Weak Form Efficient Market Hypothesis For Emerging Markets
The issue of market efficiency in emerging markets is of great significance for both foreign investors and policy makers in emerging economies. This project devotes large efforts to produce a thorough and in-depth literature review for this area. This topic is to be investigated from these aspects: theoretical foundation, methodologies of tests and empirical results. Firstly, traditional efficient market hypothesis (Fama, 1970; Makiel, 1973) and behavior finance theories developed in recent decades (Barbris, 1998; Shleifer, 2000) have formed two main schools of thought for the issue of market efficiency. Secondly, the evolution for a series of methodologies is important for testing market efficiency. Thirdly, the empirical evidence is reviewed by consideration three major factors: trade volume and non-linear behavior, structural breaks and market evolution through time. Finally, it also reflects some important policy implications for emerging markets.
Many empirical studies have been widely carried to investigate the weak-form efficient market hypothesis for emerging markets, and the results are mixed. Generally, most of emerging markets are found to be inefficient. But for some countries, such as Istanbul, Egypt and Jordan, after correcting for institutional characteristics and trading conditions, such as thin trading and the presence of non-linearity, equity markets are found to be efficient. When structure break factors are taken into account, market efficiency is powerfully rejected for countries such as Argentina, Brazil, Greece and India. There is also evidence showing that initially emerging markets are inefficient, but over time they are moving toward to be more efficient, such as in Estonian, Lithuanian and Russia duo to economic liberalization policies.
These results reflect some important policy implications. Infrequent trading and illiquidity of capital markets negatively affects market efficiency, so economic policy makers should devote efforts to minimize the institutional restriction and barriers on capital flow in the financial markets and to impose strict disclosure requirements, so that investors can easily access to high quality and reliable information. Improving liquidity of capital markets can provide lower borrowing costs for investors and greater opportunities for investment diversification with lower systematic risks. In addition, equity market liberalization is important to help achieving market development. It can reduce cost of capital and increase capital productivity with better capital allocation.
Due to the increasing globalization of financial markets, fast economic growth and adoption of financial liberalization policies for equity markets in emerging economies, it is widely indicated that equity investment in emerging economies can provide superior returns. Past decades have witnessed spectacular growth in both size and relative importance of emerging equity markets. The market capitalization of emerging market economies accounts for twelve percent of world market capitalization and has more than doubled, growing from less than $2 trillion in 1995 to $5 trillion in 2006 (Nally, 2010). By 2015, it is estimated that the combined GDP of emerging-market economies will surpass that of the top 20 developed economies (ibid). In addition, emerging market returns are weakly correlated with returns in developed markets, so international diversification with these emerging equities can give lower portfolio risks (Levy & Sarnat, 1970). The potential high rates of returns and diversification benefits has attracted large number of foreign fund investors, so the investigation on whether emerging markets function efficiently is significantly important. By knowing degree of market efficiency, economy policy makers and regulators can gain insights to develop right institutional and regulatory frameworks to allocate scare resources efficiently, form favourable investment condition and obtain further economic growth.
Therefore, this essay is going to investigate the weak-form market efficiency in emerging markets. The efficient market hypothesis by Fama (1970), Random Walk module by Makiel (1973) and behaviour finance theories are directed related to this issue and form the theoretical foundations. Section 1 will critically give the theoretical review based on the two schools of thought that are EMH and behaviour finance theories. Section 2 will give a brief review of methodologies adopted in literature review. Section 3 will give empirical review of the weak-form EMH for emerging markets. Section 4 will indicate some brief policy implications for emerging economies and section 5 is the conclusion with some directions for further research.
Theoretic Review of EMH VS Behaviour Finance
Efficient Market Hypothesis
Fama (1970) defines an efficient financial market as one in which security prices always instantaneously and fully reflect all available information. No investors can earn expected abnormal return by analysing past known information. Market efficiency is attained by two key forces: investor rationality and arbitrage activities (Fama, 1970). EMH assumes that investors are rational and can process information correctly and efficiently. Although some investors are irrational and may overact or underact to new information, these judgement errors are independent and random, hence can cancel out each other without affecting prices (Fama, 1998). Therefore, on average, the whole market is efficient. In addition, since numerous profit-maximizing investors are competing to analyse, value and trade securities based on all available information to exploit arbitrage opportunities, on aggregate level, security prices are adjusted quickly to reflect the effect of new information (Fama, 1998). Security prices are driven close to intrinsic values. Expected returns implicit in the current price of a security should reflect its underlying risk, and higher returns are earned only as compensations for bearing higher risk.
There are two main modules that explain EMH: fair-game model and random walk model (RWM). The fair-game model is expressed as: Zj,t+1 =rj,t+1-E(rj,t+1｜Фt), E(Zj,t+1｜Фt)=0 (Copeland, Weston & Shastri, 2005). Information in Фt is fully utilized to determine equilibrium expected returns. On average, the expected return on an asset E(rj,t+1｜Фt) equals its actual return (rj,t+1), so that no expected abnormal return can be gained from past information. RWM gives much stronger condition for EMH. It assumes that successive price changes have a same normal distribution and are independent. Its logic is that because new information is unpredictable and reaches market randomly, so under EMH, the resulting security price changes must be also unpredictable and random (Malkiel, 1973; Malkiel, 2003). No profit can be made from past information. There are three sub-hypotheses of EMH depending on the level of available information set (Fama, 1991). Firstly, market is weak-form efficient when prices reflect all security market information such as historical prices. Secondly, market is semistrong-form efficient when prices reflect all public information such as corporate news and financial statements. Thirdly, market is strong-form efficient when prices reflect all public and private information.
Behaviour Finance Theories
Figure1: Conceptual Framework of Behaviour Finance
Source: Shleifer (2000)
However, behaviour finance challenges EMH because it argues that psychological biases lead to investor irrationality and limits to arbitrage impede exploitation of mispricing opportunities (Shleifer, 2002). Psychological bias results into systematic overreaction or underreacion among investors. Many behaviour finance theories have been successfully developed to explain some market anomalies. Conservatism biases lead people adjust slowly to new information and hence the underreaction to new information leads to short-run momentum, while representativeness heuristic makes investors believe that past good stock performance will continue and people overreact to information (Barberis, Shleifer & Vishny, 1998). Additionally, overconfidence causes investors to overestimate the precision of their own analyses and to neglect public signals (Daniel, Hirshleifer & Subrahmanyam, 1998). Under positive (negative) private signal (which is shown in following graph), informed investors overreact and security is overpriced (underpriced). When public information becomes available, biased self-attribution causes security to be even more overpriced (underpriced). Eventually, public information proves initial investment judgement is wrong, so price is driven back to intrinsic value (Daniel et al, 1998). It explains that overconfidence leads to short-run return momentum and price correction leads to long-run return reversal.
Figure 2: Overconfidence and Self-attributed bias
Source: Daniel, Hirshleifer & Subrahmanyam (1998)
Moreover, classification is a human natural instinct to process information (Barberis & Shleifer, 2003). Investors naturally classify stocks by styles, so styles returns are highly positive correlated. There are two kinds of investors: style switchers and fundamental traders. Style switchers are unsophisticated investors and chase investment styles based on past relative stock performance. When there is good news about stock X (shown following graph), they will drain funds away from less attractive style Y. It will push up stock X’s price, even higher than its intrinsic value, but further reduce stock Y’s price. However, fundamental traders recognize stock Y is underpriced (Barberis et al, 2003). They arbitrage away mispricing opportunities and drive overpriced stocks back toward intrinsic value.
Figure 3: Switchers and Fundamental Traders
Source: Barberis & Shleifer, (2003)
On the other hand, limits to arbitrage may obstruct information to be impounded into prices, duo to the fundamental risk and implementation costs. Noise trader risk would prevent rational investors from arbitraging (Delong, Summer & Waldmann, 1990). Pessimistic noise trader drive price below intrinsic value, arbitrageurs can buy the asset, but bear risk of further deviation from the intrinsic value when noise traders become even more pessimistic and price goes down even further (Delong et al, 1990). Arbitrageurs usually have short horizon and must liquidate before price recovers, so they will incur loss. The agency problems between professionals and investors also affect arbitrage (Shleifer & Vishny, 1997), so not all mispricing would be arbitraged away to lead market become efficient.
However, Fama (1998) argues that behaviour finance theories do well only on the anomalies they are specially designed to explain and cannot be generalized to the entire market. Rubinstein (2001) also argues that investor overconfidence would make market “hyper-rational”.
Methodologies Adopted to Test the Weak-form EMH
Empirical researches on testing weak-form EMH can be divided into three broad categories. Firstly, they tests security return independence. If time-series pattern of security returns shows insignificant (significant) autocorrelations, then weak-form EMH holds (is rejected) (Copeland, Weston & Shastri, 2005). Secondly, they test return momentum effect. If portfolio of stocks with higher returns in the short past continues to earn higher abnormal returns in the subsequent short term, then short-run past returns contain information that could predict future returns, so EMH will not hold (Copeland et al, 2005). Thirdly, they test technical trading rules. If no trading rules that consistently derive abnormal profits can be found, then weak-form EMH holds.
A series of research methodologies have been developed to exam the EMH. The runs test is non-parametric, which is used to determine whether successive prices changes are independent. Unit root tests involve three different methods to test the null hypothesis of a unit root: the Augmented Dickey-Fuller (ADF) test (1979), the Phillips-Peron (PP) test (1988) and the Kwiatkowski, Phillioh, Achmidt and Shin (KPSS) test (1992). Multiple variance ratio (MVR) tests are adopted to detect autocorrelation and heteroskedasticity in returns (Chow & Denning, 1993).
Empirical Results of Weak-form EMH for Emerging Markets
The research results for testing weak-form efficiency on the emerging markets are mixed. World Bank study reports significant market inefficiency for 19 emerging equity markets (Claessens, Dasgupta & Glen, 1995). Latin American emerging markets of Argentina, Brazil, Chile, and Mexico are weak-form EMH (Urrutia, 1995), but under the variance ratio test, RWH is rejected (Ojah & Karemera, 1999). Under ADF test, EMH is also generally supported for six Latin American stock markets (Choundhry, 1997). For the emerging markets in Asia, major Asian markets are weak-form inefficient, such as Korea and Taiwan (Cheung, Wong & Ho, 1993), Singpore and Thiland (Huang, 1995), but some find it is efficient for Hong Kong, Singapore and Japan (Chan, Gup & Pan, 1992). When the observed index levels are used, both RWH and EMH are rejected for three equity markets of Saudi Arabia, Kuwait, and Bahrain after adjusting for infrequent trading , but when the corrected true indices are used, RWH is accepted (Abraham et al, 2002). RWH is rejected in five Middle Eastern emerging markets, Jordan, Morocco, Egypt, Israel, and Turkey (Omran and Farrar, 2001). Weak-form efficiency is rejected for Saudi and Palestinian financial market and inefficiency might be due to delay in operations and high transaction cost, thinness of trading and illiquidity in the market (Nourredine & Kababa, 1998; Award & Daraghma, 2009). Many researches find that emerging markets are becoming more efficient due to the liberalization policies. Istanbul stock exchange was inefficient in the early times but it becomes more efficient as the country started liberalization and deregulation (Antonios, Ergul & Holmes, 1997).
4.1 Thin Trading and Non-linearity
It is argued that such mixed evidences of the weak-form EMH in emerging markets are only reliable if the methodologies adopted take accounts for the institutional characteristics and trading conditions of the markets, such as thin trading and the presence of non-linearity (Antoniou, Ergul & Holmes, 1997). Ignoring these factors may lead to statistical illusions regarding efficiency. The conventional tests of efficiency based on linear model have been developed to test markets with high levels of liquidity, sophisticated investors with access to reliable information and few institutional impediments (Antoniou, Ergul & Holmes, 1997). Therefore they are not suitable for testing EMH for emerging markets with characteristics of thin trading, low liquidity and less well informed investors with access to unreliable information. Thin trading will bring serious serial correlation (Fisher, 1996), so the observed dependence does not necessarily represent serial correlation among securities returns.
In addition, prices responds to information in a non-linear behavior especially during the early development stages of emerging markets (Schatzberg & Reiber, 1992), so if the return generating process is non-linear but a linear model is used to test efficiency, then EMH may be wrongly accepted. This is because non-linear systems such as “chaotic” ones look very similar to a random walk (Savit, 1988). However, the conventional tests cannot recognize this problem. There are several reasons for the existence of non-linear reaction of price to information in emerging markets. Transaction costs are high, information is relatively not reliable and market is illiquid or there are restrictions on trading (Stoll & Whaley, 1990). As a result, investors do not always respond instantaneously to the information, which contradicts the assumptions of investor rationality and linear response of price. Scheinkman and LeBaron (1989) and Peters (1991) also empirically support the non-linearity of stock returns.
A number of studies have researched the impact of thin trading (Fisher, 1966; Dimson, 1979; Cohen, 1978; Lo & Mackinlay, 1990). Many empirical studies also have taken account of the non-linearity in price series and remove the impact of thin trading by the AR (1) model proposed by Miller (1994). Antoniou, Ergul and Holmes (1997) find that there is apparent predictability of stock returns for Istanbul stock market, but after considering the impact of thin trading, the random walk hypothesis is accepted and the market is informationally efficient for 1990 onwards. Abuzarour (2005) examines the effect of non-trading on market efficiency for three emerging Arabian equity markets: Jordan, Egypt and Palestine using the variance ratio test and the run test during the period of 1992 and 2004. Both random walk hypothesis and weak form efficiency are rejected when the observed index levels are used. However, when the indices are corrected by the Miller, Muthuswamy and Whaley methodologies (1994) to take account for thin trading, weak-form EMH is accepted for Egypt and Jordan stock market but it is still rejected for Palestine. All these empirical researches suggest that markets become more efficient when trading volume is high, information is much reliable and institutional frameworks are appropriate.
4.2 Structural Breaks
Research on efficiency for emerging markets should not only take account for institutional characteristics and trading conditions, but also should take account for the structural breaks in the underlying series that arise from the liberalization. Ignoring structural breaks can lead to wrong inference that these indices are following random walks. Many emerging countries are liberalizing their financial markets with various degrees (IFC, 1997) and such structure changes would have affected their equity markets (Bekaert et al, 2002; Henry 2000). For instance, huge shocks occurred for equity index level for Greece, Malaysia and Philippines in late 1980s and early 1990s, which are around the same year of their market liberalization. As Perron (1989) have demonstrated that traditional standard tests for RWH in stock prices have low power against the alternative hypothesis in small samples, and the problem is especially serious when structural changes are involved. Thus failure to consider these breaking points may wrongly support the RWH.
Therefore, many empirical researches try to incorporate the structural breaks factor by more powerful test methods, such as the Zivot- Andrew sequential test (Zivot & Andrew, 1992). Chaudhuri and Wu (2001) adopt both the standard ADF test and Zivot- Andrew sequential method to test the EMH in seventeen emerging markets: Argentina, Brazil, Chile, Colombia, Greece, India, Jordan, Korea, Malaysia, Mexico, Nigeria, Pakistan, Philippines, Taiwan, Thailand, Venezuela, and Zimbabwe. Results for the ADF test without breaks to each series tend to show non-rejection of the RWH. However, results for the Zivot- Andrew test with structural breaks show that RWH can be powerfully rejected at the one percent significant level in ten markets: Argentina, Brazil, Greece, India, Malaysia, Mexico, Nigeria, Philippines, Taiwan and Zimbabwe (Chaudhuri & Wu, 2001).
4.3 Market Evolution
Although structural breaks have been taken into account in many researches, it is argued that standard techniques are still not fit to test the weak-form EHM for emerging market, because they are not able to evaluate the evolving efficiency in emerging markets. It is also argued that methods such as a time varying parameter model and Kalman Filter technique not only can indicate the movement of stock returns from inefficiency to efficiency, but also can measure the timing of the movement towards full efficiency(Rockinger & Urga, 2000; Zalewska-Mitura & Hall, 1999). It is generally agreed that emerging markets are evolving from inefficiency to efficiency with the higher disclosure degree of firm practices, high trading volume and lower institutional barriers to trade (Cornelius, 1994). According to Laurence (1986), the methods of OLS or GMM test market efficiency over the whole period and hardly capture the tendency towards efficiency, so under these methods, early inefficiency would wrongly lead to the conclusion that there are profit opportunities based on the past asset price movement. In addition, the variance of the error process in the conventional test models is not constant over time, so if this changing variance structure is omitted and has a serial correlation property, then market efficiency would be incorrectly rejected (Hall & Urga, G2002).
Hall and Urga (2002) deal with these problems by using the Kalman Filter and combing the time varying parameter model with a standard GARCH-M model (generalized autoregressive conditional heteroscedasticity in mean). They apply this procedure to the two indexes of Russian stock market from 1995 to 2000. And find that with regard to RTS index (Russian Trading System), the market is initially inefficient and it takes about two and a half years to become efficient, while for the ASPGEN Index (Skate Press Agency General), the market is still predictable. There is evidence of a tendency towards being efficient. Kvedaras and Basdevant (2002) also investigate the market efficiency in the three Baltic States: Estonia, Latvia and Lithuania by using the time-varying variance ratio statistic robust to heteroscedasticity based on time-varying autocorrelations. They find a clear trajectory to weak-form efficiency in the Estonian and Lithuanian capital markets. Its relatively small inefficiency can be explained by transaction costs and information acquiring costs (Grossman and Stiglitz, 1980). In the Latvian market, it is inefficient even at the very end of the analyzed period.
These results have some important implications for developing effective institutional and regulatory frameworks. Since infrequent trading negatively affects market efficiency and liquidity in emerging markets, economic policy makers should pay attention to minimize the institutional restriction and barriers on capital flow in the financial markets, impose strict disclosure requirements and ensure that investors can easily access to high quality and reliable information. Improving liquidity of capital markets can provide lower borrowing costs for investors and greater opportunities for investment diversification with lower systematic risks. In addition, equity market liberalization is important to help achieving market development. It can reduce cost of capital and increase capital productivity with better capital allocation.
6.1 Short Summary
In conclusion, as two main schools of thought in modern financial theories, there is a hot debate between efficient market hypothesis and behaviour finance. EMH asserts that financial markets are informationally efficient and equity stock prices instantaneously and fully reflect all known information. While behaviour finance argues that psychological biases lead to investor irrationality and limits to arbitrage impede exploitation of mispricing opportunities, so market is not efficient. There are wide empirical researches on the issue of market efficiency in emerging markets with mixed results. It is generally found that most of emerging markets are still inefficient, but after correcting for institutional characteristics and trading conditions, such as thin trading and the presence of non-linearity, some researches find that equity markets are efficient for some countries such as for Istanbul, Egypt and Jordan. When structure break factors are taken into account, market efficiency is powerfully rejected for most emerging countries such as Argentina, Brazil, Greece and India. There are also some evidence shows that duo to economic liberalization policies, many emerging markets are moving towards more efficiency such as Estonian, Lithuanian and Russia.
6.2 Limitations of Empirical Researches and Proposed Further Research
However, there are some limitations involved in these empirical researches. Some researches ignore whether the distribution is normal or not. Others using equally weighted indices may bias the results. The possible auto-correlation might be due to the noise traders but doesn’t imply return predictability (Cuthberston, 1996). Most of these studies focus on the test of time series of equity return to investigate EMH, but don not investigate the momentum effect or the profitability of technical trading to earn abnormal return. Therefore, further research can be extended in several dimensions. Firstly, it suggests trying to combine the tests of momentum effect or technical trading rules with the time series tests to make more robust conclusions. Secondly, since most of researches focus on traditional EMH, it can consider the factors of investor behaviour, such as psychologies bias and limits to arbitrage to do further in-depth testing of EMH. Finally, further researches for more novel and accurate methodologies of testing EMH are significantly essential.