Efficient Market Hypotheses –EMH (Lebanon and Arab Gulf States)

Various studies have been made to test the efficiency of market in Lebanon and Arab Gulf States (AGS) theoretically and empirically, focusing mainly on building a model using the conventional finance. The purpose of this paper is to test the efficiency of AGS financial markets as well as the Lebanese one based on stationary method; furthermore it explores the reasons behind the non-efficiency in case it exists. However, to realize this objective ,this paper overviews the different methods behind testing efficiency in order to choose and use the most common one ,furthermore it studies also the Volatility using ARCH and GARCH and the Financial Characteristics behind each market, which leads to either enhance or reduce market efficiency so far. Findings prove that Lebanese, Abu Dhabi, Qatar, and Kuwait markets tend to be efficient in its weak form whereas Dubai, Saudi Arabia and Amman markets are non-efficient based on the average monthly data series and financial characteristics of the market over a specific period of time.


Introduction
Market Efficiency has been linked to many features in different domains. However, finance has been especially interested in "Informational efficiency". Where market efficiency is considered as the moderate concept of perfect market, that's to say rational way of thinking is widely spread between investors during investment decision process leading to efficiency in the financial markets. Fama (1965) the founder of market efficiency theory, defines the efficient market as being a market where "prices fully reflect all known information in a way that even uninformed investors exchanging a diversified portfolio at the tableau of given prices by the market will obtain a rate of return as generous as that achieved by the experts". In other terms, a market is efficient when the intrinsic value of the stocks reveals correctly the information. Fama's definition of market efficiency has been largely supported by theoreticians (Malkiel, 1992).
Three main categories of the market can be considered (weak form efficiency, Semi strong form efficiency where all the publically information is considered in addition to some insider information, and the strong form efficiency where all public and private information are revealed in stock prices).Weak form efficiency market is the most important form followed by researchers, as the historical data and path of the stock is the main key behind predicting its future prices (Fama, 1970). Samuelson (1965) have stated that ,when future prices of securities in the market are being randomly assigned and not based on the historical trend of the stocks; the market is efficient in its weak form. In other terms, the market is efficient in the weak form when prices follow a random Walk process.
A large body of research, using econometric test including unit root analysis, and variance ratio frame work, suggests that Developed markets are being efficient in weak form (Choudhry, 1994;Chan and Pan, 1997;linoswski, 2005;Maria, 2007;shiller and radikoko, 2014); whereas emerging markets tend to be non-efficient (Pant and Bishnoi, 2002;Smith, 2007;Vigg et al, 2008).
However, when market is efficient it usually attracts various investors to channel big amount of funds into numerous profitable projects enhancing economic growth. As results, efficiency is positively related to economic growth (King and Levine,1993;Pagano,1993;loayza and beck,2000;Equakun,2005).On the other hand, when markets tends to be "inefficient", speculators are attracted to it ,as a result market fluctuations control all over, hence volatility increases and financial bubble is constituted (Ewah et al,2009).
The purpose of this paper is to investigate the efficiency of Arab Gulf State financial markets as well as Lebanese financial market .Furthermore this research investigates the financial characteristics for each market impacting efficiency, to realize this objective this paper overviews the main methods for testing efficiency, in addition to that this paper studies the Volatility and Financial Characteristics behind each market impacting its efficiency. The paper is organized as follows: Section 2 presents the related literature background. Section 3 overviews Market efficiency and the random walk model. Section 4 investigates the available methods for testing efficiency. Section 5 analyzes market Volatility. Section 6 presents efficiency test results and results analyses for Lebanese and Arab Gulf States financial markets, ending up with general conclusion in section 7.

Market efficiency in the literature
Wide researches analyze the market efficiency and its relation to economic growth for developed and emerging countries. However Lebanese and AGS financial market studies are limited as a result this study aims to bridge the presented gap.
Market efficiency has been examined mainly by the mid of 1970's by the famous Eugene Fame, who explained that capital market efficiency, means that the stock prices reflects fully and correctly all relevant information which implies that stock prices are responding to all the information set coming from sudden economic and political conditions by adjusting prices. Otherwise, many developed researches have appeared. Chan, Gup and Pan (1997) tested the weak form market efficiency for numerous international stock indices, collecting 384 monthly observations from 1962-1992. Using serial correlation and Run tests, they proved that weak form market efficiency exists clearly in all the studied markets. However, Choudhry (1994) has tended to analyze the weak form efficiency market for seven members of the OECD organization using co-integration test for monthly stock indices from . Results have confirmed the presence of efficiency in these seven countries. However, Portuguese stock markets has been analyzed over the period of 1993-2008 using serial correlation test for daily, weekly and monthly returns, where Portuguese stock market has been moving toward random walk (Maria,2007 -correlation in order to analyze the daily stock index return of these markets where large variations in return has appeared over the assigned period, indicating that the markets are not efficient in the weak form. furthermore; Batool Asiri (2008) and Rakain Gupta (2007) studies the presence of weak form efficiency in the Indian financial market over the period 1990-2006 based on monthly Data series ,using Autocorrelation and unit root test (phillips-perron, ADF and KPSS) where it appeared that the Indian market is not a weak form efficient one. However, obtaining different results for the same market in different studies doesn't mean that there exists contradiction in findings, but on the other side this difference may be due to the type of the used data over a certain period of time while applying different tests.
Unlike most of the existing literature that have been studying the efficiency in the middle eastern markets; this paper goes through an overview to investigate the different methods for testing efficiency in Lebanon and AGS financial markets in order to use the most convenient one to our study; However Volatility and financial characteristics of these markets will be taken into account in order to support the results obtained. All tests are carried out with E-views.

Market Efficiency and random walk Model
The term of random walk model has appeared in the beginning by Jules Regnault, (1863) and has begun to be extensively investigated and tested by Louis Bachelier, (1964). However, the period around 1990's new thoughts have been considered regarding behavioral Finance that was known as being the modern theory, where it started contradicting the Random walk theory, by focusing on the impacts of investors' behaviors. Some of the authors that denied the random walk in their articles were (Lo and MacKinley, 1999;Lo, Mamaynski and Wang, 2000). One from the most important models for testing the behaviors of the stock prices was the "Random walk model"; as TheRandomwalkhypothesisconsiders that futurechangesofprices cannot beanticipated. However, an increase of prices in a particular day does not mean an increase or diminish within the day after. Moreover in a random walk Market the series of stock market prices changes are independent, and the efficiency idea is based mainly on no excessive profit ; that means the series of stock market prices changes are independent, thus Historical data of prices can't be used to predict the future (Brealey et al, 2005).
In 2004, Cuthbertson and Nietzsche have tended to define random walk with a drift (d) where the individual Stochastic series X t behaves as: Where the "Drift" can be considered as the weighted average of the probabilities of each price of the stock.

Methods for testing Market Efficiency
From the definition of efficiency, various statistical tests can be done in order to know whether Lebanese and Arab Gulf state markets are considered efficient or not, by analyzing the average monthly database regarding stock market indices.
A. Testing efficiency using Random Walk test Cooray and Wiskremasingue (2005) used the unit root test in order to test for weak form efficiency in different countries (India, Sri-lanka, Bangladesh) where they found that these markets are efficient. Moreover, Smith (2006) has used the same method to to test for efficiency in its weak form through European countries, where it has recorded a great support for Random walk.
The most famous Random walk tests for Stationarity in the time series data are: • Augmented Dickey fuller-ADF, Phillips-Perron test with Yt = μ + αyt−1 + εt (2) and Zivot-Andrews tests were used in order to test for the same null hypothesis which is: H0: Unit root is presented in the Data series understudy. • Whereas KPSS test Yt = ξt + rt + εt where rt = rt−1 + ut (3) tests for the null hypothesis which indicates that : H0: No unit root or Stationarity in the time series understudy.
The results are decided based on the "Rule of thumb" that's to say if the calculated value statistic is greater than the critical value ; Null hypothesis will be rejected thus stationarity exists for ADF and PP tests and vice versa for KPSS.

B. Testing efficiency using Serial correlation (auto correlation)
Serial Correlation test has been used in order to examine the linear dependency of lagged price changes of returns especially in USA (jarrett, 2008). However, Butter and Malaikeh (1992) have adopted serial correlation test for Kuwait and Saudi markets, where it appeared that weak form efficiency is not supported in these markets.
Allahyani (2009) used also serial correlation and Run test to examine the weak form efficiency for Tehran stock exchange (TSE) using weekly data over the period 1999-2005, where no existence for weak form efficiency market appeared.
Serial correlation has been one from the most used Statistical tests to explore weak form efficiency of the market, where it tends to measure the dependence of a variable on its past values, that's to say measuring the relationship between the current stock price and its previous value; Statistically, the hypothesis of weak form efficiency should be rejected if stock returns (price changes) are serially correlated (rk is significantly different from zero) (Taylor, 2014). (4) Where r k is the serial correlation coefficient of daily returns of lag k; n is the number of observations; r t is the stock return over period t; r t -k is the stock return over period t-k; is the mean of stock returns, and k is the lag of the period. r t is measured as follows: Where, r t is the index of stock market in the period t, and r t−1 is the index of stock market in the period t-1.This test is frequently performed to test the joint hypothesis that all the r k , up to certain lags, are simultaneously equal to zero, instead of testing the statistical significance of any individual autocorrelation coefficient (Gujarati, 2003).

C. Testing efficiency using Co-integration
Co-integration is an econometric technique for testing the correlation between non-stationary time series variables. If two or more series are themselves non-stationary, but a linear combination of them is stationary, then the series are said to be co-integrated and a long run relationship exists, thus weak form market efficiency appears to take place and prices in turn reflects fully and correctly the available information which leads to an increase in the economic growth so far.
Two time series X t and Y t are co integrated when there exists a number α 1 in the linear equation: yt = α 1 x t + ν t (6) such that V t is a stationary process.
"Linear relationships involving integrated nonstationary time series are meaningful only if the time series are co integrated. There are various definitions of co integration; as originally proposed by Engle and Granger. A co integrating relationship between two or more time series each having unit roots (I (1) )is defined to exist if there is a linear combination that is stationary, i.e., I(0)" (Edward Herranz ,2017).

D. Testing efficiency using Run Test
"The run test, as a non-parametric test used to detect the frequency of changes in the direction of a time series. Runs test is a strong test for randomness in examining serial dependence in asset price movements.
Runs are defined as the number of sequences of consecutive positive and non-positive (negative or zero) returns. When the expected number of runs is significantly different from the observed number of runs, the test rejects the null hypothesis that returns are random (Gu and Finnerty, 2002). To perform this test, the number of actual runs (R) is compared with the expected runs (m) Where, N is the number of observations (daily returns), i is the sign of +, -, or 0, and n i is the number of observations in each run.
Then, the standard normal Z-statistic is used to test whether the actual number of runs is consistent with the random walk hypothesis which means that weak form efficient market exists. The standard normal Z-statistic is calculated as follows (Taylor, 2014): "Where R is the actual number of runs, and 0.5 is continuity adjustment. When the actual number of runs exceed (fall below) the expected runs, the result will be a positive (negative) Z value. A positive (negative) Z value indicates negative (positive) serial correlation in the return series" (Abraham, Seyyed and Alsakran, 2002).

E. Chosen Method
This paper tests for weak form market efficiency in Lebanon and Arab Gulf States by adopting the most commonly and famous method, which is the "Random Walk model" including the ADF,Phillips-Perron and KPSS tests to investigate the presence or the absence for a unit root in the assigned data series.

Volatility measure and formation
After testing for Stationarity of return series, estimating equation 1 and 2, regarding the Volatility of the market would be necessary. Volatility can be defined as a statistical measure of the dispersion of returns for a given market index or security. It can be measured by variance or standard deviation between returns from that same market index or security. When the volatility is higher, the security becomes riskier. A high volatility market is a market where prices tend to change a lot over relatively a short time. The volatility is associated with the risk, uncertainty and unexpected. In other terms, volatility can be referred to the amount of uncertainty or risk about the size of changes in the value of security. It is considered as the difference between market price and economic fundamentals that rationally justify the value of the assets concerned. The beta is a measure of volatility of a portfolio or a security in comparison to the market as a whole.
Volatility measure of index series is based on the returns of the data (R t = logp t -logp t-1 ). Generalized Autoregressive Conditional Heteroscedasticity (GARCH) formulation which has been first suggested by Bollersley (1986) and has become popular, specially, due to its explanatory power for dependence in volatility tests if the variance of returns is stationary and whether index eventually return back to the mean value. It tests then an equation specification for the mean of the return series (9) and an equation for the conditional variance (10) of the returns: Rt= logp t -logp t-1 = c + ɛ t (9) σ 2 t= ω + αε 2 t-1 + βσ 2 t-1 Where ɛ t ~ N (0, σ 2 t) and σ 2 t = E (ɛ 2 t).
This specification is often interpreted in a financial context, when an agent trader predicts variance of this period by forming a weighted average of a long term average (the constant), the forecasted variance from last period ( the ARCH term: α), and information about volatility observed in the previous period ( the GARCH term: β). If the asset return was unexpectedly large in either the upward or downward direction, then the trader will increase the estimate of the variance for the next period.

Data Source
The Data used in the proceeding part for the empirical study has been based on "Average monthly market indices" for Lebanon, Dubai, Abu Dhabi, Saudi Arabia, Kuwait, Amman and Qatar covering different periods of time, assigned exclusively for each country through its section. However the used data has been collected from the market's site of each one, in addition for "Gulfbase.com".

A. Lebanese Financial market characteristics
"Beirut stock Exchange (BSE) is the second oldest stock market in the region, as it was established in 1920 by a decree of French commissioner. Initially, trading was restricted to gold and foreign currencies. In the early 30s, trading was expanded to encompass shares of private companies set up under the French mandate to operate and manage some public services and sectors (railways, communications, post...). It was then tantamount to privatization. Some of these corporate securities and shares were listed on both BSE and Bourse de Paris at the same time." However, "Beirut Stock Exchange incorporates three types of markets: • The official market: For companies incorporated for more than three years with a capital equivalent to USD three million at least, having distributed a minimum of 25 per cent of their share capital to the public and such percentage (25 per cent) being held by 50 shareholders at least. • The junior market: For newly established companies with a capital equivalent to USD one million at least, having distributed a minimum of 25 per cent of their share capital to the public and such percentage (25 per cent) being held by 50 shareholders at least. • Over the Counter market: For Lebanese companies with a capital equivalent to USD 100.000 at least.
The shares of such companies are traded without being listed on the Beirut Stock Exchange".
Time series Data was collected from Beirut stock exchange (bse.com.lb) regarding average monthly market index over the period January 1997 till December 2017, testing for market efficiency and volatility

A.1. Stationarity Test Results
The presence of a unit root in the Lebanese market Index series is confirmed by the Augmented Dickey Fuller (ADF); Phillips-Perron and KPSS tests. The results of these tests are presented in table 1 indicating that taking the first differences, the Lebanese market Index series become stationary. In other terms, the series is integrated of order 1 (I (1)). The return series Rt is therefore stationary since p<5%. Source: Calculated by the author using Eviews.
σ 2 t= 4.56E+12 + 0.1900ε 2 t-1 + 0.37416 σ 2 t-1 (11) (0.606) (0.649) (0.8013) According to probability values in equation (11), the ARCH and GARCH coefficients (α and β) are not significant since their respective probability is greater than 5%. The sum of ARCH and GARCH (α + β) is 0.564, indicating that the Lebanese market volatility shocks are respectively low; however to insure more this point of view, studying the Lebanese market size in addition to the traded volume would be necessary.

A.3 Results Analyses
Graph 1 represent the Market Capitalization of the Lebanese Financial market from 1996 till 2018 .However, researches relate the persistence of low volatility and efficiency in the financial market (Andrew, 1997).
Graph 1: Beirut Stock exchange average yearly market capitalization (equities and funds) from 1996 to 2018.
Source: Calculated by the author based on Data for average yearly market capitalization collected from WORLDBANK/TRADINGECONOMICS.COM Graph 1 shows that that the lebanese market capitalization Index acquires an Upward trend ;as there were no flactuations from 1996 till 2018.However, a minimum of 1,000,000,000 USD recorded in 1996 to reach a peak of 13,000,000,000 USD in 2010 then it droped down to reach almost 11,000,000,000 USD in 2018, Moreover the Total market capitalization has recorded 24.44% of the GDP lately according to BSE financial report,as it may be considered as a small size market; Furthermore to investigate more in the lebanese stock market we can notice that the stocks traded represents as a total value only 1.82% of the GDP which is a very low percentage (BSE.com.lb). Therefore, the market efficiency of the lebanese market is effected by the low level of volatility in addition to the low level of traded volume (Nahlehb,2010) which may return to many reasons related to the lebanese investor; as the lack of transperancy in the obtained results and reports that fears him,and other psycological reasons that are set below behavioral Biases in finance. Electronic copy available at: https://ssrn.com/abstract=3431988

B.3. Dubai Financial market-DFM
In 2000 Dubai financial market has born as a public institution. Referring to Dubai financial market (www.dfm.ae), and in order to test for market efficiency using Unit Root Test before proceeding in the work, Dubai market index was collected starting from 2004 reaching 2018 order to test for market efficiency.

B.3.1. Test Results
As it appears in table 3, p=0.0025 that's to say less than 0.05 which means it's significant ,thus H0 is rejected and Dubai data series appears to be stationary at level, and weak form market efficiency is obtained; Phillips-perron also confirmed the same results.

B.3.2. Result analysis
According to probability values in equation (12), the ARCH and GARCH coefficients (α and β) are significant since their respective probability is less than 5%. The sum of ARCH and GARCH (α + β) is 1.4081, indicating that the volatility shocks are strongly persistent.The figure below shows clearly the huge volatility in dubai's market.

B.4. Abu Dhabi (ADX)/ UAE's index
On 15 November, 2000 ; Abu Dhabi securities exchange (ADX) was established ; Average monthly Data regarding the variation of market index was collected covering 14 years (2004 till 2018), in order to test the efficiency of the market.

B.4.1. Test Results
Running the Augmented Dickey-Fuller test statistics, phillips-perron and KPSS using E views, results were obtained in table 3.
Since p>0.05 (=0.1785) in ADF and (p=0.2095) in phillips-perron test, thus the series tends to be not stationary at level, thus the test is rerun at first difference to obtain p<0.05(=0.0002 and 0.0003) respectively, thus the series appears to be integrated of order 1; (I (1)) which means that the series become stationary.

B.4.2. Results Analysis
According to probability values in equation (13), the ARCH and GARCH coefficients (α and β) are not significant since their respective probability is more than 5%. The sum of ARCH and GARCH (α + β) is 0.802; indicating that the volatility shocks are weekly persistent.

B.5. Kuwait financial market
Referring to the National bank of Kuwait (www.cbk.gov.kw) (S.A.K.P) Data; that was established at 30, June, 1969; average monthly market price index was studied covering 14 years, from 2004 till 2018 to test for market efficiency.

B.5.1. Test results
As it appears in (table 1); p>0.05 at level thus the series tends to be non-Stationary; as a result it has been studied on the first difference level to obtain p=0.02<0.05 in ADF TEST and 0.001 in PHILLIPS-PERRON TEST ,thus the series tends to be integrated at first difference (I(1)).

B.5.2. Results Analysis
According to probability values in equation (14), the ARCH and GARCH coefficients (α and β) are significant since their respective probability is less than 5%. The sum of ARCH and GARCH (α + β) is 0.7409, indicating that the volatility shocks are persistent.

B.6. Saudi stock exchange
Average monthly Data for Saudi stock market indices were collected for 15 years from 2003 till 2018 in order to test for weak form market efficiency in Saudi stock financial market.

B.6.2. Results Analysis
According to probability values in equation (15), the ARCH and GARCH coefficients (α and β) are significant since their respective probability is less than 5%. The sum of ARCH and GARCH (α + β) is 1.942, indicating that the volatility shocks are strongly persistent.

B.7. Qatar Financial market
Based on the data obtained from (WWW.QE.COM.QA) that was established and started operating in 1997, and become one from the pioneer stock markets in the Gulf region. Average monthly Market indices were collected from 2006 till 2018 to test for market efficiency in its weak form.

B.7.1. Test results
Based on table (3), it can be seen that p=0.1114 >0.05 (ADF TEST) and p=0.1107 (PHILLIPS-PERRON TEST) thus the series is not Stationary at level; thus we tended to make the first difference where p=0.0006 was obtained (<0.05) and p=0.0009 with PHILLIPS-PERRON, thus the series tends to be integrated of level one (I (1)).

B.8. Amman stock exchange
Amman stock exchange (ASE) average monthly Data that is used mainly to measure the performance in terms of return was collected for 11 years from 2007 till 2018 in order to test for market efficiency in the weak form.

B.8.1. Test results
As presented in table (3), it appears that p=0.0107<0.05 (ADF TEST) and p=0.0001(PHILLIPS-PERRON TEST) thus Ho is rejected, series is stationary and the market is efficient in its week form at the first level.

B.8.2-Results Analysis
According to probability values in equation (17), the ARCH and GARCH coefficients (α and β) are not significant since their respective probability is greater than 5%. The sum of ARCH and GARCH (α + β) is 0.479, indicating the absence of the volatility shocks  (table 4). Nevertheless, It can be said that although Amman has appeared as an efficient weak form market (at level) but this doesn't mean that it acquires all the conditions of efficient market; but on the other side its low market capitalization and low traded volume has been the reason behind; however this may be related to many reasons that might be considered under investor's behavioral biases which may be an indicator for many other psychological factors that would be contributing so far in these results.

Conclusion
This paper presented the different methods to test efficiency of Lebanese and AGS financial markets in weak form, in addition to their financial characteristics which either enhances or reduces efficiency; nevertheless it investigates also the market volatility for the proceeding countries.
The overall results proved to show that Lebanon, Abu Dhabi, Qatar, Kuwait series data are integrated of order 1 (I (1)) thus they tend to be efficient in their weak form, whereas Dubai, Saudi and Amman are non-weak form efficient markets. However these results have been investigated deeply through analyzing their financial characteristics (market capitalization and traded volume) and Volatility for each of the above markets.
However, in our cases efficiency of the market doesn't necessary mean that historical data and path of the stocks are the main key behind predicting its future prices; but on the other side another components have entered in order to get a clear view regarding the efficiency of the market ,as "market capitalization" and "Traded Volume" that have been seen as the keystones in the analysis of Lebanon and Arab Gulf states markets; which has left us with other reasons that might lie behind the limited investing behavior of the traders, where Irrational way of thinking and behaving might be controlling which can be explained through lack of transparency in certain markets.
Nevertheless, the psychological part of the investor during decision making can't be denied, as it plays a big role during investment decisions where Irrationality is being spread all over leading to wrong behaviors so far, that can definitely leads to wrong decisions and big losses. Thus it can be insured that other factors are being contributing during decision making process and can be explained through behavioral biases and hidden under behavioral finance concepts.