STYLIZED FACTS OF THE STATISTICAL PROPERTIES OF RISK AND RETURN OF THE DHAKA STOCK EXCHANGE: Siban Shahana Kazi Iqbal Md.

Transcription

1 STYLIZED FACTS OF THE STATISTICAL PROPERTIES OF RISK AND RETURN OF THE DHAKA STOCK EXCHANGE: Siban Shahana Kazi Iqbal Md. Iqbal Hossain 6 th Dec 2017

2 RATIONALE FOR THE STUDY The 7 th Five Year Plan: GDP growth rate of 8% by 2020 Key question: how this growth will be financed Major share of financing coming from the private sector Domestic Inv. (% of GDP) FY 13 FY14 FY15 FY16 Pvt. Inv Pub. Inv What would be the distribution of equity and non-equity financing

3 RATIONALE FOR THE STUDY The role of equity finance increases with the level of economic development The economy of Bangladesh is likely to experience greater share of equity capital The role of the security market is mostly absent in the development discourse of Bangladesh Ignorance of small investors The current literature is mostly outdated Fail to answer the most fundamental first order questions

4 RESEARCH QUESTIONS We ask very basic questions regarding STOCK RETURN and its VOLATILITY The specific questions are- What are the historical average of the rate of return and volatility of the stock market ( )? How the return and volatility have changed over time? How return and volatility vary across sectors, age, quality of firms?

5 LITERATURE REVIEW Literature on historical return and volatility is very thin for developing countries The literature on stock market of Bangladesh is outdated and deals with a short span of time Chowdhury (1994) Basher et. al(2007) The efficient market or random walk hypothesis does not strictly hold in this period Does it hold for longer period?

6 CONTRIBUTION OF OUR STUDY The earlier studies dealt with small sample period and choice of study period is very arbitrary Dated studies focused only on the aggregate indices, ignoring the presence of high degree of heterogeneity among firms Firm level data for the years , containing a few million observations, allows to address the issue of heterogeneity across firms in return, volatility and their relationship

7 DATA We compile daily stock market data for all listed securities from Dhaka Stock Exchange and the annual repots of all listed firms on- closing price of the day number of trade volume of trade and total number of shares issued by the firm for the period

8 RESEARCH METHODOLOGY Calculate continuously compounded rate of return: Rate of return = 100*[log(closing price (t)) - log(closing price (t-1])] Calculate daily, weekly, monthly, and yearly returns and volatility for full sample as well as for each decade since 1991 Sector wise analysis-divide 21 categories of securities in 3 broad groups- Financial (6) Manufacturing (10) Service (5) Bank Tannery Engineering Telecom Life insurance Ceramic Cement IT General insurance Pharmaceuticals Fuel and Energy Service and real estate NBFI Food Miscellaneous Paper and printing Mutual Fund Jute Travel and leisure Bond Textile

9 RESEARCH METHODOLOGY Calculate the volatility(standard deviation) of stock return for each decade, and types of firms Testing for the efficient market hypothesis Constructing an autoregressive model and test for autocorrelation Runs test Investigate the risk-return relationship for DSE index as well as for disaggregated indices (e.g., sector, size, etc.) For this purpose we will use GARCH-M model-

10 RESEARCH METHODOLOGY It has the following specification: δ t 2 = α 0 + R t = μ t + λδ t + e t p i=1 2 α i e t i e t = δ t. t q + i=1 2 β i δ t j A significant and positive coefficient λ implies that investors are compensated with higher returns for bearing higher levels of risk A significant negative coefficient indicates that investors are penalized for bearing risk

11 RETURN Author Time Frame Daily return SD Chowdhury (1994) % 0.029%. This study % 1.35% Basher et al. (2007) % 1.55%. This study % 2.0% This study % 2.1% % 2.08% (excluding 1996 & 2010-the years of upsurge) 1996 (1st upsurge) 0.17% 3.05% 1997 (following year after the 1 st upsurge when market came back to its usual level) -0.21% 2.06% 2010 (2 nd upsurge) 0.12% 2.24% 2011 (following year after the 2nd upsurge when market came back to its usual level) -0.15% 2.88%

12 DECADE WISE RETURN Index return Mean SD Annual return

13 STYLIZED FACTS: RETURN The unweighted daily rate of return is about % which is equivalent to about 1.25% per annum for the period The 2000s was decade of high return(daily return 0.04%,14.6% annually) while the last 5 years have been the worst(-0.06%, -22% annually) Daily rate of return SD of daily rate of return

14 STYLIZED FACTS: DAILY ACROSS FIRM VARIATIONS Variations of daily return across shares remain more or less stable for the period , with being the period of high volatility (return low but daily volatility high) Monthly volatility of daily return 1.62% for the overall period. But, it increased over time(1.35% to 1.92%) Daily rate of return SD of daily rate of return Monthly volatility of daily return

15 STYLIZED FACTS: FIRM HETEROGENEITY The Manufacturing companies outperform other sectors Daily variations within a sector are very similar. Sector All securities Manufacturin g Service Financial Daily rate of return SD of daily rate of return Daily rate of return SD of daily rate of return Daily rate of return SD of daily rate of return

16 STYLIZED FACTS: NORMALILTY Leptokurtic data: fat tails Avg. Daily Log-Returns Density y Skewness 0.5 Kurtosis 20 JB Test

17 STYLIZED FACTS: PREDICTABILITY OF RETURN Returns don t follow random walk Daily return is predictable to some extent as AC and PAC suggest [-1 0 1] [-1 0 1] LAG AC PAC Q Prob >Q [Autocorrelati on] [Partial Autocor]

18 Average daily log return STYLIZED FACTS: MARKET EFFICIENCY Average daily log return jan jan jan jan jan jan2015 The diagnostic tests reject the assumption of random walk so the market is inefficient (Portmanteau test, runs test) N(res <= 0) 4892 Date N(res > 0) 4238 Portmanteau (Q) Statistic Prob > chi2(40) obs 9130 N(runs) 2884 z Prob>z 0

19 RESIDUAL RESIDUAL SQUARE Residual square jan jan jan jan jan jan2015 datevar 0 01jan jan jan jan jan jan2015 Date GRAPHICAL PLOT OF THE RESIDUAL SERIES

20 STYLIZED FACTS: RISK RETURN TRADOFF (1) (2) (3) VARIABLES Average daily ARCHM ARCH return L.arch 0.230*** ( ) L.garch 0.818*** ( ) Risk-return tradeoff coefficient *** ( ) Constant *** *** ( ) (7.34e-05) Observations 9,130 9,130 9,130

21 STYLIZED FACTS: RISK RETURN TRADOFF (1) (2) (3) VARIABLES DSE Index ARCHM ARCH L.arch 0.214*** L.garch 0.739*** Risk-return tradeoff coefficient * Constant ** 0.114*** Observations 8,398 8,398 8,398

22 STYLIZED FACT: VOLATILITY CLUSTERING The parameters of GARCH model for returns is positively significant at 1% level, implies to reject null hypothesis and accept the existence of volatility clustering There is an ARCH effect in the avg. log return, indicating that there is a direct effect between news that enters the market and the level of volatility

23 STYLIZED FACT: TRADE OFF BETWEEN RISK AND RETURN Overall, the coefficient of the risk-return parameter is positive (0.03 and in different specifications) and statistically significant suggesting that the investors are compensated with high return in times of high volatility

24 STYLIZED FACT: TRADE OFF BETWEEN RISK AND RETURN VARIABLE S Risk return trade-off (weekly return) Full sample Manufact uring Services Financial A category Firms *** *** *** *** *** B category Z category Not so Firms Firms Old firms new firms New firms *** *** *** *** *** Overall, one percentage point increase in return comes with 2.7 percentage point of risk (SD) Financial sector is the least risky sector Better the quality, lower the risk Older the security, higher the risk.

25 THANK YOU SUGGESTION PLEASE