CHAPTER 5. VARIABLES AND DATA UNDER STUDY
5.6. Descriptive statistics and normality
The tables below show the descriptive statistics of the log of market returns and the distribution of the return series. Before ARIMA was used for testing, important features of the series return were highlighted by performing a preliminary analysis. The results show for each constituent the return series with a summary of the preliminary statistics, from all the sample local currencies for the twelve-year period, the daily nominal return series was calculated from descriptive statistics. In addition, this shows the main periods from 2005 to 2016, the sub-sample for 2005-2011 and 2011-2016, three countries and five individual companies.
Also shown below are Mean, Std. Dev., Variance, Skewness, Kurtosis and Bera-Jarque. Mean estimate is based on the daily returns and reports from the descriptive statistics. Negative or positive returns generated by assets determine the mean, and asset volatility. Risk and uncertainty are measured by Variance. Therefore, there is greater risk for decision-based returns when certainty is lost, likely realised values become less certain and there is a wide dispersal of returns distribution that occurs when variance becomes larger. Volatility is another term for loss of certainty. In addition, skewness was calculated. A positive skewness coefficient is shown by a right skewed distribution, a zero-skewness coefficient is shown by symmetric distributions, and a negative skewness coefficient is shown by a left skewed distribution. There is greater risk for investors when large negative returns are generated by assets indicated by a negative skewness. In contrast, there is lower risk for investors when large positive returns are generated by assets indicated by a positive skewness. Moreover, Extreme values for a given variance and Mean are produced from the distribution of heavy tail distributions, so this heaviness is measured by the kurtosis coefficient. Therefore, this research shows that all kurtosis coefficients were greater than the standard normal
distribution of 3, so this defines all series as leptokurtic. When the kurtosis coefficient is 3, there is a normal distribution within financial econometrics of benchmarked distribution. Moreover, to test normal levels of series distribution, the Bera-Jarque test statistic was applied, where 1% significance for all constituents is strongly significant, but at 1% significance level, normality is rejected.
Table 5. 3: Statistical summary of SSM, DSM and KSM
Statistics Saudi stock market Dubai stock market Kuwait stock market reform original series Pre-reform Post- reform original series Pre- reform Post- reform original series Pre- reform Post- reform observations 2935 1445 1488 3062 1559 1503 2953 1471 1482 mean -0.000044 -.00014 .000053 .0001108 -.00027 .0005141 -.000038 .0000515 -.00012 Std. Dev. 0.017 .0213 .0117 .018 .0211 .0152 .0075 .0089 .0059 Variance 0.00029 .00045 .00013 .0003429 .00044 .0002333 .0000577 .000080 .000034 Skewness .97 -.87 -.82 .015 .05 -.013 -.60 -.511 -.84 Kurtosis 11.48 8.49 13.05 8.420357 6.95 10.41 7.75 6.31 8.81 Bera-Jarque 808.8 319.94 403.27 364.89 148.42 228.41 474.84 177.32 327.91 Source: Calculations of researcher
Table 5-3 summarises the descriptive statistics daily general index return for SSM, DSM and KSM. The results are presented in three periods based on official reforms dates. whole period from 2005- 2016, pre-reform 2005-2010 and post-reforms periods from 2011-2016 respectively.
For whole period that from 2005-2016, the daily Mean returns for the three countries are almost the same; only Dubai stock market have appositive return at .0001108. For the first half from 2005-2011, the descriptive statistics displays that the mean daily return are (- .00014) for Saudi stock market and (-.00027) for DSM, which indicated to negative return and high level of the risk and volatility. In addition, SSM has negative skewness of ( -.8762074) and a positive kurtosis of 8.49 indicating a leptokurtic distribution. daily mean returns for the two periods (pre -post Reform) show increase during the post-Reforms period when compared with post-reform for all market only KSM, there is just a slight decline during the post-reforms period from .0000515 to (-.00012).
The standard deviation which is the unconditional variance in returns, Dubai stock market had the highest risk and volatility at .0185 then Saudi stock market at 0.017 and followed by Kuwait at .0075. Overall, Standard deviation shows a decline in the post-reforms period for all markets when compared with the post reform period. For example, SSM appears low
during the post-Reforms period at .0117 when compared with .0213 during the pre-reforms period.
The kurtosis and the Bera-Jarque tests show that there is non-normal distribution of stock market returns from this financial time series and conforms to the general expectation, because higher departure from normality is shown when there is a higher frequency at which data are sampled. For example, the kurtosis is quite above three in the all markets, obviously indicating a leptokurtic distribution. However, the skewness is negative in both SSM and KSM, but not in DSM. In all-time series, their values of skewness differ from zero. In addition, The Jarque-Bera is well above the critical value with two degrees of freedom at 1%,5%,10% level of significance which is a rejection of the null hypothesis of normal distribution. Therefore, The Jarque-Bera has a high value indicating the rejection of the null hypothesis of a normal distribution for all markets and all different period.
As result, skewness, kurtosis and Jarque-Bera statistics do not follow a normal distribution. Therefore, this supports the view that there is non-normal distribution within equity market returns, and each sub-sample and the main sample, there was mostly a negative skew for the daily return, which indicates that the level of extreme negative returns has a tendency to be more than those of extreme positive returns.
Table 5. 4: Descriptive statistics of SABIC, STC, NCCY.
Statistics SABIC STC NCCY
reform Original Series Pre-reform Post- reform Original Series Pre- reform Post- reform Original Series Pre- reform Post- reform observations 3059 1562 1497 3059 1562 1497 3045 1548 1497 mean .0000114 .000109 -.000090 -.000087 -.000515 .0003559 .0003511 .0002356 .0004705 Std. Dev. .0240 .0295362 .016472 .0186791 .022021 .0143844 .0272602 .0301689 .0238925 Variance .000578 .0008724 .0002713 .0003489 .000485 .0002069 .0007431 .0009102 .0005709 Skewness -.2953 -.3098406 -.054229 -.2668 -.357923 .2712579 -.1635 -.063477 -.3533 Kurtosis 7.653 5.618112 10.46099 10.19559 8.405774 10.89269 6.679453 5.90868 7.505827 Bera-Jarque 366. 228.77 124.85 479.73 221.62 254.27 278.31 103.57 196.72 Source: Calculations of researcher
Table 5. 5: Descriptive statistics of Al Rajah Bank and Electricity Company
Statistics AL RAJHI BANK ELECTRICITY COMPANY
reform Original series Pre-reform Post-
reform Original Series Pre-reform Post-reform
observations 3059 1562 1497 3059 1562 1497 mean .0000425 .0002076 -.0001298 -.0000555 -.0004122 .0003166 Std. Dev. .0208252 .0259614 .013534 .021754 .0267321 .0148802 Variance .0004337 .000674 .0001832 .0004732 .0007146 .0002214 Skewness -.1125466 -.1228592 -.0357856 .0084393 -.0560104 .55997 Kurtosis 9.415441 6.973051 9.450149 9.272922 6.733194 13.32826 Bera-Jarque 417.8 152.40 207.25 405.0 141.51 346.20
Source: Calculations of researcher
Table 5.4 and 5.5 summarises the descriptive statistics daily general index return for SABIC, STC, NCCY, Al Rajah Bank and Electricity Company. The results are presented in three periods based on official reforms dates. That is: whole period from 2005- 2016, pre-reform 2005-2010 and post-reforms periods from 2011-2016 respectively.
During the whole period from 2005-2016, One of longest established company in this market is Al Rajhi Bank that is shown to be .0004732 and reflects the smallest variance. However, the NCCY Company was launched in the SSM in 2005, which could explain why it is shown to be (.0007431) between 2005 and 2016 and reflects the highest variance. Moreover, variance shows an increase in the post-reforms period for NCCY when compared with the post reform period from (.0301689) to (.0238925). These findings indicate that markets improve over time to become efficient.
The daily indicators on the table are quite improved when compared between the two periods (Pre-post). Therefore, the reforms have the same effect on most of the individual companies. For example, the daily mean returns for most of the companies have increased, with just a slight decline during the post-reform period for SABIC and Al Alrajhi Bank. In addition, the standard deviation declined for all companies during the post-reforms period when compared with pre-reform period. For example, Electricity Company decreases from (.0267321) to (.0148802) in the post-refrom period, which indicates lower volatility in the post-reform. Moreover, the skewness is negative in pre-reforms periods for all companies, depicted by left skewed distribution. When large negative returns are generated by assets it is indicated by a negative skewness. Overall, in the three periods, asymmetry is shown, and their values of skewness differ from zero. Moreover, kurtosis is quite above three in the three periods clearly indicating a leptokurtic distribution. The Jarque-Bera is well above the critical value with two degrees of freedom at 1%, 5%, 10% level of significance which is a rejection of the null hypothesis of normal distribution. As result, this supports the view that there is non-normal distribution with stock market return.
Conclusion, the result for the general index and the individual companies are summarized as bellow:
1. Non-normal distributions for all datasets that is confirmed by skewness, kurtosis and Jarque-Bera statistics all indicate a non-normality distribution.
2. Reform that occurred from 2005-2011 had positive effect on all general markets and the individual companies and improved the market efficiency because there was higher degree of normality distribution than pre-reform. In addition, the skewness of most of the general index and most of the companies has shifted to have a positive skewness coefficient in post-period, which indicate lower risk for investors when large positive returns are generated by asset indicated by a positive skewness. Also the stander deviation and the variance are lower in post-reform when compared with pre- reforms period.