The Relationship Between Trading Volume and Stock Returns

41 The Rela ionship Be ween T ading Volume and S ock Re u ns Chand apala Pa hi awasam Abs ac This s udy examines he ela ionship be ween ading olume and s ock e u ns. The sam- ple o he s udy consis s o 266 s ocks aded a he Colombo S ock Exchange (CSE) om 2000-2008. This s udy ollows he con en ional me hodology used by Jagadeesh and Ti man (Jagadeesh and Ti man, 1993). The s udy e ealed ha s ock e u ns a e posi i ely ela ed o he con empo a y change in ading olume. Fu he , i was ound ha pas ading olume change is nega i ely ela ed o s ock e u ns. In es o misspeci ica ion abou u u e ea nings o illiquidi y o low olume s ocks can be he eason o he nega i e ela ionship be ween ading olume and s ock e u ns. Key wo ds: ading olume, s ock e u ns, Colombo S ock Exchange 1. INTRODUCTION Fama poin s ou ha a ma ke is weak- o m e icien i all he in o ma ion con ains in pas s ock p ices ully e lec in cu en p ices (Fama, 1970, 1991). This implies ha pas secu i y p ices can no be used o p edic he u u e p ice changes and hence, echnical analysis ools ha e no alue. In con as , echnical analys s belie e ha in o ma ion con ained in pas secu i y p ices is no ully inco po a ed in cu en secu i y p ices, and hence, hey belie e ha by obse ing he pas secu i y p ices, in o ma ion can be ob ained on u u e secu i y p ices. The e o e, i is an in e es ing opic in inance o asce ain whe he a ma ke is week- o m e icien . Technical analys s s ongly belie e ha “I akes olume o make p ice mo e” (Kapo , 1987). The ea ly s udies on olume-p ice ela ion sugges ha he e a e posi i e ela ions be ween he absolu e alue o daily p ice changes and daily olume o bo h ma ke indices and indi- idual s ocks. (see, o e.g., Ying, 1966; Wes e ield, 1977; Ru ledge, 1984). The ea ly s udies on olume-p ice ela ion examined con empo aneous ela ionships be ween ading olume and absolu e p ice changes. Hence, hey ha e li le ele ance on he p edic abili y o u u e s ock p ices. Adding a new pa adigm o he ading olume-p ice ela ionships, Ge a s, Kaniel and Mingel- g in (KGM) in es iga e he ole o ading ac i i ies in e ms o he in o ma ion i con ains abou u u e p ices (Ge a s, Kaniel and Mingelg in, 2001). In o he wo ds hey a e in e es ed in he powe o ading olume in p edic ing he di ec ions o u u e p ice mo emen s. They ind ha indi idual s ocks whose ading olume usually la ge (small) o e pe iod o a day o a week, end o expe ience la ge (small) e u ns o e he subsequen mon h. Li e a u e e eals ha mos o he s udies on olume-p ice ela ionships ha e been done based on de eloped ma ke s. The e o e his s udy is o empi ically examine he ading olume-p ice ela ionships in eme ging ma ke o Colombo S ock Exchange. This s udy has wo objec- i es, i s , o empi ically examine he con empo a y ela ionship be ween he ading olume joc_3-2011en_ 3.indd 41 30.9.2011 16:58:59 Jou nal o Compe i i eness | Issue 3/201142 change and s ock e u ns. Second, o empi ically examine he in o ma ion con en o pas ading olume in p edic ing he u u e di ec ion o s ock e u ns. The s udy used mon hly s ock p ices and ading olume o 266 companies du ing he pe iod om Feb ua y 2000 o Decembe 2008. This s udy ollows he con en ional me hod widely adop ed on he momen um li e a u e (Jegadeesh and Ti man, 1993) o o m olume based ading s a egies. The s udy inds ha ading olume is posi i ely ela ed wi h s ock e u ns in he con empo a y pe iod and he ela ionship is nega i e when he pas ading olume is ela ed wi h s ock e u ns. This pape is o ganized as ollows. In he nex sec ion li e a u e ela e o ading olume–p ice ela ionship is b ie ly desc ibed. Sample, da a, hypo heses and me hod o compu a ion explain in sec ion III. In he sec ion IV, main esul s a e p esen ed while las sec ion concludes he pape . 2. REVIEW OF LITERATURE S udies on olume–p ice ela ion go back o 1950s. Osbo ne shows a heo e ical ela ion be- ween olume and p ice (Osbo ne, 1959). Mos o he ea ly s udies ind posi i e co ela ion be ween he absolu e alue o daily p ice changes and daily olume o bo h ma ke indices and indi idual s ocks. G ange and Mo gens e n conduc an ea ly empi ical s udy based on New Yo k S ock Exchange (NYSE) composi e index om 1939-1961(G ange and Mo gens e n, 1963). He inds ha he e is no ela ion be ween absolu e alue o daily p ice changes and daily olume. Howe e , subsequen s udies ind a ela ionship be ween absolu e p ice change and olume change (see, e.g., Ying, 1966; Co ouch, 1970; Epps and Epps, 1976; Ha is, 1986). In ecen s udies esea che s ha e ound con empo a y and lag ela ion be ween s ock e u ns and ading olume (see, e.g., Chen, Fi h and Rui, 2001; Khan and Rizwan, 2001; Lee and Rui, 2002; Pised asalasai and Gunaseka age, 2008). Taking a new app oach o olume– e u n ela ionship GKM de eloped isibili y hypo hesis based on iewpoin s o Mille and Maysha (Mille , 1977; Maysha , 1983). Visibili y hypo hesis holds ha when in es o s ha e di e se opinions abou he alue o a s ock, he ade s who bough he s ock a e op imis ic abou i s alue. Fu he , when he s ock’s supply is limi ed by sho selling o ma gin ading hen he opinion o he pessimis ic in es o s will no inco po- a e in o he s ock p ice and he s ock p ice will be bias. GKM poin ou ha unde such si u- a ion, any posi i e shock in numbe o ade s gi ing a en ion o he s ock (GKM named as inc ease in he s ock’s isibili y) will inc ease demand o he s ock (because numbe o buye s inc ease). Howe e , supply o he s ock will emain cons an (no excess in sales). Hence, ol- ume and p ice mo e posi i ely. The e o e, he isibili y hypo hesis sugges s ha unde ma ke cons ain s, i mo e ade s’ a en ion is a ac ed on a s ock i s ading olume and p ice will inc ease. Visibili y hypo hesis is i s es ed by GKM and subsequen ly by Huang and Heian (Huang and Heian, 2010). GKM examine he ela ionship be ween cu en ading olume wi h u- u e e u ns o NYSE om 1963 o 1996 o bo h daily and weekly da a. Numbe o sha es aded is used as he measu e o ading olume. Po olios a e o med in acco dance wi h he Jegadeesh and Ti man app oach (Jegadeesh and Ti man, 1993). They o med high, medium joc_3-2011en_ 3.indd 42 30.9.2011 16:58:59 43 and low olume po olios based on daily and weekly da a and wi hou ebalancing kep hese po olios o 1, 10, 20, 50 o 100 ading days. S udy inds ha po olios wi h high ading olume ended o be ollowed by high e u ns and ice e sa. This high- olume e u n p e- mium is ue when he o ma ion pe iod is day o a week. I las ed a leas 20 ading days and a mos 100 ading days. Huang and Heian examine he isk adjus ed high alue p emium based on all i ms lis ed on NYSE and AMEX om Augus 1963 o Decembe 2005 (Huang and Heian, 2010). They use he con en ional me hod widely used by momen um li e a u e (Jegadeesh and Ti man 1993) o es he s a egy. Fo ma ion pe iod is 26 weeks and he holding pe iod a ied om 1 week o 52 weeks. They ind s a is ically signi ican abno mal e u ns o high olume minus low ol- ume po olio o holding pe iods 1-4 weeks. Howe e , hey u he o ind ha as he holding pe iod inc ease beyond 8 weeks, abno mal e u ns dec ease signi ican ly. 3. DATA AND METHODOLOGY 3.1 Da a All he da a used in he s udy is aken om he CSE da a lib a y. The sample pe iod co e s 9 yea s om Feb ua y 2000 o Decembe 2008. The sample o he s udy includes all he lis ed s ocks in he CSE. This sample includes e en delis ed s ocks in o de o add ess he p oblem o su i o ship bias1. The e o e he o al sample includes 266 companies. 3.2 Me hodology Volume e u n- ela ionships a e examined in wo ways. Fi s , he con empo a y ela ion be- ween changes in olume- e u n is examined and nex he ela ionship be ween pas ading olume and u u e s ock e u ns a e examined. Change in olume and mon hly s ock e u ns a e compu ed as ollows. The a iables used in he s udy a e mainly mon hly indi idual s ock p ices and numbe o s ocks aded. Pe cen age mon hly e u ns a e compu ed using indi idual s ock p ices as ol- lows. 100 1, 1,, , u    i i i i P PP R i R , = Capi al gain e u ns o he i h sha e in he mon h . i P , = P ice o he i h sha e a he end o mon h . = P ice o he i h sha e a he end o he p e ious mon h. 1,  i P Pe cen age mon hly e u ns a e adjus ed o di idends, igh issues and bonus issues on he basis o ein es men assump ion a he end o he mon h in which ex-da e occu ed. The ol- lowing o mula is used o compu e he mon hly ading olume changes. 1 Ko ha i, Shanken and Sloan show ha he da a selec ion biases including a su i o bias signi ican ly a ec ed on he anomalies (Ko ha i, Shanken and Sloan, 1995). joc_3-2011en_ 3.indd 43 30.9.2011 16:58:59 Jou nal o Compe i i eness | Issue 3/201144 1, 1,, )(    i i i Vol VolVol W he e, i Vol , = he ading olume o company i o mon h 1,  i Vol = he ading olume o company i o p e ious mon h In o de o examine he con empo a y change in olume- e u n ela ionships, each mon h s ocks a e anked and g ouped in o h ee po olios as high olume (HV), middle olume and low olume (LV) po olios based on hei olume change du ing he pas 3, 6 9 and 12 mon hs and hei a e age e u ns a e compu ed o 3, 6 9 and 12 mon hs unde each pe iod. Nex , he ela ionship be ween pas ading olume and u u e s ock e u ns a e examined. Each mon h s ocks a e anked and g ouped in o h ee po olios as HV, middle olume and LV po olios based on hei olume change du ing he pas 3, 6, 9 and 12 mon hs and hei a e age e u ns a e compu ed o nex 3, 6, 9 and 12 mon hs unde each pe iod. 3.3 Hypo heses I he change in ading olume ca ies in o ma ion on s ock e u ns, he e should ha e a signi ican ela ionship be ween change in olume and s ock e u ns. A e age e u ns o HV and a e age e u ns o LV po olios a e compa ed o de ec ela ions be ween ading olume change and s ock e u ns. The e o e, he null hypo hesis (H0) and he al e na i e hypo hesis (H1) a e de eloped as ollows. H 0 :   0  LVHV RRE H 1 :   0z VLHV RRE Whe e, HV R = a e age e u ns o high olume po olio LV R = a e age e u ns o low olume po olio The null hypo hesis explains ha HV po olio and LV po olio has he same expec ed e- u ns while he al e na i e hypo hesis explains ha expec ed e u ns o HV po olio is di e - en om ha o LV po olio. 3.4 Tes o signi icance The signi icance o he e u ns di e ence on ading olume po olios is measu ed using he pai ed sample – es s. The – alues a e compu ed as ollows. nVa RnVa R RR LVHV LVHV //   joc_3-2011en_ 3.indd 44 30.9.2011 16:59:00 45 4. RESULTS This s udy has wo objec i es. Fi s , o examine he con empo a y ela ionship be ween ad- ing olume and s ock e u ns and he second objec i e is o examine he ela ionship be ween pas pe iod ading olume and cu en s ock e u ns (lag olume ela ionship). 4.1 Con empo a y ading olume e ec The able 1 shows a e age mon hly e u ns o HV, LV and high minus low olume (HmLV) po olios o 16 es s. Each mon h s ocks a e anked and di ided in o h ee po olios based on 3,6,9 and 12 mon h change in ading olume and unde each change in olume po olio a e age e u ns a e compu ed o 3, 6, 9 and 12 mon hs. Tab. 1 - Con empo a y T ading Volume E ec om 1995-2008. Sou ce: own J=Fo ma ion Pe iod, K= Holding Pe iod K=3 K=6 K=9 K=12 J=3 HV LV HmLV 1.90 0.17 2.07*** 2.00 0.58 1.42*** 1.80 0.87 0.93** 1.72 1.07 0.65*** J=6 HV LV HmLV 1.08 0.51 0.57*** 2.02 0.41 1.61** 1.83 0.80 1.03*** 1.72 1.03 0.69*** J=9 HV LV HmLV 1.26 0.77 0.49*** 1.47 0.91 0.56*** 1.92 0.75 1.16*** 1.78 1.00 0.78*** J=12 HV LV HmLV 1.14 0.97 0.17*** 1.42 0.97 0.45*** 1.47 1.01 0.46*** 1.76 0.91 0.85*** ** Signi ican ly di e en om ze o a he 5% le el. *** Signi ican ly di e en om ze o a he 1% le el. A a i s look, he able e eals ha all he HV po olios e lec posi i e a e age e u ns and hey ou pe o m he a e age e u ns o LV po olios. Po olios o med based on 3 mon h ading olume change and hei 3 mon hs e u ns show highes e u ns di e ence be ween HV and LV po olios (2.07% a =11.86). The lowes e u ns di e ence be ween high and low olume po olios epo s when po olios o med based on 12 mon h a e age ading olume change and held o 3 mon hs. The HV po olios ou pe o m he LV po olios o all he es po olios epo ed in he able. The e o e, able 1 shows ha he e is a con empo a y ela ion- ship be ween ading olume change and s ock e u ns. joc_3-2011en_ 3.indd 45 30.9.2011 16:59:00 Jou nal o Compe i i eness | Issue 3/201146 4.1 Lag ading olume e ec Tab. 2 - Lag T ading Volume E ec om 1995-2008. Sou ce: own J=Fo ma ion Pe iod, K= Holding Pe iod K=3 K=6 K=9 K=12 J=3 P1 P3 P1-P3 0.74 0.77 -0.03 1.12 1.26 -0.13 1.25 1.40 -0.15 1.37 1.47 -0.10 J=6 P1 P3 P1-P3 0.63 0.92 -0.29*** 1.10 1.46 -0.36*** 1.28 1.55 -0.27*** 1.36 1.61 -0.25*** J=9 P1 P3 P1-P3 0.64 0.88 -0.24** 1.17 1.49 -0.32*** 1.32 1.64 -0.31*** 1.45 1.08 0.23*** J=12 P1 P3 P1-P3 0.69 0.84 -0.15*** 1.11 1.55 -0.44*** 1.33 1.68 -0.35*** 1.39 1.66 -0.27*** The able 3 e lec s a e age mon hly po olio e u ns when po olios a e o med based on pas pe iod ading olume changes. Each mon h om Feb ua y 2000 o Decembe 2008, s ocks a e anked and o m in o 3 equally weigh ed po olios based on 3, 6,9 and 12 mon h pas change in ading olumes. Subsequen ly, a e age po olio e u ns a e compu ed o he nex 3, 6, 9 and 12 mon hs o each po olio in he o ma ion pe iod. This makes 16 di e en es po olios. A a glance, he able shows ha low olume po olios ou pe o m he high olume po olios o all he po olios epo ed in he able. The a e age e u ns di e ence is no s a is ically signi ican o he po olios wi h o ma ion pe iod 3 mon hs. The a e age e u ns di e ences o all he o he po olios epo ed in he able gene a e s a is ically signi i- can nega i e excess e u ns. The ou pe o mance o LV po olio o e he HV po olio is con a y o he p e ious ind- ings (see, Ge a s e al., 2001; Huan and Heian 2010). Fu he , he inding is no in acco dance wi h he sequen ial a i al o in o ma ion heo y o Copeland (1976) and mix u e o dis ibu- ion hypo hesis o Epps and Epps (1976). The ou pe o mance o HV po olio e u ns by LV po olio e u ns can be jus i ied wi h wo easons. Fi s , he highe (lowe ) u u e e u ns o LV(HV) s ocks can be due o in es o mispe cep ions abou u u e ea nings. Lee and Swami- na han ound nega i e ela ionship be ween ading olume and e u ns and hey poin ed ou ha (Lee and Swamina han, 2001): … analys s p o ide lowe (highe ) long- e m ea nings g ow h o ecas s o low(high) olume s ocks. Howe e , low (high) olume i ms expe ience signi ican ly be e (wo se) u u e ope a ing pe o mance. Mo eo e , we ind ha sho -window ea nings announcemen e u ns a e signi ican ly mo e posi i e (nega i e) o low (high) olume i ms o e each o he nex eigh qua e s. The same pa e n is obse ed in his s udy also and as Lee and Swamina han poin ed ou ha his can be due o in es o mispe cep ions abou u u e ea nings o low olume i ms (Lee joc_3-2011en_ 3.indd 46 30.9.2011 16:59:00 47 and Swamina han, 2001). As a esul o ha ma ke is “su p ised” by he sys ema ically highe (lowe ) u u e ea nings o low (high) olume i ms. The second jus i ica ion o he nega i e ela ionship be ween ading olume and e u n is he liquidi y o he asse s. Da a , Naik and Radcli e ind nega i e ela ionship be ween u u e s ock e u ns and liquidi y measu ed by s ock u no e a e o all non inancial i ms on he NYSE om July 31, 1962 h ough Decembe 31, 1991 (Da a , Naik and Radcli e, 1998). As Da e e al. poin ou he nega i e sign be ween s ock e u n and ading olume con i ms ha illiquid s ocks o e highe a e age e u ns han liquid s ocks (Da e e al., 1998). This ac is ele an o CSE also because esea che s ha e ound ha mos o he s ocks a CSE do no ade equen ly ( see, e.g., Sama akoon, 1996; Pa hi awasam and Idi isinghe, 2011). 5. CONCLUSION This s udy examines he ading olume e ec on s ock e u ns a Colombo S ock Exchange om 2000-2008. The s udy adds some impo an indings o he exis ing li e a u e as ading olume e ec is p o en on ex ensi ely in de eloped ma ke s while li le e idence on de elop- ing ma ke s. This s udy examines he ela ionship be ween ading olume change and s ock e u n in wo s ages. Fi s con empo a y ela ionship be ween ading olume change and s ock e u ns a e examined and nex , he ela ionship be ween pas pe iod ading olume change and s ock e u ns a e examined. The s udy inds ha con empo a y ading olume change is posi i ely ela e wi h he s ock e u ns. Howe e , he ela ion be ween pas pe iod ading olume change and cu en pe iod s ock e u ns is nega i e. This means s ocks wi h low ading olume change ou pe o m he s ocks wi h high ading olume change in he subsequen pe iod. The ou pe o mance o high olume po olio e u ns by low olume po olio e u ns can be jus i ied wi h wo easons. The highe (lowe ) pe o mance o low olume (high olume) s ocks can be due o in es o misspeci ica ion abou u u e ea nings. O else, illiquidi y o low olume can be he eason o ou pe o mance o low olume s ocks. As he ading olume has p edic i e powe on s ock e u ns, in es o s can make ading olume based s a egies o make p o i s and heo e ically his p o ides e idence o weak o m ine iciency o he CSE. Re e ences Chen, G., Fi h, M. & Rui, O. M. (2001). The dynamic ela ion be ween s ock e u ns, ad- ing olume, and ola ili y. The Financial Re iew, ol. 38, 153-174. Cohen, K., Maie , S. F., Schwa z, R. A. & Whi comb, D. K. (1979). Ma ke Make s and he Ma ke Sp ead: A Re iew o Recen Li e a u e. Jou nal o Financial and Quan i a i e Analysis, ol.14, 813-835. C ouch. R. L. (1970). A Nonlinea Tes o he Random-Walk Hypo hesis. Ame ican Economic Re iew, 199-202. Da a , V.T., Naik & Radcli e, R. (1998). Liquidi y and s ock e u ns: An al e na i e Tes . Jou nal o Financial Ma ke s, ol. 1, 203-219. 1. 2. 3. 4. joc_3-2011en_ 3.indd 47 30.9.2011 16:59:00 Jou nal o Compe i i eness | Issue 3/201148 Epps, T. W., Epps, M. L. (1976). The S ochas ic Dependence o Secu i y P ice Changes and T ansac ion Volumes: Implica ions o he Mix u e-o -Dis ibu ions Hypo hesis. Econome - ica, ol. 44, 305-321. Fama, E. F. (1970). E icien Capi al Ma ke s: A Re iew o Theo y and Empi ical Wo k. Jou nal o Finance, ol. 25, 383 – 417. Fama, E.F. (1991). E icien Capi al Ma ke s: II. Jou nal o Finance, ol. 46, 1575 – 1617. G ange . C. W. J. & Mo gens em, O. (1963). Spec al Analysis o New Yo k S ock Ma ke P ices. Kyklos, ol. I6, 1-27. Ge ais, S., Kaniel, R. & Mingelg in, D.H. (2001). “The high- olume e u n p emium”, The Jou nal o Finance, ol. 51, 877-919. Ha is, L. (1983). The Join Disl ihuiion o Specula i e P ices and o Daily T ading Volume. Wo king Pape , Uni e si y o Sou he n CA . Khan, S.U., Rizwan, F. (2008). T ading olume and s ock e u ns: e idence om Pakis an’s s ock ma ke . In e na ional Re iew o Business Resea ch Pape s, ol. 4, 151-162. Huang, Z. & Heian, J.B. (2001). T ading- olume shocks and s ock e u ns: An empi ical analysis. The Jou nal o Financial Resea ch, ol. 38, 153-177. Jegadeesh, N. & Ti man, S. (1993). Re u ns o buying winne s and selling lose s: Implica ions o s ock ma ke e iciency. Jou nal o Finance, ol. 48, 65-91. Ka po , J. (1987). The ela ion be ween p ice changes and ading olume: A su ey. Jou nal o Financial and Quan i a i e Analysis, ol. 22, 109 -125. Ko ha i, S.P., Shanken, J. & Sloan, R.G. (1995). Ano he look a he c oss-sec ion o expec ed s ock e u ns. Jou nal o Finance, ol. 50, 185- 223. Lee, S.B. & Rui, O. M. (2002). The dynamic ela ionship be ween s ock e u n and ading olume: Domes ic and c oss-coun y e idence. Jou nal o Banking and Finance, ol. 26, 51-78. Maysha , J. (1983). On di e gence o opinion and impe ec ions in capi al ma ke s. Ame ican Economic Re iew, ol. 73, 114-128. Mille , E.M. (1977).Risk, unce ain y, and di e gence o opinion. Jou nal o Finance, ol. 32, 1151-1168. Osbome. M. F. M. (1959). B ownian Mo ion in he S ock Ma ke . Ope a ions Resea ch, 1 (Ma ch- Ap il 1959), 145-173. Pa hi awasam, C., Idi isinghe, I.M.S.K. (2011). Ma ke E iciency, Thin T ading and Non- linea Behe io : Eme gin Ma ke E idence om S i Lanka. E+M Economics and Managemen , ol.8, 112-122. Pised asalasai, A. & Gunaseka age, A. (2007). Causal and dynamic ela ionships among s ock e u ns, e u n ola ili y and ading olume: E idence om eme ging ma ke s in Sou h- Eas Asia. Asia-Paci ic Financial Ma ke s, ol. 14, 277-297. Ru ledge. D. J. S. (1984) T ading Volume and P ice Va iabili y: New E idence on he P ice E ec s o Specula ion. In Peck A. E. (ed.), Selec ed W i ings on Fu u es Ma ke s: Resea ch Di ec ions in Commodi y Ma ke s (237-251). Chicago: Chicago Boa d o T ade. Sama akoon, L.P. (1996). P edic abili y o Sho -ho izon Re u ns in he S i Lankan S ock Ma ke . S i Lankan Jou nal o Managemen , ol. 1, 207-224. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. joc_3-2011en_ 3.indd 48 30.9.2011 16:59:00 49 Wes c ie d. R. (1977). The Dis ibu ion o Common S ock P ice Changes: An Applica ion o T ansac ions Time and Subo dina ed S ochas ic Models. Jou nal o Financial and Quan i a i e Analysis, ol. 12, 743-765. Ying. C. C. (1966). S ock Ma ke P ices and Volumes o Sales. Ecomme ica, ol. 34, 676-686. Con ac in o ma ion Chand apala Pa hi awasam. B.Sc, M.Sc, MBA Tomas Ba a Uni e si y in Zlin, Facul y o Managemen and Economics Mos ní 5139, 760 01 Zlín, Czech Republic. Facul y o Comme ce and Managemen S udies, Uni e si y o Kelaniya, S i Lanka Depa men o Comme ce and Financial Managemen E-mail: Pa [email protected] 24. 25. JEL Classi ica ion: G12 joc_3-2011en_ 3.indd 49 30.9.2011 16:59:00