4 zijmr vol2 issue5 may 2012 - copy

ZENITHInternational Journal of Multidisciplinary Research Vol.2 Issue 5, May 2012, ISSN 2231 5780

STOCK MARKET EFFICIENCY IN NEPALJEETENDRA DANGOL**Lecturer, Public Youth Campus, Tribhuvan University, Nepal.

ABSTRACT The paper examines random-walk behaviour and weak-form market efficiency on daily market returns of All Share Price Index (ASPI) and Sensitive Index (SI) on the Nepal Stock Exchange (NEPSE) using Lo and MacKinlays (1988) variance-ratio tests and runs tests for the period between September 13, 2006 to May 13, 2010. The study finds that the randomwalk hypothesis is strongly rejected for both indices. There is no evidence for weak-form efficiency in either series. It implicates that market participants have opportunities to predict future price and earn abnormal returns from the Nepalese stock market. KEYWORDS: Market efficiency, Random walk hypothesis, Runs tests, Variance ratio test. ___________________________________________________________________________ I. INTRODUCTION In an efficient stock market, share prices reflect all information available to market participants and that, by implication, share prices cannot be predicted, thus precluding any abnormal profit returns. From view-point of market participants, the stock price behaviour is very important to determine future abnormal returns. Thus, this paper intends to measure the behaviour of the stock returns in the emerging Nepalese stock market. Once the behaviour of the stock returns is determined, then one can better understand the market and the economy. It makes stock prices reflect the true picture of the company as well as the condition of the overall economy. It can provide better confidence to decision-makers on their investment decisions and help in reducing the level of risk. The prior assumption of the study is that the market is efficient and the series follow a random-walk. If this is true, then past information including past prices are irrelevant in predicting future stock prices for the companies listed in the Nepalese stock market. If successive returns are independent, then, the market is said to be efficient in its weak-form.www.zenithresearch.org.in

The Section II of the paper briefly surveys the related literature. Section III contains details about the data, methodology and empirical results, while Section IV consists of the conclusion and implications. II. LITERATURE REVIEW For weak form tests, information can include only past history of security prices. Tests for weak-form market efficiency are, more generally, referred to as test of return predictability (Fama 1991). The weak-form of market efficiency is investigated by examining whether stock prices in equity markets exhibit specific patterns, which allow future prices to be predicted. For a market to be efficient in weak-form then no such patterns should exist and

40


prices should follow a random walk. The weak-form inefficiency of the stock market provides an opportunity to the traders for predicting the future prices and earning abnormal profits. Fama and French (1988) reported that NYSE has negative serial correlation (mean reverting) in market returns over observation intervals of three to five years, i.e., stock returns are predictable. They argued that autocorrelation may reflect market inefficiency or time-varying equilibrium expected returns generated by rational behaviour. On contrary to Fama and French (1988) study results; Lo and MacKinlay (1988) using a simple volatility based specification test, concluded that the NYSE-AMEX return indices showed positive serial correlation in market returns and the random-walk model is strongly rejected. They argued that the negative serial correlation in Fama and Frenchs (1988) study for long (three- to five-years) holding-period returns was, on purely theoretical grounds, not necessarily inconsistent with positive serial correlation for shorter holding-period returns. They also claimed that the sum of a random-walk and mean-reverting process cannot be a complete description of stock-price behaviour. Similarly, Lo and MacKinlay (1988) opined that the rejection of the random-walk model does not necessarily imply the inefficiency of stock price formation. Jegadeesh (1990) showed that the monthly returns on individual stocks exhibited significant negative first-order serial correlation and significantly positive higher-order (longer lags) serial correlation. The study also showed that the return in January was significantly different from other months. The stock returns showed a specific pattern. It is a strong evidence of predictable behaviour of security returns. It indicates the rejection of the hypothesis that the stock prices follow a random-walk. The author pointed out that the predictability of stock returns can be attributed either to market inefficiency or to systematic changes in expected stock returns. Fama and French (1988), Lo and MacKinlay (1988), and Jegadeesh (1990), the studies in the developed markets, showed the predictability of future returns and concluded that the market was inefficient in weak-form, i.e., price formation is dependent on or follow specific patterns. But, these studies did not explain the economic implication of the inefficient markets. Similarly, using monthly index prices in local currency, Urrutia (1995) tested the efficiency of Latin American countries: Argentina, Brazil, Chile and Mexico. The time series behaviour of sample Latin American equity prices did not seem to fit mean-reverting processes either, since variance-ratios larger than unity imply positive return autocorrelation. Thus, results of the variance ratio test rejected the random walk hypothesis for all sample equity markets. However, findings from the run tests indicate that the four Latin American equity markets are weak-form efficient. Author argue that both the economy and the capital markets of developing countries have been growing at an unusually rapid pace, and it is likely that positive autocorrelations are indicators of economic growth rather than evidence against the efficient market hypothesis. Thus, based on the results of the run test, the author concluded that the four Latin American emerging equity markets were weak-form efficient. Urrutia (1995) is successful to link between the market inefficiency and economy, which are lacking in the previous studies; for example, Fama and French (1988) and Lo and MacKinlay (1988).

41

www.zenithresearch.org.in


Recently, Worthington and Higgs (2009) examined efficiency in the Australian stock market for long period of 12,519 daily and 1,575 monthly observations. They reported that the monthly Australian stock returns followed a random-walk, but daily returns did not because of short-terms autocorrelation in returns. In the context of Nepal, using autocorrelation and runs tests, Bhatta (2010) and Dangol (2010) found that the Nepalese stock market did not follow random-walk hypothesis as well as inefficiency in weak-form for daily, weekly and monthly market returns series. On the contrary, Pradhan and KC (2010) reported inconclusive results of their study regarding the random-walk hypothesis and weak-form of market efficiency using autocorrelation and runs tests for weekly stock prices of 26 individual companies on three years period between midJuly 2005 and mid-July 2008. The previous studies show that the mixed results regarding the random-walk hypothesis and weak-form of market efficiency. The reasons for inefficiencies are observed due to autocorrelation structures in their returns series. The developed markets show autocorrelation on its returns series, may be systematic changes in expected stock returns or rational behaviour of the investors. On the other hand, the majority of the emerging equity markets provide positive autocorrelation indicating unusual rapid economic growth. III. DATA, METHODOLOGY AND RESULTS The paper employed daily market returns of the two indices, namely, All Share Price Index (ASPI)1 and Sensitive Index (SI)2 of Nepal Stock Exchange (NEPSE) from September 13, 2006 to May 13, 2010. The paper used the above said periods because to capture the following political and non-political events: (1) NEPSE just move to semi-automation process from traditional trading system (August 24, 2007); (2) political uncertainties, for example, election of Constitution Assembly, the Maoist led first republican government after the Constitution Assembly and was replaced after nine months by the UML party; (3) increased lending as well as deposit interest rates by financial institutions; (4) experience of liquidity crisis in financial institutions; (5) increased capital gain tax from 10 to 15 per cent and then reduced it from 15 back to 10 per cent; (6) mandatory provision to declare income source while trading on shares worth one million Nepalese rupees and above; (7) huge size of rights-share and initial public offerings; and (8) capturing of both up (bullish) and down (bearish) market trends3. The study used stock market returns as an individual time-series variable. Market returns are calculated from the daily price indices. Daily Market returns (Pt) are calculated from the price indices as follows:www.zenithresearch.org.in

1

The ASPI is based on market prices of all stocks listed with the NEPSE. At present 173 companies are listed in NEPSE. 2 The SI is based on the market prices of group A shares listed with NEPSE. The Compan ies should fulfill the following criteria for group A share: (1) the number of shareholders must be at least 1000; (2) the company must be in profit since last three years; (3) the paid up capital of the company must be at least Rs 20 million; (4) the book value per share must not be less than its paid up value; and (5) submission of the financial statement within six months from the closure of the fiscal year. At present 94 companies are classified under group A category in Nepal. Nepal formally implemented the sensitive index from September 13, 2006. 3 Highest NEPSE index of 1175.38 points on August 31, 2008, after that the market has showed downward trend till date.

42


Pt = Ln

PI t PI t -1

.......................................................................... (1)

Where, Pt refers to market return in period t; PIt, price index at period t; PIt-1, the price index at period t-1 and Ln refers to natural log. The reasons to take logarithm returns are justified by both theoretically and empirically. Theoretically, logarithmic returns are analytically more tractable when linking returns over longer intervals. Empirically, logarithmic returns are more likely to be normally distributed, which is a prior condition of standard statistical techniques (Strong, 1992). To test the weak-form of market efficiency, the paper has first examined the normal distribution of stock returns. If stock returns series follow a normal distribution, it belongs to the assumption of random-walk model; hence the market is accepted as having the weakform of efficiency. The paper tests normality using the skewness, kurtosis and Jarque-Bera statistic. Descriptive statistics can be interpreted to test the informational efficiency of stock market. Generally, values for zero skewness and kurtosis at three represent that the observed distribution is normally distributed. Table 1 shows the descriptive statistics of daily returns of All Share Price Index (ASPI) and Sensitive Index (SI). The distribution of daily stock returns have slightly negative-skewed but it is highly leptokurtic (peaked). Therefore, skewed and leptokurtic frequency distribution of daily market returns series indicate that the distributions are not normal. Jarque-Bera test also rejects the null hypothesis of normal distribution for both indices; ASPI and SI. It gives evidence that the frequency distribution is not normal. But the positive mean return and low variance indicate that the Nepalese stock market involves low risk. TABLE 1: DESCRIPTIVE STATISTICS OF DAILY RETURNS ASPI Observations Mean Median Maximum Minimum Standard Deviation Skewness Kurtosis Jarque-Bera 838 0.00021 -0.00033 0.04894 -0.07228 0.01402 -0.13769 4.91434 130.60640 SI 838 0.00017 -0.00018 0.05707 -0.07129 0.01534 -0.25450 5.21239 170.95110www.zenithresearch.org.in

43


Probability

0.00000

0.00000

VARIANCE-RATIO TESTS OF RANDOM-WALK The paper uses the variance-ratio method of Lo and MacKinlay (1988) to test for randomwalk in the Nepal stock exchange (NEPSE). The idea behind the variance-ratio test is that if the natural logarithm of a time series Yt is a pure random-walk, the variance of its qdifferences grows proportionally with difference q. That is, the variance of the increments in a random-walk is linear in the sampling interval. Therefore, if a time series follows a randomwalk process, the variance of its q-differences should by q times the variance of its first differences. The variance- ratio, VR(q), is defined as: VR(q) = 2 (q ) 2 (1)

........................................................................ (2)

Where, 2(q) is 1/q the variance of the q-differences and 2(1) is the variance of the first differences. According to Lo and MacKinlay (1988), formulas for the calculation of 2(q) and 2(1) are as follows: 2(q) =1 q q(nq - q + 1)(1 ) nqnq

(Ytt q

Yt

q

q ) 2 .. ................... (3)

and 2(1) = where, 1 (Ynq nq Y0 ) 1 ( nq 1)nq

(Ytt 1

Yt

1

) 2 ............................................ (4)

(q) =

2( 2 q 1)( q 1) ............................................................. (5) 3q ( nq )

Under heteroskedasticity, the asymptotic variance can be expressed as:q 1

*(q) =k 1

2( q k q

2

( k ) ..................................................... (6)

44


Y0 and Ynq are the first and last observations of the time series. The test is performed under both homoskedastic and heteroskedastic specifications. Under homoskedasticity, the asymptotic variance of the variance ratio is expressed as follows:

ZENITHInternational Journal of Multidisciplinary Research Vol.2 Issue 5, May 2012, ISSN 2231 5780nq

(Yt

Ytnq

1

) 2 (Yt Yt

k

Yt )2

k 1 2

) 2

where, ( k )

t k 1

(Ytt 1

1

The homoskedasticity and heteroskedasticity consistent Z-statistics are denoted by Z(q) and Z*(q) and expressed as follows: Z(q) = and Z*(q) =VR ( q ) 1 * (q )

VR ( q ) 1 (q )

~ N(0,1) . ..................................................... (7)

~ N(0,1) .

(8)

Under a single variance-ratio test, the null hypothesis is that VR(q) = 1 or that the chosen index follows a random-walk. If the null hypothesis is rejected and VR(q) > 1, then the computed Z(q) and Z*(q) are positive and returns are positively serially correlated. If the null hypothesis is rejected and VR(q) < 1, then the computed Z(q) and Z*(q) are negative and returns are negatively serially correlated, i.e., mean reverting. Table 2 reports the variance-ratio tests, which are computed for interval q = 2, 4, 8, 16 daily observation interval. The rejection of the random-walk hypothesis under homoskedasticity is not sufficient on its own, as it could be due to heteroskedasticity or autocorrelation in the examined series. Hence, it is important to focus mainly on heteroskedasticity consistent Zstatistics. As per Table 2 of the variance-ratio test, Z-statistics are negative with statistically significant for both homoskedasticity and heteroskedasticity in ASPI and SI. Hence, the random-walk hypothesis is strongly rejected for both indices. Similarly, the empirical findings reveal that the null hypothesis of random-walk for both selected indices cannot be accepted for all levels of interval q at the one per cent level of significance. The both sample indices: ASPI and SI, variance-ratio values below one and they decrease with the interval q increases. It indicates negative serial correlation in the returns and potential mean reversion. In other words, if stock price-returns do revert means, they should be negatively serial correlated, and the variance ratio should get smaller and smaller than unity as the interval q increases. This type of behaviour is generally observed in emerging financial markets that may suffer a bubble effect (Summers, 1986).

45



TABLE 2: VARIANCE-RATIO TEST Indices ASPI Q 2 4 8 16 SI 2 4 8 16 VR(q) 0.7408 0.3685 0.1869 0.0935 0.7227 0.3686 0.1842 0.0942 Z(q) -7.4996* -9.7664* -7.9520* -5.9583* -8.0226* -9.7644* -7.9793* -5.9533* Z*(q) -4.4722* -6.5124* -6.1421* -5.1226* -4.7936* -6.5957* -6.3354* -5.2553*

Note: Number of observations: 838 for each index. The asterisk denotes statistical significance at the 0.01 level with a critical value equal to 2.57. Similarly, the observations are split up into two equal sub-samples and the results of variance-ratio tests are reported not different with overall sample observations. The paper provides the evidence of variance-ratios lesser than one suggesting negative returns autocorrelation. The results tend to disagree with Lo and MacKinlay (1988) who find positive autocorrelation (variance-ratio larger than one) for the New York Stock Exchange (NYSE) and American Stock Exchange (AMEX). Similarly, the results of negative autocorrelation are contradicted with Worthington and Hinggs (2009) in Australia who finds value of variance-ratio are larger than one showing positive autocorrelation. But the meanreverting process of stock returns is documented by Fama and French (1988), and Jegadeesh (1990). TESTS OF WEAK-FORM OF MARKET EFFICIENCY The paper has also tested for weak-form market efficiency using run test, which is a nonparametric test used for detecting the frequency of the changes in the direction of a time series. The run test is accomplished by computing the expected runs and the actual runs for the sample returns. The expected number of runs is represented as under: E(R) =2 n1n2 n n

...................................................................... (9)

Where n represents the total number of observations, n1 and n2 represent observations that equals or above and below the sample mean (or median), and R represents the observed number of runs. The standard error can be written as under:

46



(R) =

2 n1n2 ( 2 n1n2 n ) ...................................................... (10) n 2 ( n 1)

The asymptotic (and approximately normal) Z-statistic can be written as follows: Z(R) =R E( R ) ........................................................................ (11) ( R )

Table 1 depicts that the both series are not normally distributed. Thus, non-parametric runs test, which tests for independence between successive events (in a series) without requiring normality of distribution, is used to test for weak-form efficiency of the selected indices ASPI and SI. Table 3 presents the tests of independence, i.e., runs test. The negative Z-values for both returns series indicates that the actual number of runs falls short of the expected number of runs under the null hypothesis of returns independent at the 0.01 level. The negative Z values for the both indices indicate positive serial correlation. The run test shows that the successive returns for both indices under study are not independent at the one per cent significant level. There is no evidence for weak-form efficiency in either series. TABLE 3: RUNS TESTS OF DAILY MARKET RETURNS Indices ASPI SI Mean 0.00021 0.00017 N 838 838 n1 434 435 n2 404 403 E(R) 419.463 419.389 R 260 292 Z(R) -11.038 -8.819 p-value 0.000 0.000

Note: The results of runs test are based on mean value. The runs tests were also performed with the median as base and it also reported similar results. Similarly, the observations are split up into two equal sub-samples and the results of runs tests are reported not different with overall sample observations. VI. CONCLUSION AND IMPLICATIONS This paper examines the random-walk hypothesis and weak-form market efficiency in the Nepalese stock market employing variance-ratio and runs tests. The tests used on two important daily market indices; namely, ASPI and SI. Since the variance-ratio is less than unity, the random-walk hypothesis for both indices is strongly rejected. Similarly, the runs tests have rejected the independence of stock price movements that indicated the Nepalese stock market as inefficient in weak-form. The Nepalese stock market is inefficient in daily returns series suggesting that past movements in stock prices can be used to predict their future movements. It provides market players bring the possibility of earning higher returns than expected. The presence of a random-walk in the stock data has an important implication for portfolio investors, the allocation of capital within an economy and hence overall economic development. It is, therefore, relevant to suggest that there should be an effective regulatory framework and its implementation; and a more effective role by all the stakeholders should be helpful in making the market reflective of a true picture of the economy.

47



The possible reasons for the market inefficiency in the emerging stock market like Nepal may be the poor institutional infrastructure, weak legal framework, lack of supervision, slow development of the market infrastructure, poor corporate governance and accountability, low level of capacity of major market players and lack of transparency of market transaction. The study provides the time series behaviour of a less developed market of Nepal. The processing of new information in Nepal is rather weak, it may result from persistent large number of non-actively traded shares, limited role of mutual fund and lack of professionally managed investment and broker houses. The absence of qualified analysts, institutional investors and investment-friendly environment is a well-known constraint in the emerging market like Nepal. The main challenges of the Nepalese stock market include frequently changing government policies, policy-level interventions, tussles of regulatory authorities, lack of commitments on economic agenda from political parties, and slow economic growth. Other challenges are to widen the use of automation, strengthen regulations and supervision, and educate investors. Similarly, functional autonomy of SEBON (regulatory body), and NEPSE (implementation body) are also the need of the day. Further research is recommended to assess whether this papers findings are verifiable using alternative econometric procedures. REFERENCES Bhatta, G. P. (2010). Does Nepalese Stock Market Follow Random Walk?, SEBON Journal, Volume 4, p. 18-58 Dangol, J. (2010). Efficient Market Hypothesis and the Emerging Capital Market in Nepal, In Efficient Market Hypothesis: Few Empirical Evidences from Nepalese Stock Market, Kathmandu: Quest Publication Fama, E. F. (1991). Efficient Capital Markets: II, Journal of Finance, Volume 46, Issue 5, p. 1575-1617 Fama, E. F. and French, K. R. (1988). Permanent and Temporary Components of Stock Prices, Journal of Political Economy, Volume 96, Issue 2, p. 246-273 Jegadeesh, N. (1990). Evidence of Predictable Behavior of Security Returns, Journal of Finance, Volume 45, Issue 3, p. 881-898 Lo, A. W. and MacKinlay, A. C. (1988). Stock Market Prices Do Not Follow Random Walks: Evidence from a Simple Specification Test, The Review of Financial Studies, Volume 1, Issue 1, p. 41-66www.zenithresearch.org.in

Pradhan, R. S. and KC, S. (2010). Efficient Market Hypothesis and Behaviour of Share Prices: The Nepalese Evidence, SEBON Journal, Volume 4, p. 104-117 Summer, L. H. (1986). Does the Stock Market Rationally Reflect Fundamental Values?, Journal of Finance, Volume 41, Issue 3, p. 591-601 Urrutia, J. L. (1995). Tests of Random Walk and Market Efficiency for Latin American Emerging Markets, Journal of Financial Research, Volume 18, Issue 3, p. 299-309 Worthington, A. C. and Higgs, H. (2009). Efficiency in the Australian Stock Market, 18752006: A Note on Extreme Long-Run Random Walk Behaviour, Applied Economics Letters, Volume 16, p. 301-306

48

4 zijmr vol2 issue5 may 2012 - copy

Documents