Journal of Finance and Economics, 24, 112-120

Berges, A., J. McConnell and Schlarbaum, G. (1984) The turn-of-the-year in Canada, Journal of Finance, 39, 185-192

Brown, P., D. B. Keim, A.W. Kleidon and T. A. Marsh (1983) Stock return seasonalities and the “tax-loss selling” hypothesis: Analysis of the arguments and Australian evidence, Journal of Financial Economics, 12, 105-127

Chen, G., Kwok, C. and Rui, O. (2001) The day-of-the-week regularity in the stock markets of China, Journal of Multinational Financial Management, 11, 139-63

Clare, A. and Thomas, S. (1995) The overreaction hypothesis and the UK stockmarket, Journal of Business Finance & Accounting, 22, 961-973

Cooper, M.C. (2003) The politics of China’s shareholding system, Shorenstein APARC, June 2003 Doran, J., Jiang, D. and Peterson, D. (2008) Gambling in the New Year? The January idiosyncratic

volatility puzzle, MPRA Paper 8165, University Library of Munich, Germany

Fabozzi, F. J., Tunary, R. and Wu, T. (2004) Modeling volatility for the Chinese equity markets, Annals of economics and finance, 5, 79-92

Fan, X. and Dong W. (2007) The month-of-the-year effects in Chinese stock market: Based on the consumption habits, Journal of Sun Yatsen University, No.3

Fountas S. and Segredakis K. N. (2002) Emerging stock markets return seasonalities: The January effect and the tax-loss selling hypothesis, Applied Financial Economics, 12, 291-299

Gao L. and Gerhard K. (2005) Calendar effects in Chinese stock market, Annals of Economics and Finance, 6, 75-88

Harjeet, S. B., Upinder S. D. and Gabriel G. R. (1999) A November effect? Revisiting the tax-loss-selling hypothesis-statistical data included, Financial Management, December 22, 1999

Jiang, Z., Xie C. and Zhang Z. (2008) The month-of-the-year effect in Chinese stock markets, Financial Theory and Practice, No.8 (in Chinese)

Josef, L., Andrei S. and Robert V. (1991) Window dressing by pension fund managers, AEA Papers and Proceedings, May 1991

Keim, D. B. (1983) Size-related anomalies and stock return seasonality, Journal of Financial Economics, 12, 13-32

Keim, D. B. (1989) Trading patterns, bid-ask spreads, and estimated security returns: The case of common stocks at calendar turning points, Journal of Financial Economics, 25, 75-97

Li, R. (2003) The month-of-the-year effect in the stock markets of China, Quantitative & Technical Economics Research, No. 7 (in Chinese)

(35)

Liu, F., Sun, P. and Chen, Q. (2003) Closing stock prices and their window show effect: Empirical study based on stocks traded in Shanghai, Securities Market Herald, No.10, 2003 (in Chinese)

Liu, F. and Chen, J.(2004) The windows dressing explanation of turn-of-the-month effect: An empirical research on Shanghai stock market, Quantitative and Technical Economics, No.21, 2004

Miller, E. M. (1990) Explaining the January small firm effect by the interaction of procedurally rational investors and seasonal traders, Quarterly Journal of Business and Economics, 29, 36-54

Mitchell, J.D. and Li L.O. (2006) Seasonalities in China’s stock markets: cultural or structural, IMF Working Paper, WP/06/4

Mookerjee, R. and Yu, Q. (1999) Seasonality in returns on the Chinese stock markets: The case of Shanghai and Shenzhen, Global Finance Journal, 10, 93-105

Rozeff, M. and Kinney, W. (1976) Capital market seasonality: The case of stock returns, Journal of financial economics, 3, 379-402

Poterba, J. M. and Weisbenner, S. J. (2001) Capital gains tax rules, tax-loss trading and the turn-of-the-year returns, Journal of Finance, 56, 353-68

Sebastian, H. (2002) The Chinese stock market: The pitfalls of a policy-driven market, China Analysis, No.15

Raj, M. and Thurston, D. (1994) January or April? Tests of the turn-of-the-year effect in the New Zealand stock market, Applied Economics Letter, 1, 81-83

Wei Y. G. (2001) The statistical analysis of policy intervention on Shanghai stock market, Statistical Research, No.2

Xu, C. K. (2000) The microstructure of the Chinese stock market, China Economic Review, 11, 79-97

Yartey, C. A. (2008) The determinants of stock market development in emerging economics: Is South Africa different? IMF Working Paper, No. 08/32

(36)

Appendix

Table 1.Descriptive statistics for all the five indexes

Jan Feb March April May June July Aug Sept Oct Nov Dec

SHSE-A Mean 0.01 0.14 0.22 -0.01 0.07 -0.01 -0.04 0.02 -0.03 -0.12 -0.01 -0.02 Median 0.11 0.10 0.40 0.00 -0.01 -0.02 0.03 0.08 -0.10 -0.15 0.06 -0.22 Upper 0.29 0.95 0.62 0.46 0.69 1.31 0.47 0.55 0.54 0.30 0.46 1.20 Lower -0.78 -0.22 -1.21 -0.71 -0.58 -1.09 -0.79 -0.42 -0.45 -0.78 -0.94 -0.51 Std. Dev 0.32 0.31 0.53 0.40 0.41 0.62 0.37 0.29 0.30 0.32 0.39 0.53 SHSE-B Mean 0.16 0.1% 0.22 0.15 0.42 -0.13 -0.10 -0.15 -0.01 -0.19 0.05 -0.01 Median 0.25 0.04 0.01 0.04 0.00 -0.17 0.04 -0.09 -0.25 -0.15 0.09 0.03 Upper 1.38 0.44 2.63 1.13 1.80 1.94 1.08 0.48 0.82 0.70 0.91 0.52 Lower -0.88 -0.22 -1.12 -0.33 -0.38 -0.85 -1.41 -1.27 -0.39 -1.32 -0.54 -0.47 Std. Dev 0.72 0.22 0.95 0.47 0.84 0.81 0.75 0.49 0.40 0.56 0.43 0.37 SZSE-A Mean 0.13 0.15 0.25 0.07 0.11 -0.13 0.04 0.00 -0.09 -0.20 0.02 -0.03 Median 0.19 0.15 0.43 -0.08 0.02 0.00 0.01 0.06 -0.12 -0.21 0.09 -0.21 Upper 0.86 0.61 0.51 1.12 0.65 0.68 0.79 0.58 0.26 0.29 0.68 0.85 Lower -0.41 -0.29 -0.99 -0.42 -0.51 -1.30 -0.81 -0.82 -0.48 -0.70 -0.80 -0.47 Std. Dev 0.38 0.28 0.46 0.43 0.39 0.55 0.43 0.39 0.25 0.26 0.40 0.45 SZSE-B Mean 0.08 0.08 0.48 0.10 0.27 0.05 -0.17 -0.17 -0.12 -0.13 0.05 -0.01 Median 0.31 0.13 0.15 0.13 0.11 -0.12 -0.04 -0.19 -0.17 -0.17 0.03 0.00 Upper 0.92 0.29 4.16 0.72 1.31 2.31 0.61 0.34 0.51 0.63 0.58 0.74 Lower -0.88 -0.14 -0.81 -0.50 -0.58 -0.71 -1.49 -0.52 -0.72 -1.00 -0.72 -0.46 Std. Dev 0.54 0.14 1.34 0.41 0.56 0.86 0.64 0.29 0.42 0.52 0.35 0.38 HSCEI Mean -0.15 0.13 -0.02 0.21 0.13 0.28 0.09 0.02 -0.13 -0.02 0.00 0.08 Median -0.16 0.24 0.04 0.01 0.17 0.22 0.17 -0.02 -0.21 -0.16 0.18 -0.05 Upper 0.61 0.71 0.65 1.22 0.73 1.83 0.87 0.57 0.66 0.78 0.43 0.95 Lower -1.17 -0.99 -0.88 -0.63 -0.29 -0.75 -0.85 -0.53 -0.91 -0.92 -0.80 -0.36 Std. Dev 0.58 0.48 0.40 0.54 0.30 0.67 0.54 0.32 0.47 0.53 0.43 0.49

(37)

(38)

C. Table 4. ADF test for all five indexes

Variable Coefficient SE t-value p-value Variable Coefficient SE t-value p-value

ADF test for SHSE-A ADF test for SHSE-B

SHSE-A(-1) -0.7279 0.1244 -5.8488 0 SHSE-B(-1) -0.7277 0.1214 -5.9925 0

Constant 0.0001 0.0004 0.3483 0.7283 Constant 0.0004 0.0006 0.6686 0.5051 ADF test statistic ___ ___ -5.8488 0** ADF test statistic ___ ___ -5.9925 0**

1% level -3.4866 1% level -3.4866 5% level -2.8861 5% level -2.8861 Test critical values

10% level -2.5799

Test critical values

10% level -2.5799

ADF test for SZSE-A ADF test for SZSE-B

SZSE-A(-1) -0.6242 0.1186 -5.2642 0 SZSE-B(-1) -0.8537 0.0912 -9.3572 0

Constant 0.0002 0.0004 0.4903 0.6249 Constant 0.0005 0.0006 0.9130 0.3631 ADF test statistic ___ ___ -4.2642 0** ADF test statistic ___ ___ -9.3572 0**

1% level -3.4866 1% level -3.4861 5% level -2.8861 5% level -2.8859 Test critical values

10% level -2.5799

Test critical values

10% level -2.5798

ADF test for HSCEI

HSCEI-A(-1) -0.9949 0.0908 -10.9557 0

Constant 0.0006 0.0004 1.3486 0.1801 ADF test statistic ___ ___ -10.9557 0**

1% level -3.4861 5% level -2.8859 Test critical values

10% level -2.5798

Are the month-of-the-year effects in China caused by government interventions?