Mahedi Masuduzzaman Senior Assistant Secretary, Ministry of Finance Finance Division, Macroeconomic Wing

Transcription

1 Impac of Macroeconomic Variables on he Sock Marke Reurns in Bangladesh: Does a Meaningful Impac Exis? Mahedi Masuduzzaman Senior Assisan Secreary, Minisry of Finance Finance Division, Macroeconomic Wing Absrac: This paper srives o invesigae he long-run relaionship and he shor-run dynamics beween Bangladesh sock marke index (DGEN -Key marke-racking index of Dhaka Sock Exchange) and key macroeconomic variables such as Consumer Price Index (CPI), Exchange rae of BDT agains USD (Exrae), Broad money supply (M2), Indusrial Producion (IP) and Ineres rae (Inrae). This sudy analyse monhly daa for he above variables beween he periods spanning from July 2006 o Ocober 2012 and employ differen economeric ools. The Engle-Granger and bivariae Johansen co-inegraion ess produces nonexisence of long erm relaionship. However, Johansen mulivariae co-inegraion ess indicae ha he Bangladesh sock marke index and chosen five macroeconomic variables are co-inegraed, his is indicaive of a long-run relaionship. Alhough, here is a long erm relaionship, i is only IP, CPI and M2 ha adjus any disequilibrium once he sysem is shocked. Bu, he oal adjusmen power is very low in he sysem. This sudy finds no shor-run causal relaionship, which furher srenghened by he Impulse Response Funcion(IRF) and Variance Decomposiion(VDC) ess. IRF indicaes, shock o macroeconomic variables does no generae a significan response o Bangladesh sock marke. The VDC esablished ha, a leading proporion of he variabiliy in he sock index was explained by is own innovaions while only a marginal was explained by he seleced variables. These resuls recognized, sock prices in Bangladesh canno be prediced using macroeconomic variables. The findings of no shor-run dynamic and very weak long run relaionship in his sudy help asue invesors and policy makers in efficien and appropriae decision making regarding Bangladesh sock markes. Key words: Macroeconomic variables, Invesor, Bangladesh, Reurns, Sock marke, causaliy, impulse response, variance decomposiion. JEL Classificaion: C22, E44, G15 1

2 I. Inroducion The impac of macroeconomic variables on sock marke reurns has generaed a lo of ineres in he financial marke lieraure. The relaionship beween sock marke reurns and macroeconomic variables has been widely invesigaed, especially in developed markes. The early sudies on he US sock markes by Linner (1973), Bodie (1976), Jaffe and Mandelker (1977) and Fama and Schwer (1977) mainly examined wheher he financial asses were hedges agains inflaion. These sudies have repored a negaive relaion beween sock reurns and changes in he general price level. However, Fama (1981) found a direc posiive relaionship beween sock marke reurns and real economic aciviies such as indusrial producion. Chen e. al. (1986), esed wheher a se of macro-economic variables explained unexpeced changes in sock marke reurns. I is recognized ha sock markes play a pivoal role in growing indusries and commerce of a counry ha evenually affec he economy. The imporance of he sock markes has been well acknowledged in policy makers, porfolio managers, indusries and invesors perspecives. The sock marke avail long-erm capial o he lised firms by collecing funds from various poenial invesors, which allow hem o expand in business and also offers invesors alernaive invesmen avenues o pu heir surplus funds in (Naik and Padhi, 2012). I is very ineresing o inves in sock marke bu also a very risky rench of invesmen. So, poenial invesors always ry o guess he movemen of sock marke prices o achieve maximum benefis and minimize he fuure risks. By concerning wih he relaionship beween sock marke reurns and macroeconomic variables, invesors migh guess how sock marke behaved if macroeconomic indicaors such as exchange rae, indusrial producions, ineres rae, consumer price index and money supply flucuae (Hussainey and Ngoc, 2009). Macroeconomic indicaors such a composiions of daa which frequenly used by he policy makers and invesors o gahering knowledge for curren and upcoming invesmen prioriy (Masuduzzaman, 2012). The issue wheher he sock marke performance leads or follows economic aciviy is now becoming very conroversial in Bangladesh as he sock marke has gained much aracion in he las few years. Almos all he indicaors such as marke capializaion, rading volume, urnover and he marke index have shown remendous growh, alhough i has volailiy. In his end, how does and a wha exen he sock marke reurns of Bangladesh respond o he changes in macroeconomic deerminans remains an open empirical quesion. Undersanding he main macroeconomic variables, which may impac he Bangladesh sock marke index, wih he recen daa could be helpful for policy makers, invesors and all oher sakeholders. The presen sudy herefore, aemps o explore wheher here are long-run and shor-run dynamic ineracions beween Bangladesh sock marke index (DGEN-Key marke-racking index of Dhaka Sock Exchange -DSE) and key macroeconomic variables namely, Consumer Price Index (CPI), Exch ange rae of BDT agains USD (Exrae), Broad money supply (M2), Indusrial Producion (IP) and Ineres rae (Inrae) for Bangladesh, by using uni roo, Engle-Granger, Johansen co-inegraion, error correcion model (ECM), Granger causa liy, Variance Decomposiion (VDC) and Impulse Response Funcion (IRF). The auhor finds no sudy ha paricularly invesigaed he associaion beween sock price of Bangladesh and he leading macroeconomic deerminans by employing Impulse Response Funcion (IRF) and Variance Decomposiion (VDC). So his will creaes exra imporance in he case of Bangladesh sock markes. The secion one of his sudy is he inroducory par. The res of he sudy is srucured in five secions. The second secion of he sudy will presen an overview of relaed lieraures ha will give us a sound concepion of he facs and secion hree gives an overview on developmen of Bangladesh sock markes. The secion four provides an avenue regarding he research mehodological approach and he relevan informaion on he ime series daa ses ha are used for his sudy, while secion five discussed he empirical resuls. Finally, secion six will provide he conclusion ha will poin ou he possible recommendaions of he sudy as well. 2

3 II. Review of Previous Empirical Sudies The heoreical consruc linking he impac of macroeconomic volailiy on sock reurns is based on he ransmission mechanism of leading macroeconomic variables, namely; inflaion, exchange rae, ineres rae, money supply and indusrial producion (Smih and Sims, 1993; Flannery and Proopapadakis, 2002). Using US monhly sock reurns and inflaion ime spanning from 1953 o 1974, Nelson (1976) found a negaive relaionship beween sock reurns, in boh expeced and unexpeced inflaion. Furhermore, empirical ess on he response of sock reurns o inflaion in he 1980s by Fama (1981), and Solnik (1983), amongs ohers, also yielded similar resuls of a negaive relaionship. Mukherjee and Naka (1995) invesigaed he relaionship beween sock marke reurns and he main macroeconomic variables ( exchange rae, inflaion, money supply, indusrial producion index, he long-erm governmen bond rae and call money rae) in Japan. Their resul confirms ha, he reurn of Japanese sock marke was closely co-inegraed wih he leading macroeconomic variables. This indicaes a long-run equilibrium relaionship beween he sock marke reurn and he macroeconomic variables. Nasseh and Srauss (2000) did heir analysis on macroeconomic deerminans for Germany, UK and oher indusrialized counries in Europe. They used quarerly daa during he period of o and heir findings suggesed ha consumer price index (CPI) and indusrial producion exis wih large posiive coefficiens in he said counries sock markes. They also found negaive coefficiens on long erm ineres raes. This sudy also argue ha German indusrial producion and sock prices also posiively influence on he reurn of oher European sock markes like Holland, France, Ialy, Swizerland and he UK. Ibrahim and Aziz (2003) esimaed vecor auo-regression model o explore he dynamic linkages beween sock prices and four macroeconomic variables for he case of Malaysia. Empirical resuls of he analysis suggesed he presence of a long-run relaionship beween hese variables and he sock prices and subsanial shor-run ineracions among hem. They also saed ha, he sock marke is playing somewha predicive role for he macroeconomic variables. Gunasekarage e al. (2004) examined he causal nexus beween sock prices and five macroeconomic variables in Sri Lanka, by employing a vecor error correcion model. The resuls showed lagged values of consumer price index, M1 and Treasury bill rae have powerful influences on he sock marke reurns. Boh VDC and IRF analyses revealed ha shocks o macroeconomic facors explained only a minoriy of he forecas variance error of he marke index. Abugri (2008) finds ha Chile, Argenina, Brazil and Mexico sock markes reurns are changed by individual macroeconomic facor such as indusrial producion, exchange rae, money supply ec as well as he US hree monh T-bill yield. Kizys and Pierdzioch (2009) race ou ha inernaional co movemen of monhly equiy reurns in indusrialized economies is no sysemaically relaed o macroeconomic shock. Macroeconomic variables influence no only domesic sock marke bu also inernaional sock marke. Mial and Pal (2011) finds ha exchange rae shows significance influence in Bombay Sock Exchange (India). The sudy conclusively argues ha sock marke indices are dependen on macroeconomic variables. In he Bangladesh conex here has been very few sudy in our area of ineres. The auhor finds no sudy ha paricularly examined he relaionship beween sock price of Bangladesh and he macroeconomic deerminans by employing Variance Decomposiion (VDC) and Impulse Response Funcion (IRF). However, Rahman and Uddin (2009) have invesigaed he ineracions beween sock prices and exchange raes in hree emerging counries of Souh Asia namely, Bangladesh, India and Pakisan for he period beween January 2003 and June This sudy claimed ha here is no co-inegraing relaionship beween sock prices and exchange raes. Applying Granger causaliy es, hey find ou no causal relaionship beween sock prices and exchange raes. Afzal and Hossain (2011) examined he relaionship beween sock prices and selecive macroeconomic variables using co-inegraion es and Granger causaliy es for he ime spanning from July 2003 o Ocober The resuls of his sudy sae ha co-inegraion exiss beween sock prices wih M1, M2 and inflaion rae ha implies he long-run relaionship. This sudy also esablished unidirecional causaliy running from sock marke index o exchange rae and M1 in he shor-run. Using daa se for he period from January 1987 o December 2010, Ali (2011) have examined he direcion of he causal relaionship beween sock prices and cerain macroeconomic variables in Dhaka sock Exchange(DSE) by applying co-inegraion and Granger causaliy es. Empirical resuls of he analysis suggesed he sock prices do no granger cause CPI, deposi ineres rae, expor receip, GDP, invesmen, indusrial producion index, lending ineres rae and naional 3

4 income deflaor. Bu unidirecional causaliy is found from DSI o broad money supply and oal domesic credi. In addiion, bi-direcional causaliy is also idenified from DSI o exchange rae, impor paymen and foreign remiances. III. Brief Overview on Bangladesh Sock Exchange The recen changing and growing Bangladesh sock markes are consised of Dhaka Sock Exchange (DSE) and Chiagong Sock Exchange (CSE). The DSE formal rade began in 1956 wih paid up capial of approximaely 04 billion of Bangladesh Taka (BDT). DSE is a public limied company which is officially regulaed by Securiies and Exchanges Commission (SEC) Ac There is anoher sock marke operaing righ now namely, Chiagong sock exchange ha began formally rade in Dhaka Sock Exchange is larges and mos acive sock marke in Bangladesh accouning for around 90 per cen of he value of he counry s oal sock ransacion (Turnover US $48 million) as on Ocober 23,2012 wih marke capializaion US$ billion (DSE,2012). Acually, DSE is he represenaion of Bangladesh sock marke scenario. Afer he financial liberalizaion in Bangladesh since 1990 has open up he marke for he inernaional invesors and he marke expanded rapidly ha reflecs significan economic growh. DSE General Index (DGEN ) is a key markeracking index ha reflecs he real scenario of Dhaka Sock Exchange (DSE) performance and Bangladesh sock marke performance as well. The figure-1 provides an idea abou he share price performance of DGEN during he period considered. The abnormal behaviour observed in share price in beween lae 2010 and he early 2011, made he reurns volaile. Figure-1: Monhly Share Price Index and Reurns of Dhaka Sock Exchange Dhaka Sock Exchange Index (DGEN) 9,000 8,000 7,000 6,000 5,000 4,000 3,000 2,000 1,000 III IV I II III IV I II III IV I II III IV I II III IV I II III IV I II Monhly Sock Reurns III IV I II III IV I II III IV I II III IV I II III IV I II III IV I II Source: Auhor s calculaion based on DSE daa The Dhaka Sock Exchange (DSE) has experienced rapid growh in erms of marke capializaion, urnover and share of marke capializaion o Gross Domesic Produc (GDP) in he recen pas. According o he (Agarwal, 2001), marke capializaion is posiively correlaed wih he abiliy o mobilize capial and diversify risk on an economy wide basis. The figure-2 reveals ha he marke capializaion of DSE is relaively low compared o 4

5 he GDP of he counry, alhough i s increased rapidly since June Increasing rend of marke capializaion indicaes ha resrucuring and effors of he privaizaion are well under way (Arayssi e al ). Figure- 2 also depics he urnover of he sock marke of Bangladesh. From June 2000 o June 2012, i increased form 23.8 crore BDT o crore BDT. The daily urnover peaked a BDT in June 2010, bu showing a declining rend aferwards. Graph-2 also exhibis, marke capializaion (% of GDP) has been significanly increased over ime bu he growh of he lised companies is no increased significanly raher i has remained sagnan during he period beween June, 2000 and June, Figure-2: Performance of he DSE Source: Auhor s calculaion based on Bangladesh Bank daa. a) Daa Descripion IV. Daa Descripion and Mehodology In he empirical analysis we use daa from July 2006 o Ocober We choose monhly frequency o maximize he number of observaions for a robus esimaion of he model. The daa on Bangladesh sock marke index (DGEN-Key marke-racking index of DSE), Money supply (M2), Exchange rae of BDT agains USD (Exrae) and Ineres rae (Inrae) has been obained from Bangladesh Bank (Cenral Bank of Bangladesh). The daa on Indusrial producion (IP) index and Consumer Price Index (CPI) has been sourced from Bangladesh Bureau of Saisics (BBS). All he variables, wih only excepion of Ineres rae are expressed in naural logarihm form. To provide an overall undersanding of he chosen sock marke index and he macroeconomic variables, we include summary saisics of he daa se in able-1. The sandard deviaion ha measures volailiy is very high for ineres rae and sock marke index compared o oher four variables. Skewness of he sock marke index is negaive implies ha he disribuion has a long righ ail. The Kurosis value of sock index, indusrial producion, CPI and M2 relaively small implying ha he disribuions are low relaive o normal. The value of skewness and kurosis indicae he lack of symmeric in he disribuion. The significan coefficien of Jarque- Bera saisics of exchange rae, ineres rae and M2 also indicaes ha he frequency disribuions of considered series are no normal. 5

6 Table-1: Summary Saisics of he chosen variables GENINDEX IP CPI EXRATE INTRATE M2 Mean Median Maximum Minimum Sd. Dev Skewness Kurosis Jarque-Bera Probabiliy bservaions Source: Auhor s calculaion based on BB and BBS daa The following able-2 shows ha, Bangladesh sock marke have posiive average reurns. The highes average rae of reurns was 1.54 per cen. The highes monhly reurns around 27 per cen and he minimum reurn negaive 36 per cen during he sample period. The sandard deviaion ha measures he volailiy of he sock marke reurns is very high during he sample periods. Table-2: Sock reurns of differen sample periods Mean Maximum Minimum Sd. Dev. Sock Reurns Noes: Reurns have been calculaed as follows, r ln( p / p 1) 100 b) Mehodology The Uni Roo Tes The ime series daa are non-saionary ha always leads o spurious resuls. Therefore, i is he prerequisie o perform he uni roo es before economeric analysis of any ime series. Apar from ha, according o Engle and Granger (1987), a long-run relaionship could only exis when he variables of ineres are inegraed o he same order. There are several ess in he lieraure o examine he uni roo in ime series variables and each es has is own advanages and disadvanages. The Augmened Dickey-Fuller (ADF) es is widely used for es saionariy of he variables (Dickey, Fuller, 1979, 1981). Phillips and Perron (1988) proposed a modificaion of he Dickey-Fuller (DF) es and have developed a comprehensive heory of uni roos. In his consideraion, we adop he Phillips Perron mehodology o es uni roos in he chosen six variables. The es is conduced a individual variables in level log form and he firs differenced log form. If he log forms or firs differenced log forms rejec he null hypohesis (H 0 : series has a uni roo), he ime series is saionary. The PP es is based on he following model Y 1... (1) by Where = firs difference operaor, = consan, = error erm. Co-inegraion Tes and Error Correcion Mechanism (ECM) Afer confirming wheher he six variables used in his sudy are inegraed of he same order, we have o conduc co-inegraion es. For robusness of he co-inegraion es, his sudy employed boh he Engle and Granger (1987) and he Johansen and Juselius (1990) approaches. According o Lukepohl (1982), i is imporan o noe ha, he bivariae co-inegraion es or regression may be counerfei due o he omission of he imporan variables. In his consideraion, his sudy employed boh bivariae and mulivariae procedure. The primary sep for Engle-Granger es is o run he Ordinary Leas Square (OLS) regression and hen he second sep is o predic i s residual. If he residual is saionary hen we can say ha he sock marke index and he macroeconomic variables are inegraed. To move ino he firs sep we consider he following model, 6

7 y 1 0 x e (2) Where y = Sock marke index of Bangladesh and x represens he chosen macroeconomic variables. We address he residual e by using he following Augmened Dickey Fuller (ADF) uni roo es. n e ˆ e ˆ ie ˆ 1 1 i i2... (3) In equaion (3) α represens he esimaed parameers, while is he error erms. To deerminaion of long run relaionship, we conduced hypohesis es on he coefficiens of. If he es saisic of he coefficiens exceeds he criical value, hen we could rejec he null hypohesis and conclude ha he residual is saionary. The rejecion of null hypohesis implies he linkages of long run relaionship among he variables. Johansen mehod indicaes he maximum likelihood procedure o he idenificaion of exisence of coinegraing vecors for chosen non-saionary ime series. This model direcly invesigaes he inegraion in Vecor Auo-regression (VAR). The primary sep in he Johansen co-inegraion es is o deermine he opimal lag lengh for each VAR model. This sudy idenified he opimal number of Lags by using Schwarz Crierion (SC) and considered he minimized crierion value. The Johansen s mulivariae co-inegraion es is based on he following vecor auo regression equaion: z c A z A z p p... (4) Where z represens n 1 vecor ha inegraed I (1) and be re-wrie following way is n 1-vecor innovaions. The VAR model (4) can z c z p z 1 1 i i (5) i1 Where p i1 A i I and p i A j ji1 If he rank r<n in he co-efficien marix of above equaion, here is a possibiliy of exising n r marices namely α and β each wih rank r and i can be wrien in he following way ' =α (6) Where α=n 1 column vecor ha represens he speed of shor-run adjusmen o disequilibrium. On he oher ' hand, = 1 n co-inegraing row vecor, represens he long run coefficien marix. There are wo differen likelihood raio es proposed by he Johansen namely, Trace Tes= race k = -T r j 1 ln(1- ) (7) ˆ j Maximum Eigen Value Tes= = -T ln(1- r 1 ) (8) Where T= Sample size and max ˆ ˆ j = Esimaed values of characerisic roos ranked from larges o smalles. 7

8 The equaion (7 ) ha is: race es ( race ) ess he null hypohesis of co-inegraing vecor agains he alernaive hypohesis of n co-inegraing vecors and equaion (8) ha is: maximum Eigen value es ( ess he null of r co-inegraing vecors agains he alernaive hypohesis of r+1 co-inegraing vecors. max ) Afer confirming wheher he six variables used in his sudy are co-inegraed; ha is, here is a long erm or equilibrium, relaionship beween sock prices and he macroeconomic variables, we can move in o examine ECM. Of course, in he shor-run here may be disequilibrium. Therefore, one can rea he error erm of any regression model as he equilibrium error. This is a means of reconciling he shor-run behaviour of economic variables wih is long-run behaviour(gujrai and Porer-2009). The Engle-Granger represenaion heorem Engle and Granger (1987) saes ha if wo series are co-inegraed, he Error Correcion Mechanism (ECM) can inerpre he dynamic changes in he shor-run and can include variaions in he parial adjusmen mode. For he purpose of capuring he long-run behaviour an Error Correcion Term (ECT) which is lag of residuals generaed from he model esimaed in level is included in shor-run equaion (Mial and Pal -2011). The ECM specificaion is represened below: LogDGEN LogCPI LogIP LogExrae LogM 2 Log in rae ECT (9) We also conduced he causaliy es based on Granger s (1969) approach in order o see any influence beween sock marke index and he macroeconomic variables considered in his sudy. If we consider x (sock marke index) and way: y (macroeconomic variables) as wo differen ime series hen he model express as following m n x 0 x i y i1.. (10) m n y y i 1 2x i (11) Where is he difference operaor, n and m are he lag lenghs of he variables. 1 and 2 are he disurbance erms. In he firs equaion, i is assumed ha he values of variable X are a funcion of is lagged values and he lagged values of variable Y. In he second equaion i is assumed ha he presen values of variable Y are a funcion of is lagged values and he lagged values of variable X. The null hypohesis is ha X does no Granger-cause Y in he second model and ha Y does no Granger-cause X in he firs model. Variance Decomposiion (VDC) and Impulse Response Funcion (IRF) The sandard Granger causaliy analysis inerpreed wihin he sample period only. In his regard, variance decomposiion analysis could be an imporan ool o make proper inference regarding he causal relaionships beyond he sample period. Acually, Variance Decomposiion indicaes he percenage of he forecas error variance in one variable ha is due o errors in forecasing iself and each of he oher variables (Tarik, 2001). The impulse response funcion is designed o infer how each variable responds a differen ime horizon o an earlier shock in ha paricular variable and o shocks in oher macroeconomic variables. Paricularly, we invesigae he response of he DGEN(Sock marke index of Dhaka Sock Exchange) o one sandard deviaion shocks o he equaion for DGEN and macroeconomic variables and also he he response of macroeconomic variables o one sandard deviaion o he equaion for he DGEN. This sudy use Cholesky decomposiion o make error erm orhogonal. A wo-sandard-deviaion confidence inerval will be repored for each impulse response funcion. I is noed ha, if a confidence inerval conaining zero, implies lack of significance and he said confidence inerval for IRF will be compued from Mone Carlo simulaions. As VAR resuls are very much sensiive o he choice of appropriae lag lengh, his sudy idenified he opimal number of Lags by using Akaike Informaion Crieria (AIC) and considered he minimized crierion value o derive VDC and IRF from he vecor error correcion model. 8

9 V. Empirical Resuls The uni-roo es is performed on Bangladesh sock marke ime series and he chosen macroeconomic variables o deermine wheher he ime series is saionary. We employed PP uni roo ess. The resuls of he uni-roo es are shown in Table-3. According o he resuls none of he variable is saionary a heir naural log level a sandard 05 per cen level of significance. Bu null hypohesis ( H 0 : Non-saionary) are rejeced in heir log firs differences a 01 per cen level of significance. This implies ha all he variables are saionary in he log firs difference ha is I (1). Therefore, i is possible o proceed o he second sep of he analysis, esing for coinegraion beween sock marke prices and macroeconomic variables. Table 3: Phillips-Peron uni roo ess wih Bangladesh s Sock marke indices and oher macroeconomic variables Variables Inercep Inercep and Trend Level Firs Difference Level Firs difference T-saisics P-value T-saisics P-value T-saisics P-value T-saisics P-value DGEN * 0.00 CPI * 0.00 IP * 0.00 Exrae * 0.00 M * 0.00 inres * 0.00 Noes: The criical values and deails of he es are presened in Phillips and Perron (1988). *indicaes significan a 1%. Resuls of Engle-Granger co-inegraion es are repored in able-4. These resuls are obained by employing ADF es for co-inegraion on he residuals generaed from he esimaed model (3) in naural log levels. The resuls indicae ha he null hypohesis of no co-inegraion is no rejeced in he esimaed models where DSE general index (DGEN) is dependen variable and CPI, IP, Exchange raes, M2 and ineres rae are individually explanaory variables. This esablished he fac ha when he marke forces are reacing acively, his macroeconomic variable does no affec he Bangladesh sock marke in he long run. Furhermore, when DSE general index (DGEN) is dependen variable and oher macroeconomic variables acs collecively as dependen variables in he model (3), he null hypohesis of no co-inegraion is no rejeced a sandard 05 per cen level of significance ha confirmed here is no long-run relaionship beween sock marke indices and he chosen macroeconomic variable. Table-4: ADF ess resuls on Engle-Granger co-inegraion es residuals ADF Tes ADF Tes ( rend and inercep) (wihou rend) Variables T-saisics P-value T-saisics P-value DGEN and M DGEN and CPI DGEN and Exrae DGEN and Inrae DGEN and IP DGEN and M2, CPI, Exrae, Inrae, IP As we menioned earlier ha, for robusness of Engle-Granger es we also used boh he bivariae and mulivariae Johansen co-inegraion es and consider boh he race saisic and maximum Eigen value saisic es. The primary sep in he Johansen co-inegraion es is o deermine he opimal lag lengh for each VAR 9

10 model. This sudy idenified he opimal number of Lag by using Schwarz Informaion Crierion (SIC) and considered he minimized crierion value. The empirical resul of he Bivariae Johansen co-inegraion es (Panel A of Table 5) do no suppor he presence of he co-inegraing vecor. The null hypohesis Sock marke index and he individual macroeconomic variables are no co-inegraed (r=0) agains he alernaive of one coinegraing vecor (r 1) is no rejeced because he race saisic ( ) value is no exceed he corresponding 05% criical value. The maximum Eigen value saisic ( max We have seen ha boh he race and maximum eigenvalue of mulivariae co-inegraion es recommend he presence of he long-erm relaionship exiss in he sysem. Hence, esimaing hem in a Vecor Error Correcion Model (VEC) is required. The resuls obained from ECM specificaion represened by model (9) are shown in able-6 wih common diagnosic ess. The adjused R 2 is very low, when sock index acs as a dependen variable, oher fairly high compare o sock marke index equaion. The lagged error correcion erms for he equaions are saisically no significan a sandard 05 per cen level excep IP, CPI and M2 bu he coefficien is very low ha is 0.027, and respecively. This implies ha if he sysem is exposed o a shock, i will be converged o he long run equilibrium a 2.7, 0.36 and 0.31 per cen per monh by IP, CPI and M2 10 race ) also produce same resul. We do no find any evidence of co-inegraion on a bivariae basis beween Sock marke indices and he macroeconomic variables. However, we wan o see wheher he sock marke index and he macroeconomic variables are co-inegraed as group. In his regard, his sudy also carried ou he mulivariae Johansen coinegraion es. The resul of he mulivariae es (Panel B of Table 5) implies ha long-erm relaionship exis beween sock marke indices and chosen macroeconomic variables because race saisic and maximum Eigen value saisic exceed he corresponding 05% criical value. Since he marke index and he chosen macroeconomic variables have a leas one co-inegraing vecor, i is reasonable o assume ha hey move ogeher in he long run equilibrium pah. This resul is suppored by Afzal and Hossain (2011), who esablished ha long-run relaionship exised beween socks prices and he macroeconomic variables in Bangladesh. The finding of his research is also on he line as Mial and Pal (2011) have esablished he long-run relaionship beween sock price and he cerain macroeconomic variables in India. Table 5: Resul of he Johansen Co-inegraion Tes (Wihou Trend) Panel A: Resul of Bivariae Johansen Co-inegraion Tes (Wihou Trend) Trace Saisic ( race ) 05% Criical Value Max Eigen Value Saisic( ) max 5% Criical Value r= DGEN and M2 r DGEN and CPI r=o r DGEN and Exrae r=o r DGEN and Inrae r=o r DGEN and IP r=o r Panel B: Resul of Mulivariae Johansen Tes (Wihou Trend) DGEN and M2, CPI, Exrae, Inrae, IP r=o * * r r r r r Noes: The 5% criical values provided by MacKinnon e al. (1999) indicae no co-inegraion. Noe: * Indicae significan a 05% level.

11 respecively, ha is oal only 3.37 per cen. So he adjusmen power is very low in he sysem. This is he indicaion of weak long run relaionship beween Bangladesh sock marke indices and he chosen macroeconomic variables. This implies ha Bangladesh sock marke and he chosen macroeconomic variables do no have srong co-inegraing relaionship in he long run. The resul (Table-6) also shows ha none of he variable shows a significan relaionship wih Bangladesh sock marke in he shor-run. I is noed ha, indusrial producion have shown an upward rend in he recen pas ( BBS, 2012), herefore i is expeced ha indusrial producion should posiively relaed o sock marke. Bu, he sudy reveals ha hese variables have insignifican relaionship wih sock marke reurns. To invesigae he possible endogenous relaionship beween Sock reurns and chosen macroeconomic variables, his sudy perform he Granger causaliy es o examine wheher macroeconomic developmen is encouraging he sock marke performances. The Granger-causaliy es was applied o firs differences of he DSE general index pairwise wih each of he five macroeconomic variables and he resuls are shown in Table- 7. Since his es is highly sensiive o he lag orders of he righ-hand-side variables, he Schwarz Informaion Crieria (AIC) was used o deermine he opimal lag lengh. The resuls sugges ha, here is no bidirecional Granger causaliy exising beween sock reurns and he chosen macroeconomic deerminans. Again, here is no causaliy in eiher direcion found beween he sock price and he macroeconomic variables. The Bangladesh sock marke price does no appear o be caused by he macroeconomic variables, nor does i have a significan influence on hem. Table-7: Pairwise Granger Causaliy Tes for Sock Marke Reurns and Macroeconomic Variables Null Hypohesis: F-Saisic Prob. 11

12 IP does no Granger Cause GENINDEX GENINDEX does no Granger Cause IP CPI does no Granger Cause GENINDEX GENINDEX does no Granger Cause CPI EXRATE does no Granger Cause GENINDEX GENINDEX does no Granger Cause EXRATE INTRATE does no Granger Cause GENINDEX GENINDEX does no Granger Cause INTRATE M2 does no Granger Cause GENINDEX GENINDEX does no Granger Cause M Impulse Response Funcion (IRF) The paern of he impulse response of Bangladesh sock marke o a shock in he chosen macroeconomic variables is examined, in order o obain addiional insigh ino he srucure of he sock marke linkages. Figure-3 simulaed for IRF, laeral axis denoes ime horizon and longiudinal axis denoes response degree for innovaion shock. Solid line (blue) denoes value of simulaion, dashed line (red) denoes confidence inerval of wo imes sandard deviaion. In figure-3 (Firs sample) i is possible o observe ha he response of DGEN o is own shocks is always posiive. The response of sock marke index o M2 is posiive bu insignifican and vice versa. The response of sock marke index o exchange rae and ineres rae is negaive bu insignifican. The response of IP and CPI is firs wo monhs is negaive hen i s posiive bu insignifican. The response exchange rae and ineres rae o of sock marke index is negaive, bu again i insignifican. These resuls again indicae ha sock prices and macroeconomic variables in Bangladesh do no have causal link. Only he responses of sock marke index o sock index have significan posiive impac up o six monhs. 12

13 Figure-3: Impulse Response Funcion for Bangladesh sock marke (Response o one S.D. Innovaions).2 Response of GENINDEX o GENINDEX.2 Response of GENINDEX o IP.2 Response of GENINDEX o CPI Response of GENINDEX o EXRATE Response of GENINDEX o INTRATE Response of GENINDEX o M Response of IP o GENINDEX Response of CPI o GENINDEX Response of EXRATE o GENINDEX Response of INTRATE o GENINDEX Response of M2 o GENINDEX Variance Decomposiion Analysis Afer discussing he findings for he impulse response funcions in he previous sub-secions, we now urn o he resuls for he variance decomposiion. The variance decomposiion analysis was employed o supplemen he Granger causaliy resuls o reinvesigae he ou of sample impac. The resul indicaes how much of he DGEN own shock is explained by movemens in is own variance and he chosen macroeconomic variables over he 12 monhs forecas horizon. According o he resuls, shown in figure 4, he amoun of variance of he DGEN explained by own goes down when he ime horizon increased up o 12 monhs. A horizon wo, 96 per cen of DGEN variance is explained by own. A horizon 12, 87 per cen of DGEN variance is explained by iself. This indicaes ha a longer horizons, he variance of DGEN may be caused by variance of oher macroeconomic facors. A horizon 6, he exchange raes explain 7 per cen of he variances of he DGEN. When he ime 13

14 horizon goes up, he acual amoun of variance of he DGEN explained by he exchange raes also goes up. This discloses ha a longer horizons, he variance of DGEN may be caused by variance of oher macroeconomic facors as he chosen macroeconomic variables especially do no have significan explanaory power. Therefore, i is esablished ha here is no shor run relaionship exiss beween Bangladesh sock marke and he chosen macroeconomic variables. However, here is very weak long run relaionship exiss. I is also esablished ha, here is no causal relaionship exiss beween Bangladesh sock marke and he chosen macroeconomic variables in he shor run. These resuls indicae ha sock prices in Bangladesh canno be prediced using macroeconomic variables. 14

15 VI. Conclusion and Policy Implicaions The paper has made an aemp owards examinaion of he impac of macro-economic variables on Bangladesh sock marke applying differen ime series echniques of Engle-Granger co-inegraion, Johansen co-inegraion, Granger- Causaliy, Error Correcion Model(ECM), Impulse Response Funcion (IRF) and Variance Decomposiion (VDC), in a sysem incorporaing he macroeconomic variables such as Consumer Price Index (CPI), Exchange rae of BDT agains USD (Exrae), Broad money supply(m2), Indusrial Producion (IP) and Ineres rae (Inrae) have been aken as explanaory variables and o represen sock index (DGEN) of Bangladesh sock marke is dependen variable beween he period from July 2006 o Ocober By applying ADF and PP es, we find all he six variables are non-saionary in he level and he firs differences are saionary. By employing Engle-Granger co-inegraion es we do no find evidence of coinegraion ha esablished nonexisence of long-run relaionship beween Bangladesh sock price and he chosen macroeconomic variables. We also employed he mos sophisicaed Johansen co-inegraion es for robusness of Granger co-inegraion es resuls. The bivariae Johansen co-inegraion es produces same resul like Engle-Granger ess. While, Johansen mulivariae co-inegraion ess indicae ha he Bangladesh sock marke index and chosen five macroeconomic variables are co-inegraed, his is indicaive of a long-run relaionship. Alhough here is a long erm relaionship among sock prices and he chosen variables, i is only IP, CPI and M2 ha adjus any disequilibrium once he sysem is shocked. Bu, of course, he oal adjusmen power is very low in he sysem. Therefore, error correcion erm does no suppor he long run relaionship srongly. The resuls from he ECM also indicae ha none of he variable shows a significan relaionship wih Bangladesh sock marke in he shor-run. I is noed ha, indusrial producion has shown an upward rend in he recen pas in Bangladesh ( BBS, 2012). Therefore, i is expeced ha indusrial producion supposed o be posiively relaed o sock marke. Bu, he sudy reveals ha hese variables have insignifican relaionship wih sock marke reurns. This implies ha here are many oher macroeconomic facors which affec he flucuaion in Bangladesh sock marke reurns. These resuls indicae ha sock prices in Bangladesh canno be prediced using only macroeconomic variables. The Granger Causaliy es implies ha Bangladesh sock marke price does no appear o be caused by he macroeconomic variables, nor does i have a significan influence on hem excep uni-direcional causaliy runs from exchange rae o sock reurns. The auhor finds no sudy ha paricularly examined he relaionship beween sock price of Bangladesh and he macroeconomic deerminans by employing Impulse Response Funcion (IRF) and Variance Decomposiion (VDC). The impulse response analysis shows ha, shock o macroeconomic variables does no generae a significan response o Bangladesh sock markes. The VDC analyses revealed ha a major proporion of he variabiliy in he sock marke index was explained by is own innovaions while only a minoriy (insignifican) was explained by macroeconomic variables. Overall, i is esablished ha here is no shor run relaionship exiss beween Bangladesh sock marke and he chosen macroeconomic variables. However, here is very weak long run relaionship exiss. I is also esablished ha, here is no causal relaionship exiss beween Bangladesh sock marke and he chosen macroeconomic variables in he shor run. These resuls indicae ha sock prices in Bangladesh canno be prediced using macroeconomic variables. The findings of no shor-run dynamic and he mixed long run relaionship in his sudy help asue invesors and policy makers in efficien and appropriae decision making regarding Bangladesh sock markes. The findings of no causaliy and feedback relaionships verify ha Bangladesh sock markes do no follow he fundamenal and heoreical linkages beween sock prices and macroeconomic variables. Acually, Bangladesh marke is no properly explained by he economic fundamenals. There are some reasons behind ha. In Bangladesh, a big amoun of invesmens are rend-driven and rumour-driven as well (Haque, R. 2011). Weak form of insiuional infrasrucure, lack of accounabiliy and ransparency, quesionable ransparency of marke ransacions and weak form of corporae governance are he common phenomena of Bangladesh sock marke (Khoda and Uddin, 2009). Apar from he above, Dhaka sock marke is familiar wih irraional gains and losses. This marke los 25 per cen price jus wihin a couple of monhs in early 2011, while prices advanced over 82 per cen in lae 2010 (Ahmed, 2011). The seep rise and fall in indices canno follow he macroeconomic deerminans. This scenario in Bangladesh sock marke happens frequenly ha increased he vulnerabiliy and led he markes far away from economic fundamenals. So, for he beermen of Bangladesh sock markes, he above menioned hindrances should be minimised. 15

16 REFERENCES Ahmed, G.T. (2011). Dhaka Bourse now Asia's Wors. The Daily Sar Newspaper, Available a hp:// Accessed on Ocober 31, 2012 Ali, M.B. (2011). T-Y Granger Causaliy beween Sock Prices and Macroeconomic Variables: Evidence from Dhaka Sock Exchange (DSE). European Journal of Business and Managemen, 3(8), Afzal, N. and Hossain, S.S. (2011). An empirical Analysis of he Relaionship beween Macroeconomic Variables and Sock Prices in Bangladesh. Bangladesh Developmen Sudies, XXXIV( 4), Agarwal, S. (2001). Sock Marke Growh and Economic Growh: Preliminary Evidence from African Counries. Journal of Susainable Developmen in Africa. 3(1), Arayssi, M.; Elfakhani,S. and Smaha, H.A. (2008). Globalizaion and Invesmen Opporuniies: A Coinegraion Sudy of Arab, US and Emerging Sock Markes. The Financial Review, 43, BBS (2012), Bangladesh Bureau of saisics, Available a Accessed on Sepember 12, BB (2012), Bangladesh Bank (Cenral Bank of Bangladesh), Available a Accessed on Sepember 10, Bodie, Z., (1976). Common socks as a hedge agains inflaion. Journal of Finance, 31, Dhaka Sock Exchange (DSE-2012). Bangladesh, Available a hp:// Accessed on Ocober 31, Dickey, D., Fuller, W.A. (1979). Disribuion of he Esimaes for Auoregressive Time Series wih a Uni Roo. Journal of he American Saisical Associaion. 74, Dickey, D., Fuller, W.A. (1981). Likelihood Raio Saisics for Auoregressive Time Series wih a Uni Roo. Economerica, 49, Engle, R.F. and Granger, C.W.J. (1987). Co-inegraion and Error Correcion Represenaion, Esimaion and Tesing. Economerica, 55 (1,) Fama, E.F., (1981). Sock Reurns, Real Aciviy, Inflaion and Money. American Economic Review, 71, Flannery, W.J. and Proopapadakis, A.A. (2002). Macroeconomic facors do influence aggregae sock reurns. Review of Financial Sudies, 15, Fama, E.F. and Schwer, W., (1977). Asse reurns and inflaion. Journal of Financial Economics, 5, Granger, C.W.J., (1969). Invesigaing causal relaions by economeric models and cross-specral mehods. Economerica, 37, Gujarai, D. and Porer, D. (2009). Basic Economerics. 5h revised ediion, London, McGraw Hill higher educaion. Gunasekarage, G., Pisedasalasai, A. and Power, D.M. (2004). Macroeconomic influence on he sock marke: evidence from an emerging marke in Souh Asia. Journal of Emerging Marke Finance, 3(3), Hussainey, K. and Ngoc, L. K. (2009). The Impac of Macroeconomic Indicaors on Vienamese Sock Prices. The Journal of Risk Finance, 10(4), Ibrahim, M. H., and Aziz, H. (200 3). Macroeconomic Variables and he Malaysian Equiy Marke: A View hrough Rolling Subsamples. Journal of Economic Sudies, 30, Iqbal, A., Khalid, N. and Rafiq, S. (2011). Dynamic Inerrelaionship among he Sock Markes of India, Pakisan and Unied Saes. World Academy of Science, Engineering and Technology, 73, Jaffe, J.F. and Mandelker, G., (1977). The Fisher Effec for risky asses: An empirical invesigaion. Journal of Finance, 32, Khoda, N.A.K.M. and Uddin, M.G.S (2009). Tesing Random Walk Hypohesis for Dhaka Sock Exchange: An Empirical Examinaion. Inernaional Research Journal of Finance and Economics, 33, Linner, J., (1973). Inflaion and common sock prices in a cyclical conex. Naional Bureau of Economic Research Annual Repor, USA. Lukepohl, H. (1982). Non-causaliy due o Omied Variables. Journal of Economerics, 19, Masuduzzaman, M; (2012). Impac of he Macroeconomic Variables on he Sock Marke Reurns: The case of Unied Kingdom and Germany. Global Journal of Business and Managemen Research, 12(16), Mial, R. and Pal, K. (2011). Impac of Macroeconomic Indicaors on Indian Capial Markes. The Journal of Risk Finance, 12(2) Mukherjee, T.K. and Naka, A. (1995). Dynamic relaions beween Macroeconomic Variables and he Japanese Sock Marke: An applicaion of a Vecor Error Correcion Model. The Journal of Financial Research, 2,

17 Naik, P.K., & Padhi, P. (2012). The Impac of Macroeconomic Fundamenals on Sock Prices Revisied: Evidence from Indian Daa. Eurasian Journal of Business and Economics, 5 (10), Nasseh, A. and Srauss, J., (2000). Sock Prices and Domesic and Inernaional Macroeconomic Aciviy: A Coinegraion Approach. The Quarerly Review of Economics and Finance, 40(2), Nelson, C. R. (1976). Inflaion and raes of reurn on Common Socks. Journal of Finance, 31(2), Pesaran, M., Shin, Y., (1998). Generalised impulse response analysis in linear mulivariae models. Economics Leers, 58, Phillips, P.C., Perron, P. (1988). Tesing for a Uni Roo in Time Series Regression. Biomerika, 75(2), Rahman, L. and Uddin, J. (2009). Dynamic Relaionship beween Sock Prices and Exchange Raes: Evidence from Three Souh Asian Counries. Inernaional Business Research, 2(2), Smih, K. and Sims, A. (1993). Sock marke performance and macroeconomic variables. Applied Financial Economics, 3, Solnik, B. (1983). The relaion beween sock prices and inflaionary expecaions: he inernaional evidence. Journal of Finance, 39(1),