Asian Economic and Financial Review EMERGING STOCK PREMIA: SOME EVIDENCE FROM INDUSTRIAL STOCK MARKET DATA. Michael Donadelli. Marcella Lucchetta

Transcription

1 Asan Economc and Fnancal Revew journal homepage: EMERGING STOCK PREMIA: SOME EVIDENCE FROM INDUSTRIAL STOCK MARKET DATA Mchael Donadell Department of Economcs and Fnance, LUISS Gudo Carl, Rome (Italy) Marcella Lucchetta Department of Economcs, Ca Foscar Unversty, Vence (Italy) ABSTRACT Ths paper studes the behavor of emergng stock excess returns n an ndustry-by-ndustry context. We examne stock market performance for 23 countres and ten ndustres over 17 years from 1995 to 2012 a perod that ncludes major changes n captal market regulatons, the removal of trade barrers, the IT bubble, the 9/11 terrorst attacks, and the subprme mortgage crss. In addton, we examne stock market co-movement and rsk exposure for ten ndustres n eght emergng/developng stock markets. We obtan four key emprcal fndngs. Frst, at ndustry level, we confrm that the equty rsk premum n emergng markets s hgher than n developed markets. We also confrm the tme-varyng nature of emergng stock market excess returns. Second, at country level, we dentfy those ndustres that manly contrbute to the presence of the emergng stock prema. Thrd, we show that some ndustres are more exposed to global rsk factors than others. Fourth, gven the ncreasng degree of co-movement between nternatonal stock markets, we observe that some cross-ndustry portfolo dversfcaton benefts are stll explotable. Our analyss yelds nterestng mplcatons for fnancal applcatons. In partcular, we argue that the presence of a strong tme-varyng component n the ndustry-betas mght have strong mpact on the estmaton of the cost of captal. Keywords: Emergng and Developng Economes, Industral Stock Markets, Tme-Varyng ERP JEL Classfcaton: D82, F36, F44 G11, G15 398

2 INTRODUCTION AND MOTIVATIONS The behavor of the emergng and developng equty rsk prema has been largely studed n the recent fnancal lterature. 1 Common emprcal results are: () the equty rsk premum n emergng markets s hgher than n developed markets; () emergng stock market nvestments ncrease portfolo dversfcaton benefts; () emergng stock markets are hghly volatle; (v) the extent to whch emergng stock markets reward nvestors s varyng through tme. 2 In recent years, researches have focused on the effects of the fnancal lberalzatons and global ntegraton process on emergng economc growth and nternatonal stock market prces. 3 It s largely accepted that the removal of barrers to nternatonal captal flows has decreased the cost of captal (.e. excepted returns) n emergng stock markets (Bekaert and Harvey, 2000). For example, De Jong and De 1 See Bekaert, G., Market ntegraton and nvestment barrers n emergng equty markets. World Bank Economc Revew, 9: , Bekaert, G. and C.R. Harvey, Emergng equty market volatlty. Journal of Fnancal Economcs, 43: 29-78, Bekaert, G. and C.R. Harvey, Captal flows and the behavor of emergng market equty returns. Nber chapters. Captal Flows and the Emergng Economes: , Bekaert, G., B.E. Claude, C.R. Harvey and T.E. Vskanta, Dstrbutonal characterstcs of emergng market returns and asset allocaton. Journal of Portfolo Management: , Donadell, M. and L. Prosper, 2012a. The equty rsk premum: Emprcal evdence from emergng markets. CASMEF Workng Paper., Donadell, M. and L. Prosper, 2012b. The equty premum puzzle: Ptfalls n estmatng the coeffcent of relatve rsk averson. Journal of Appled Fnance & Bankng, 2(2): , Donadell, M. and L. Prosper, 2012c. On the role of lqudty n emergng markets stock prces. Research n Economcs, 66(4): , Grootveld, H. and R. Salomons, The equty rsk premum: Emergng vs. Developed markets. Emergng Markets Revew, 4(2): , among many others. 2 For a detaled dscusson on the dversfcaton benefts of nvestng n emergng markets, see Barry, C.B., I. Peavy, J.W. and M. Rodrguez, A convenent way to nvest n emergng markets. Emergng Markets Quarterly, 1(1): and Claessens, S., S. Dasgupta and J. Glen, Return behavor n emergng stock markets. World Bank Economc Revew 9(1): See Bekaert, G. and C.R. Harvey, Captal flows and the behavor of emergng market equty returns. Nber chapters. Captal Flows and the Emergng Economes: , Henry, P., Stock market lberalzaton, economc reform and emergng market equty prces. Journal of Fnance, 55(2): , De Jong, F. and F.A. De Roon, Tme-varyng market ntegraton and expected returns n emergng markets. Journal of Fnancal Economcs, 78: , Kefela, G.T., Drvng forces of globalzaton n emergng market economes and developng countres. Asan Economc and Fnancal Revew, 1(2): , Donadell, M. and L. Prosper, 2012a. The equty rsk premum: Emprcal evdence from emergng markets. CASMEF Workng Paper, Donadell, M. and L. Prosper, 2012c. On the role of lqudty n emergng markets stock prces. Research n Economcs, 66(4): , Hye, Q.M.A. and S. Wzarat, Impact of fnancal lberalzaton on economc growth: A case study of pakstan. Asan Economc and Fnancal Revew, 3(2): , among many others. 399

3 Roon (2005), fnd that the average annual decrease n segmentaton reduces the cost of captal by about 11 bass ponts, and reduces stock returns by about 4.5%. Donadell and Prosper (2012a), n a dynamc context fnd an opposte results. Ther results suggest that fnancal and real market openness ncrease, ex-post, expected excess returns n emergng stock markets. In lne wth Donadell and Prosper (2012a), Karadagl (2012) fnds that the overall level of globalzaton sgnfcantly mproves frm performance n emergng countres. Exstng emprcal fndngs also document an ncreasngly degree of co-movement between nternatonal stock markets (see Fg. C.1) as well as between busness cycles (see Fg. 4.1). 4 It turns out that emergng markets tend to be more exposed to global macroeconomc shocks. To analyze the behavor of nternatonal stock market prces, all these studes have focused on natonal stock market ndces. We argue that few works have employed ndustral stock market ndces. Usng the total return ndex for 23 countres and ten ndustres, we mprove the exstng lterature nto three man drectons. Frst, we provde new emprcal evdence based on an extensve emergng/developng stock markets ndustry-based dataset. The ndustry-based analyss allows us to capture the man sources of the observed emergng stock prema (.e. equty rsk premum n emergng markets s hgher than n developed markets). We fnd that some emergng ndustral stock markets (e.g. healthcare and utltes) have generated hgher average excess returns. In addton, we confrm that emergng ndustral stock markets have a strong-tme varyng component, and are, on average, ncreasngly ntegrated. Second, we show that the exposure to the global market factor across ndustres s relatvely heterogeneous and heavly contngent on state and tme (e.g. emergng crss, 9/11 terrorst attacks and Lehman Brothers collapse). Thrd, we observe that our emprcal results have strong mplcatons for mean-varance portfolo dversfcaton strateges and, n a consumpton-based asset prcng framework, reflect consumpton-smoothng motve (.e. nsurance-motve). Not surprsngly, we fnd that most emergng ndustral stock markets are strongly correlated. Nevertheless, few ndustres stll allow for consumpton-smoothng motve and portfolo dversfcaton benefts. For nstance, focusng on the last fve years, the technology, consumer goods, consumer servces and telecommuncatons sectors would allow for a wse portfolo composton. The rest of the paper s organzed as follows. Secton 2 revews the lterature. Secton 4 See Bekaert, G. and C.R. Harvey, Emergng equty market volatlty. Journal of Fnancal Economcs, 43: 29-78, Bekaert, G. and C.R. Harvey, Captal flows and the behavor of emergng market equty returns. Nber chapters. Captal Flows and the Emergng Economes: , Bekaert, G., B.E. Claude, C.R. Harvey and T.E. Vskanta, Dstrbutonal characterstcs of emergng market returns and asset allocaton. Journal of Portfolo Management: , Carrer, F., V. Errunza and K. Hogan, Characterzng world market ntegraton through tme. Journal of Fnancal and Quanttatve Analyss, 42(4): , Donadell, M. and L. Prosper, 2012a. The equty rsk premum: Emprcal evdence from emergng markets. CASMEF Workng Paper., Henry, P., Stock market lberalzaton, economc reform and emergng market equty prces. Journal of Fnance, 55(2): , among others. 400

4 3 descrbes data and analyzes emergng stock prema n a statc context. Secton 4 analyzes the performance and the rsk exposure of eght emergng ndustral stock markets n a dynamc context. Secton 5 concludes. RELATED LITERATURE The behavor of the emergng equty rsk premum has been largely studed n the last two decades. As mentoned, most of these emprcal studes employ natonal stock market ndces. In addton, exstng emprcal results have been found usng pre-2000 data. In contrast, a lmted number of studes have employed an updated ndustry-based dataset. In ths paper, we re-evaluate the mportance of ndustres performances n explanng emergng market stock prema. In general, the fnancal lterature employs ndustral stock market ndces to construct ndustry-based rsk factors amed at explanng varaton n emergng excess returns. Ths study s unque n that we drectly analyze ndustres performances and ther exposure to the market factor. Roll (1992) fnds that the ndustry component s sgnfcant, whch means that stocks from dfferent countres, but from the same ndustry, are correlated. Therefore, countres whose stock markets are smlar n terms of ndustry composton wll be nterdependent (.e. ndustry composton s sgnfcant n explanng stock market correlaton). Serra (2000) draws an opposte concluson. She fnds that country effects are the most mportant factors drvng the behavor of emergng markets ndvdual stock returns. In other words, emergng market ndces are drven by country factors and cross-market correlaton does not seem to be affected by the ndustral composton of the ndces. The same result of Serra s found by Ang et al. (2009). They present evdence that the negatve relaton between lagged dosyncratc volatlty and future average returns s observed across a broad sample of nternatonal developed markets. They fnd also how the negatve spread n returns between stocks wth hgh and low dosyncratc volatlty n nternatonal markets strongly co-moves wth the dfference n returns between U.S. stocks wth hgh and low dosyncratc volatltes. They conclude that there are not easly dversfable factors behnd ths effect. In lne wth our paper, Brooks and Del (2002) clam that dversfcaton across ndustres results to be more effectve than dversfcaton across countres. Nevertheless, n contrast wth our study, they explore the mplcaton of ndustry only n a regonal framework. Smlarly, Elng et al. (2012) show that nternatonal returns are prmarly drven by ndustry and currency rsk factors, but they analyze the G7 countres only. In contrast to our paper, all these studes focus on the role of ndustres (or ndustry factors) n explanng varaton n emergng stock returns, and as n our paper, they use ndustral stock market ndces. 401

5 Internatonal Excess Returns: Some Stylzed Facts We download country-by-country and ndustry-by-ndustry TRIs from Datastream Global Equty Indces (DGEI). 5 For each country, ten or less dfferent ndustral TRIs are avalable. All ndces are monthly total returns denomnated n US dollars (.e. dvdends are ncluded) and run from January 1995 (or later) to June2012. We examne the behavor of the followng ndustral stock market excess returns: Ol & Gas, Basc Materals, Consumer Goods, Consumer Servces, Industres, HealthCare, Fnancals, Technology, Telecom and Utltes. Detals on DGEI are gven n appendx B. The full set of ndustry-based TRIs s avalable for eght emergng countres (.e. Brazl, Chle, Chna, Malaysa, Israel, Sngapore, Thaland, Turkey). We decde to restrct the large part of our dynamc analyss on to these eght markets. We argue that they represent a reasonable set of stock markets n whch a US nternatonal nvestor mght be nterested. For comparson purposes, we use the US ndustral stock market ndces. Stock returns are computed for each ndustry n country k. Formally, k, t k, t GEI TRI GEI TRI 1 R k, t (3.1) GEI TRI k, t 1 where GEI TRI k, t represents the DGEI of ndustry n the emergng country k at tme t. We obtan the excess return by subtractng a rsk-free rate proxy from Eq. (3.1). Formally, f k, t R k t Rt (3.2) ExR, where f Rt s the one-month Treasury bll rate (from Ibbotson Assocates). Defnton 1: The excess return spread of ndustry n country k s gven by k, t k, t US t Spread ExR ExR, where ExR k, t s the excess return of ndustry n the emergng country k and ExR US, t s the excess return of ndustry n the US. Summary statstcs for the 23 equty market excess returns are reported n Table B.1. For each country, we compute the mean (frst lne), standard devaton (second lne) and Sharpe rato (thrd lne). The fourth lne of Table B.1 reports the ndustry average spread (as defned n Def. 1) for 5 Datastream Global Equty Indces break down nto sx levels. Level 1 s the market ndex, ths covers all the sectors n each regon or country. Level 2 dvdes the market nto 10 ndustres and covers all the sectors wthn each group n each regon or country. Source: Datastream. 402

6 each emergng stock market k. As expected, the emergng average ndustral stock market excess returns are hgher than the US average ndustral stock market excess returns. The average spread appears to be postve both across ndustres and countres (.e. longtudnal and vertcal averages of lne four n Table B.1). We argue that negatve spreads are manly nfluenced by local shocks (.e. emergng crss of the late 90s and early 00s). 6 Clearly, a post-crses (or dynamc) analyss wll delver dfferent results (see Fg. 3.1). Defnton 2: The country-by-country ndustral average spread s gven by Country Avg 1 I 1 T Spreadk Spread k I 1 T t 1, t where represents the ndustry and k denotes the emergng country. Emergng country - cross-ndustry - average spreads (.e. Def. 2) are llustrated n Table 3.1. In lne wth exstng emprcal fndngs on emergng stock markets performances, our estmates suggest that emergng stock markets tend to perform better than the US stock market. For the sample January June 2012 the emergng countres average spread s negatve only n four out of 22 stock markets. On annual bass, the average spread n Argentna, Phlppnes, Sr Lanka and Tawan s equal to -1.80%, -1.49%, % and -0.47%, respectvely. The postve spread ranges from a mnmum of 0.85% (Hungary) to a maxmum of 21.80% (Russa). Results n Table. B.1 suggest that the performance of the Russan stock market has been drven by the consumer servces sector. Lne 4 (Russa) of Table 4.1 shows that the consumers servces ndustry annual average excess return s equal to 46.8%. Table-3.1. Emergng Countres: Country Average Spread. Average values (computed as s Def. 2) are n annual terms and expressed n percentage ponts. Sample: January 1995 (or later) June Country Mean Country Mean ARGENTINA PAKISTAN 2.34 BRAZIL 9.83 PERU CHILE 2.18 PHIL CHINA POLAND For example, we obtan negatve spread average values n the followng countres (ndustres): Argentna (ol&gas, consumer goods, fnancals, utltes); Mexco (consumer goods, ndustrals); Phlppnes (ol&gas, basc materals, consumer goods, consumer servces, fnancals). E.g. crss dates: Argentna (1995, ); Mexco ( ), Phlppnes ( ). 403

7 CZ REP 5.96 RUSSIA HONG-KONG 5.35 SINGAPORE 3.51 HUNGARY 0.85 SOUTH AFRICA 6.57 INDIA 5.89 SRI LANKA ISRAEL 4.24 TAIWAN MALAYSIA 3.04 THAI 3.19 MEXICO 6.00 TURKEY Defnton 3: The ndustry-by-ndustry average spread s gven by 1 K 1 T Industry Avg Spread Spread K k 1 T t 1 where represent the ndustry and k denotes the country., t k Table 3.2 reports the emprcal counterparts of Def. 3 for two sets of emergng stock markets: a full set contanng all emergng countres and a set composed by our eght benchmark emergng economes. For both sets, we fnd that the average spread s postve across all ndustres (.e. emergng ndustry-erp s hgher than the US ndustry-erp). In both sets, the hghest cross-country ndustry average spread has been generated by the healthcare ndustry. For the set composed by all the emergng countres, Table 2.3 shows that the annual average spread s equal to 12.05%. For the set composed by the eght benchmark economes, the spread s equal to 16.05%. In other words, our smple results suggest that the healthcare emergng ndustral stock market has domnated all the other ndustral stock markets over the last 15 years. It turns out that such ndustry has provded a substantal contrbuton to the extra premum pad by the emergng stock market world to nternatonal nvestors. Table B.1 suggests that emergng healthcare extra premum has been largely drven by the Chnese healthcare ndustry (.e. an annual average spread for the Chnese healthcare ndustry s 49.2%). Table-3.2. Emergng Countres: Industry Average Spread. Average values (computed as s Def. 3) are annualzed and expressed n percentage ponts. Sample: January 1995 (or later) June Country Average Average (All Emergng) (8 Emergng) Ol & Gas Basc Materals Consumer Goods Consumer Servces Industrals HealthCare Fnancals Telecom Technology Utltes

8 The result s obvously sample and state senstve. Gven that average emergng excess returns have a strong tme-varyng component, we conduct our ndustry-by-ndustry performance analyss n a tme-varyng framework. Fg. 3.1 llustrates the dynamcs of the ndustral stock market average (computed per unt of rsk ) excess returns for the followng countres: Unted States, Brazl, Chle, Chna, Israel, Malaysa, Sngapore, Thaland, Turkey. In all subplots the black dashed lne represents the US ndustres. Both across ndustres and countres, we observe smlar Sharpe ratos patterns. Sharpe ratos are estmated usng a rollng wndow of 60 months. 7 Emergng ndustry Sharpe ratos are much more volatle than the US ndustral Sharpe ratos. The result confrms the unstable ndustral structure of the emergng economes. A more generous average compensaton s also evdent. The performance of the healthcare ndustry (corrected per unt of rsk) stll seems to domnate all the others (see subplot HealthCare of Fg. 3.1). A relevant contrbuton s also gven by the consumer goods and utlty ndustres. Ther cross-wndows average performances are 9.94% and 10.01%, respectvely. 8 Fgure-3.1. Industry-by-Industry Sharpe ratos. Sharpe ratos - computed as the rato between the excess return and the standard devaton of each ndustry TRI - are estmated usng a rollng sample of 60 months. Formally, k ShR ExR, k, w / Sdk w, where w represents the wndow n whch the rato s estmated. The sample perod goes from January 1995 (or later) to June The number of observatons per estmaton s 60. The frst estmaton wndow s January 1995-December The second s February 1995-January There are 150 estmaton wndows. The fnal estmaton wndow s July 2007-June Note: the average s computed over the total number of estmated Sharpe ratos (.e. total number of wdows). 405

9 406

10 To capture the tme-varyng exposure of the emergng ndustral stock markets to the global market rsk factor, we estmate a standard one-factor asset prcng lnear model n a rollng-wndow framework. We focus on the standard formulaton put forward by Sharpe (1964) and Lntner (1965). In partcular, as n Black et al. (1972), we consder the followng lnear regresson k, t k, t k, m t k t (3.3) ExR, where ExR k, t s the excess return of ndustry n country k, t the excess return on the market (.e. R R ), m f k, t and k, m are the regresson parameters, and k, t s the error term. The ntercept, k, t, measures the average monthly abnormal excess return (.e. Jensen s alpha). 9 The frst smple way to check f the CAPM holds, t s to run a test of sgnfcance on the ntercept of ths lnear regresson, proceedng asset by asset. In lne wth the purpose of ths study, we do not focus on testng the valdty of the model. Instead, we focus on the average estmated coeffcent k, t across emergng and US ndustral stock markets. The ntercept s a value extensvely employed n fnance to evaluate the performance of asset and fund managers. In partcular, t ndcates whether or not on average the observed returns on an asset are larger (or smaller) than the value consstent wth the CAPM. The sample perod goes from January 1995 (or later) to June 2012 and the market excess return s from the Kenneth French Data Lbrary. 10 We estmate Eq. (3.3), va standard OLS, n a tme-varyng context. 11 In partcular, the parameters are estmated usng a rollng wndow of 60 months. k, t and k, m 9 See Jensen, M.C., The performance of mutual funds n the perod Journal of Fnance, 23(2): The excess return on the market R m R f s publcly avalable at k m 11 Formally, k ˆ ˆ, k k ; ˆ z ˆ k zm ˆ 2 m 407

11 Defnton 4: The estmated alpha average spread s gven by Alpha Avg Spread ke 1 W W w 1 ˆ 1 w W ke, w ˆ US, w w 1 where denotes the ndustry, k e denotes the emergng country, and w defne the wndow n whch the ntercept (α) s estmated. The sample counterparts of Def. 4 are reported n Table 3.3. We fnd that Brazl, Chna and Turkey delver the hghest annual rollng average Jensen s alpha spreads (.e , and , respectvely). In Chna and Turkey, the healthcare ndustry dsplays the greatest ndvdual performance. The technology sector seems to domnate n Brazl, Chle and Turkey. Outstandng performances are acheved also by the Basc Materals and Consumer Goods ndustres n Brazl, Chna, Israel, Thaland and Turkey. The utltes sector has the hghest alpha average spread n Sngapore. The results are obvously sample senstve. Our CAPM alpha spreads seem to support the average performances presented n Table 3.1. Not surprsngly, the healthcare ndustry delvers the hghest cross-country average alpha spread (.e ). Table-3.3. Average Alpha Spread. The US stock market s used as benchmark. Intercepts of Eq. (3.3) are estmated va standard OLS usng a rollng sample of 60 months. Estmated values are annualzed. Standard errors are Newey and West (1987). The sample goes from January 1995 (or later) to June BRAZIL CHILE CHINA ISR MAL SING THAI TUR Basc Materals Cons. Goods Cons. Servces Fnancals Health Care na Industrals Ol & Gas Technology na Telecom Utltes Avg Mn Max where k z and z m are the sample means for the excess returns of ndustry (n country k) and the market portfolo, respectvely, whle and 2 ˆ m sample varance of the market portfolo return. k ˆ, m s the sample covarance between the two excess returns 408

12 Internatonal Busness Cycles and Stock Market Co-Movements The effects of the global ntegraton process on emergng stock market returns have been wdely dscussed n lterature. Fg. 4.1 also suggests that the nternatonal economes are becomng ncreasngly ntegrated. It s popularly beleved that fnancal and economc ntegraton are strctly related. 12 Recent studes document a reducton n cross-country dversfcaton benefts. Fg. C.1 reports the dynamcs of the correlaton coeffcents between the emergng and the US ndustral stock market excess returns. At the ndustry level, we observe, ex-post, that few cross-ndustry dversfcaton benefts can be exploted. Fg 4.2 reports the dynamcs of the market beta (estmated va Eq. (3.3)). In contrast to exstng emprcal fndngs (see (Grootveld and Salomons, 2003), manly based on natonal stock market ndces, we fnd that some ndustres cannot be classfed as hgh-beta stock markets. In few countres, these ndustres pay less than the market average excess returns. The result s dynamcally consstent. Betas are found to be constantly less than one n the fnancals and utltes sectors n the followng countres: Chle, Malaysa and Israel. At the begnnng of the sample, all emergng betas, n all emergng economes, are less than one (and negatve n few cases). In all subplots of Fg. 4.2, the dashed black lne represents the estmated US ndustral stock market betas. As expected, we fnd that the US betas are less volatle than the emergng betas. Fgure-4.1. Internatonal Busness Cycles. Ths fgure shows the cyclcal components of the nomnal (US$) GDP growth rate for the Unted States, Brazl, Chle, Chna, Israel, Malaysa, Thaland and Turkey. Cyclcal components are extracted usng the Hodrck and Prescott (1980) busness cycle flter. Followng emprcal practce, we use a smoothng parameter λ = 100. Data are annual and run from 1995 to Source: IMF 12 See, for example, Phylakts, K. and F. Ravazzolo, Measurng fnancal and economc ntegraton wth equty prces n emergng markets. Journal of Internatonal Money and Fnance, 21:

13 Our emprcal regulartes are key, and provdes practcal mplcatons. Frst, we observe that some cross-ndustry dversfcaton opportuntes should be exploted. In contrast, recent emprcal fndngs suggest that natonal ndexes tend to be much more exposed to global factors. Second, we fnd that some ndustres carry a small, but explotable, nsurance component. We refer to those ndustres that show a low-beta. Gven that the market represents an ndcator of the state of the economy, assets that are less volatle than the market wll be much more desrable n bad states of the economy. To hold these assets (.e. to be less exposed to bad states), an nternatonal nvestor has to pay a premum (.e. lower return). To conclude, we argue that the presence of a strong tmevaryng component n the dynamcs of the rsk exposure (.e. ndustral stock market betas) mght have strong mplcatons for the estmaton of the cost of captal. Fgure-4.2. Industry-by-Industry Rollng Betas (One-Factor Model). The beta parameter of Eq. (3.3) are estmated, va standard OLS, usng a rollng sample of 60 months. Standard errors are Newey and West (1987). Sample: January 1995 (or later) June

14 CONCLUSION Over the last twenty years, and especally after lberalzatons, emergng stock markets have captured the attenton of many scholars as well as many practtoners. Emergng markets emprcal regulartes are well known (e.g. hgh returns, hgh volatlty, tme-varyng moments). We mprove the exstng lterature by provdng an ex-post country-by-country and ndustry-by-ndustry standard performance analyss. We obtan two man results. Frst, we show that some emergng ndustral stock markets are more generous than others. In partcular, we fnd that the average extra prema pad by emergng markets to nternatonal nvestors has been manly drven by the healthcare sector. Second, we fnd that some ndustres stll ncorporate dversfcaton benefts as 411

15 well as nsurance propertes. Our results have mplcatons for fnancal applcatons. In partcular, they provde nsghts for the mplementaton of portfolo dversfcaton strateges and the estmaton of the cost of captal. ACKNOWLEDGEMENTS We are grateful to Alessa Varan for excellent research assstance. We thank Gudo Cazzavllan, Lauren Persha, and Paolo Vtale for ther suggestons, and one anonymous revewer. Any errors and omssons are of course our own. REFERENCES Ang, A., R.J. Hodrck, Y. Xng and X. Zhang, Hgh dosyncratc volatlty and low returns: Internatonal and further u.s. Evdence. Journal of Fnancal Economcs, 9: Barry, C.B., I. Peavy, J.W. and M. Rodrguez, A convenent way to nvest n emergng markets. Emergng Markets Quarterly, 1(1): Bekaert, G., Market ntegraton and nvestment barrers n emergng equty markets. World Bank Economc Revew, 9: Bekaert, G., B.E. Claude, C.R. Harvey and T.E. Vskanta, Dstrbutonal characterstcs of emergng market returns and asset allocaton. Journal of Portfolo Management: Bekaert, G. and C.R. Harvey, Emergng equty market volatlty. Journal of Fnancal Economcs, 43: Bekaert, G. and C.R. Harvey, Captal flows and the behavor of emergng market equty returns. Nber chapters. Captal Flows and the Emergng Economes: Black, F., M.C. Jensen and M. Scholes, The captal asset prcng model: Some emprcal tests. New York: Studes n the Theory of Captal Markets, Praeger. Brooks, R. and N. Del, M., Internatonal stock returns and market ntegraton: A regonal perspectve. Reserve Bank of Atlanta Workng Paper. Carrer, F., V. Errunza and K. Hogan, Characterzng world market ntegraton through tme. Journal of Fnancal and Quanttatve Analyss, 42(4): Claessens, S., S. Dasgupta and J. Glen, Return behavor n emergng stock markets. World Bank Economc Revew 9(1): De Jong, F. and F.A. De Roon, Tme-varyng market ntegraton and expected returns n emergng markets. Journal of Fnancal Economcs, 78:

16 Donadell, M. and L. Prosper, 2012a. The equty rsk premum: Emprcal evdence from emergng markets. CASMEF Workng Paper. Donadell, M. and L. Prosper, 2012b. The equty premum puzzle: Ptfalls n estmatng the coeffcent of relatve rsk averson. Journal of Appled Fnance & Bankng, 2(2): Donadell, M. and L. Prosper, 2012c. On the role of lqudty n emergng markets stock prces. Research n Economcs, 66(4): Elng, E., B. Gerard, P. Hllon and F.A. de Roon, Internatonal portfolo dversfcaton: Currency, ndustry and country effect revsed. Journal of Internatonal Money and Fnance, 31: Grootveld, H. and R. Salomons, The equty rsk premum: Emergng vs. Developed markets. Emergng Markets Revew, 4(2): Henry, P., Stock market lberalzaton, economc reform and emergng market equty prces. Journal of Fnance, 55(2): Hodrck, R.J. and E.C. Prescott, Postwar u.s. Busness cycles: An emprcal nvestgaton. Carnege Mellon Unversty dscusson. Hye, Q.M.A. and S. Wzarat, Impact of fnancal lberalzaton on economc growth: A case study of pakstan. Asan Economc and Fnancal Revew, 3(2): Jensen, M.C., The performance of mutual funds n the perod Journal of Fnance, 23(2): Karadagl, E.C., The effects of globalzaton on frm performance n emergng markets: Evdence from emergng-7 countres. Asan Economc and Fnancal Revew, 2(7): Kefela, G.T., Drvng forces of globalzaton n emergng market economes and developng countres. Asan Economc and Fnancal Revew, 1(2): Lntner, J., The valuaton of rsk assets and the selecton of rsky nvestments n stock portfolos and captal budgets. Revew of Economcs and Statstcs, 47(1): Newey, W.K. and K.D. West, A smple, postve sem-defnte, heteroskedastcty and autocorrelaton consstent covarance matrx. Econometrca, 55(3): Phylakts, K. and F. Ravazzolo, Measurng fnancal and economc ntegraton wth equty prces n emergng markets. Journal of Internatonal Money and Fnance, 21: Roll, R., Industral structure and the comparatve behavor of nternatonal stock market ndces. Journal of Fnance, 42: Serra, A.P., Country and ndustry factors n returns: Evdence from emergng markets stocks. Emergng Markets Revew, 1:

17 Sharpe, W.F., Captal asset prces - a theory of market equlbrum under condtons of rsk. Journal of Fnance, 19(3): APPENDIX A. Data Table-A.1. Datastream Global Equty Indces (DGEI). All returns are denomnated n US$. DGEI break down nto sx levels. Level 1 s the market ndex, ths covers all the sectors n each regon or country. Level 2 dvdes the market nto 10 ndustres and covers all the sectors wthn each group n each regon or country. Levels 3-6 subdvde the level 2 classfcatons nto sector classfcatons n ncreasng detal. Sample: December 1994 (or later) - June Source: Datastream. Industral Stock Market Indces Perod Industral Stock Market Indces Perod US-DS Ol & Gas ARGENTINA-DS Ol & Gas US-DS Basc Mats ARGENTINA-DS Basc Mats US-DS Consumer Gds ARGENTINA-DS Consumer Gds US-DS Consumer Svs ARGENTINA-DS Consumer Svs US-DS Industrals ARGENTINA-DS Industrals US-DS Health Care ARGENTINA-DS Fnancals US-DS Fnancals ARGENTINA-DS Telecom US-DS Telecom ARGENTINA-DS Utltes US-DS Technology BRAZIL-DS Ol & Gas US-DS Utltes BRAZIL-DS Basc Mats HONG KONG-DS BRAZIL-DS Consumer Gds Ol & Gas HONG KONG-DS BRAZIL-DS Consumer Svs Basc Mats Feb 02 Jun 12 HONG KONG-DS Consumer Gds BRAZIL-DS Industrals HONG KONG-DS BRAZIL-DS Health Care Consumer Svs Nov 07 Jun 12 HONG KONG-DS BRAZIL-DS Fnancals Industrals HONG KONG-DS BRAZIL-DS Telecom Fnancals HONG KONG-DS BRAZIL-DS Technology Telecom Mar 06 Jun 12 HONG KONG-DS BRAZIL-DS Utltes Technology HONG KONG-DS CHILE-DS Ol & Gas Utltes ISRAEL-DS Ol & CHILE-DS Basc Mats Gas 414

18 ISRAEL-DS Basc Mats ISRAEL-DS Consumer Gds Jan 98 Jun 12 ISRAEL-DS Consumer Svs Jan 98 Jun 12 ISRAEL-DS Industrals ISRAEL-DS Health Care ISRAEL-DS Fnancals ISRAEL-DS Telecom ISRAEL-DS Technology ISRAEL-DS Utltes Jan 98 Jun 12 SINGAPORE-DS Ol & Gas SINGAPORE-DS Basc Mats SINGAPORE-DS Consumer Gds SINGAPORE-DS Consumer Svs SINGAPORE-DS Industrals SINGAPORE-DS Health Care SINGAPORE-DS Fnancals SINGAPORE-DS Telecom SINGAPORE-DS Technology SINGAPORE-DS Utltes Jan 01 Jun 12 TAIWAN-DS Ol & Gas Jan 04 Jun 12 TAIWAN-DS Basc Mats TAIWAN-DS Consumer Gds TAIWAN-DS Consumer Svs TAIWAN-DS Industrals TAIWAN-DS Fnancals TAIWAN-DS Telecom Sep 00 Jun 12 CHILE-DS Consumer Gds CHILE-DS Consumer Svs CHILE-DS Industrals CHILE-DS Health Care CHILE-DS Fnancals CHILE-DS Telecom CHILE-DS Technology CHILE-DS Utltes COLOMBIA-DS Ol & Gas COLOMBIA-DS Basc Mats COLOMBIA-DS Consumer Gds COLOMBIA-DS Consumer Svs COLOMBIA-DS Industrals COLOMBIA-DS Fnancals COLOMBIA-DS Telecom COLOMBIA-DS Utltes MEXICO-DS Basc Mats MEXICO-DS Consumer Gds MEXICO-DS Consumer Svs MEXICO-DS Industrals MEXICO-DS Health Care MEXICO-DS Fnancals MEXICO-DS Telecom PERU-DS Ol & Gas PERU-DS Basc Mats PERU-DS Consumer Gds Jan 98 Jun 12 Oct 03 Jun 12 Jul 98 Jun 12 Avr 04 Jun

19 TAIWAN-DS PERU-DS Consumer Svs Technology Mar 96 Jun 12 Jan 01 Jun 12 CHINA-DS Ol & PERU-DS Industrals Gas CHINA-DS Basc PERU-DS Fnancals Mats Dec 99 Jun 12 CHINA-DS PERU-DS Telecom Consumer Gds CHINA-DS PERU-DS Utltes Consumer Svs Aug 96 Jun 12 CHINA-DS CZECH REP.-DS Consumer Industrals Gds CHINA-DS Health CZECH REP.-DS Consumer Care Mar 04 Jun 12 Svs Mar 95 Jun 12 CHINA-DS CZECH REP.-DS Industrals Fnancals CHINA-DS CZECH REP.-DS Fnancals Telecom Nov 02 Jun 12 CHINA A-DS CZECH REP.-DS Telecom Technology Jun 08 Jun 12 Mar 95 Jun 12 CHINA-DS Utltes Jul 95 Jun 12 CZECH REP.-DS Utltes INDIA-DS Ol & HUNGARY-DS Ol & Gas Gas Dec 95 Jun 12 INDIA-DS Basc HUNGARY-DS Basc Mats Mats INDIA-DS HUNGARY-DS Consumer Gds Consumer Gds INDIA-DS HUNGARY-DS Industrals Industrals May 97 Jun 12 INDIA-DS Health HUNGARY-DS Health Care Care INDIA-DS HUNGARY-DS Fnancals Fnancals INDIA-DS Telecom HUNGARY-DS Telecom Nov 97 Jun 12 INDIA-DS HUNGARY-DS Technology Technology May 99 Jun 12 INDIA-DS Utltes HUNGARY-DS Utltes INDONESIA-DS POLAND-DS Ol & Gas Basc Mats Feb 06 Jun 12 INDONESIA-DS POLAND-DS Basc Mats - Consumer Gds INDONESIA-DS POLAND-DS Consumer Gds Consumer Svs Jul 07 Jun 12 Jul 96 Jun 12 INDONESIA-DS POLAND-DS Consumer Svs Industrals Nov 95 Jun 12 INDONESIA-DS POLAND-DS Industrals Health Care Sep 96 Jun 12 INDONESIA-DS POLAND-DS Fnancals Fnancals INDONESIA-DS POLAND-DS Telecom Telecom Nov 98 Jun 12 INDONESIA-DS Jul 09 Jun 12 POLAND-DS Technology Feb 98 Jun

20 Technology INDONESIA-DS Utltes Dec 03 Jun 12 MALAYSIA-DS Ol & Gas MALAYSIA-DS Basc Mats MALAYSIA-DS Consumer Gds MALAYSIA-DS Consumer Svs MALAYSIA-DS Industrals MALAYSIA-DS Health Care Avr 01 Jun 12 MALAYSIA-DS Fnancals MALAYSIA-DS Telecom MALAYSIA-DS Technology Mar 10 Jun 12 MALAYSIA-DS Utltes PAKISTAN-DS Ol & Gas PAKISTAN-DS Basc Mats PAKISTAN-DS Consumer Gds PAKISTAN-DS Consumer Svs PAKISTAN-DS Industrals PAKISTAN-DS Health Care PAKISTAN-DS Fnancals PAKISTAN-DS Telecom PAKISTAN-DS Utltes PHILIPPINE-DS Ol & Gas PHILIPPINE-DS Basc Mats PHILIPPINE-DS Consumer Gds PHILIPPINE-DS Consumer Svs PHILIPPINE-DS Industrals Asan Economc and Fnancal Revew, 2013, 3(4): POLAND-DS Utltes RUSSIA-DS Ol & Gas RUSSIA-DS Basc Mats RUSSIA-DS Consumer Gds RUSSIA-DS Consumer Svs RUSSIA-DS Industrals RUSSIA-DS Health Care RUSSIA-DS Fnancals RUSSIA-DS Telecom RUSSIA-DS Utltes SOUTH AFRI-DS Ol & Gas Nov 00 Jun 12 Feb 98 Jun 12 Nov 01 Jun 12 Jan 03 Jun 12 Avr 99 Jun 12 Feb 05 Jun 12 Sep 07 Jun 12 Avr 98 Jun 12 Feb 98 Jun 12 Feb 98 Jun 13 SOUTH AFRI-DS Basc Mats SOUTH AFRI-DS Consumer Gds SOUTH AFRI-DS Consumer Svs SOUTH AFRI-DS Industrals SOUTH AFRI-DS Health Care SOUTH AFRI-DS Fnancals SOUTH AFRI-DS Telecom TURKEY-DS Ol & Gas TURKEY-DS Basc Mats TURKEY-DS Consumer Gds TURKEY-DS Consumer Svs TURKEY-DS Industrals TURKEY-DS Health Care Jan 96 Jun 12 Jul 00 Jun 12 TURKEY-DS Fnancals PHILIPPINE-DS TURKEY-DS Telecom 417

21 Fnancals PHILIPPINE-DS Telecom PHILIPPINE-DS Utltes SRI LANKA-DS Ol & Gas Nov 96 Jun 12 SRI LANKA-DS Consumer Gds SRI LANKA-DS Consumer Svs SRI LANKA-DS Industrals SRI LANKA-DS Fnancals SRI LANKA-DS Telecom Jan 03 Jun 12 SRI LANKA-DS Technology Jul 11- Jun 12 THAILAND-DS Basc Mats THAILAND-DS Consumer Gds THAILAND-DS Consumer Svs THAILAND-DS Industrals THAILAND-DS Health Care THAILAND-DS Fnancals THAILAND-DS Telecom THAILAND-DS Technology THAILAND-DS Utltes Jan 95 Jun 12 TURKEY-DS Technology TURKEY-DS Utltes B. Summary Statstcs Table-B.1. Excess Returns: Summary Statstcs. Statstcs are computed for 10 dfferent ndustres 23 dfferent countres. For each country, the frst and second lne report the mean and standard devaton of the ndustres excess returns. Lne 3 shows the average Sharpe ratos. For emergng markets only, the last lne provde the average spread between the emergng and the US stock excess returns, computed as n Def. 1. Values are computed on monthly bass and expressed n percentage ponts. The sample goes from January 1995 (or later) to June Count ry Ol&Gas BasMa ts ConsGd s ConsS vs Indust r HC Fn Telec Te ch Utlt 418

22 USA ARG n/a n/a n/a n/a n/a n/a n/a n/a -0.7 BRA ZIL CHIL E CHIN A CZ REP n/a n/a n/a n/a n/a n/a n/a n/a 0.9 HK n/a n/a n/a n/a HUN n/a n/a n/a n/a INDI A n/a n/a n/a

23 n/a ISRA EL MAL MEX n/a n/a n/a 1.4 n/a n/a n/a 9.4 n/a n/a n/a 15.0 n/a n/a n/a 0.9 PAK n/a n/a n/a n/a 0.0 PERU n/a n/a n/a n/a n/a n/a n/a n/a 0.2 PHIL n/a n/a n/a n/a n/a n/a n/a n/a 0.1 POL n/a n/a n/a n/a RUSS IA n/a n/a n/a n/a 1.3 SING SA n/a n/a n/a n/a n/a n/a n/a n/a 420

24 SrLa nka 1.8 n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a -9.8 TAIW AN n/a n/a n/a n/a n/a n/a n/a n/a THAI TUR KEY C. Internatonal Stock Markets Co-Movements Fgure-C.1. Industry-by-Industry Rollng Correlaton Coeffcents. Correlaton coeffcents are estmated usng a rollng sample of 60 months. The US ndustral stock markets are used as benchmark. The sample perod goes from January 1995 (or later) to June

25 422