DO STOCK MARKETS HAVE ANY IMPACT ON REAL ECONOMIC ACTIVITY?

Transcription

1 ACTA UNIVERSITATIS AGRICULTURAE ET SILVICULTURAE MENDELIANAE BRUNENSIS Volue Nuber 1, DO STOCK MARKETS HAVE ANY IMPACT ON REAL ECONOMIC ACTIVITY? Kateřina Krchnivá 1 1 Departent of Accounting and Taxation, Faculty of Business and Econoics, Mendel University in Brno, Zeědělská 1, Brno, Czech Republic Abstract KRCHNIVÁ KATEŘINA Do Stock Markets Have Any Ipact on Real Econoic Activity? Acta Universitatis Agriculturae et Silviculturae Mendelianae Brunensis, 64(1): There is no doubt about the existence of connection between the stock arkets and the real econoic activity. Many researchers indicated that the stock arkets causally affect the econoic activity with the lag of three onths. Contrary, the other group of researchers suggested, that these relationships are reversal, oreover soe of the concluded that these relationships are reciprocal. The paper analyses the relationship between the stock arkets and the econoic activity in seven countries with the research objective to identify these relationships in relation of cause and effect. As the proxy of the stock arkets are stock indices considered, while the econoic activity is expressed by the Gross Doestic Product at constant prices. The correlation analysis and the Granger causality test applied on suggested vector autoregressive odels are eployed for the research in the paper. The paper concludes that stock arkets ay be considered as the significant predictor of the real econoic activity with the lag of one quarter, however, there is no reciprocal links between the. Keywords: econoic activity, correlations analysis, the Granger causality, gross doestic product, utual and unilateral relations, stock indices, stock arkets, vector autoregressive odels INTRODUCTION It is unquestionable that the stock arkets are closely linked with the real econoic activity throughout a variety of channels. However, the direction of these links is not clearly identified by the theory. For exaple, Morck et al. (1990) identified five ain channels of the relation between stock prices and the real econoic activity. Their results are supported by the fact that the investent decisions of anagers of big corporations are influenced by the inforation provided by the stock arkets. Moreover, they also indicated that stock prices reflect the present value of paid dividends in the future. A nuber of different clais about the relationship between stock arkets and the econoic activity were provided epirically. Soe researchers (Faa, 1981; Schwert, 1990; Mauro, 1995) confired the existence of this relationship, contrary the other group of researchers (Binswanger, 2000 and 2004; Mao and Wu, 2007) argued that any possible relationships between stock arkets and the econoic activity were breached at the beginning of the eighties of 20 century and therefore any change in stock prices cannot be explained on the basis of changes in the econoy. In a nuber of studies are also analysed the causal relationships between the stock arkets and the econoy. Deetriades and Adriánová (2003) evaluated the direction of these causal links. One possibility is that the financial arket reacts to the econoic developent of a respective country. If the real econoy grows, the volue of savings in the financial syste grows as well. Thus, the financial syste is able to provide ore financial sources to those who need the. The other possibility is that the well-functioning financial arket stiulates the econoic growth. In accordance to these two different possible links reciprocal causal relationships ay be expected. However, any causal effect of financial arkets on the econoic activity were not unabiguously proved. This is probably caused by any factors which ay be as follows: investents into unproductive activities as the result of icroeconoic inefficiency of the banking syste whereas the banks are unable 283

2 284 Kateřina Krchnivá to solve the probles with the transission of inforation quickly and effectively. An alternative explanation for inefficiency of banking activities ay be the inappropriate political interference in the banking syste. The causal effect of the financial arket on the econoic developent ay also indicate soe acroeconoic probles, such as a high degree of political or econoic uncertainty appearing for exaple in the for of unpredictable inflation. Many authors focused on the analysis of the causal relationships between financial arkets and the econoy within one country rather than within ultiple set of countries. Mao and Wu (2007) identified a long-ter utual causal relationship between the Australian stock arket and the econoic activity, while Faa (1990), Schwert (1990) and Mauro (1995) argued that increasing perforance of the stock arkets ay have beneficial effects for the enhancing of econoic efficiency. However, studies eploying the Norwegian (Gjerde and Saette, 1999) or the Korean data (Know and Shin, 1999) rejected that the econoic output is deterined by the efficiency on the stock arket. In general, the results of researchers eploying the data of one country differ according to the econoic developent of a respective country. For exaple, Kaplan (2008) eploying the Turkish data identified a long-ter relationship between the stock arket and the econoy and proved that the stock arket has the causal effect on the econoic output. Based on his results the stock arket ay be considered as the ain deterinant of the future econoic activity. Adaopoulos (2010) analysing the relationship between the Geran econoy and the stock arket concluded siilarly. However, the results of Kaplan study (2008) were refused by ore recent study of Goktas and Hepsag (2011). They confired the existence of a causal relationship between stock arket s outputs and the econoic activity with the lag of six onths, however the identified direction is fro the econoy to the stock arket output These controversial results ay be caused by different ethodology; since Goktas and Hepsag (2011) took into consideration the seasonal behaviour of the tie series and therefore utilized the HEGY seasonal unit root test, while Kaplan (2008) did not. As was entioned above, the results of studies analysing the direction of the relationship between stock arkets and the econoy ay differ depending on the econoic developent of a respective country. In accordance to that the authors recently switched their interest fro developed countries to eerging ones. Aong the, Chakraborty (2008) exained the relationship between the Indian financial arket and econoy. He concluded that the developent of the Indian financial arket affects the econoic activity, but this relationship is relatively weak. Further, Ibrahi (2010) stated for the Malaysian saple that the stocks returns ay be considered as the ain indicator of the future econoic activity, but only for the period shorter of one year. The influence of stock arkets of five European countries (Gerany, France, Netherlands, Italy and the Great Britain) on their econoic efficiency was analysed by Siliverstovs and Duong (2006). Although the identified links were not statistically significant, they noted that the econoic developent positively responds to positive shocks taking place on the stock arket and also that the stock arkets of analysed countries iply certain inforation for the forecast of the econoic activity. Siilarly, Doain and Louton (1997) analysing the causal relationship between stock arket returns and the econoic efficiency taking into account the asyetry of the business cycle noted that the sharp declines in the econoic perforance are caused by the decline on stock arket returns, whereas the positive perforance of the stocks arket is followed by the increase of econoic output. Faa (1990) and Binswanger (2000) in their studies also suggested that onthly data have a little explanatory power for the analysis of the relationship between stock arkets and the econoic activity and quarterly or annual data see to be ore appropriate. Ibrahi (2010) also noted that irrespective to the fact that any studies contest the predictive ability of stock prices, their easy availability and exact easureents favouring the as the predictive indicator in coparison with other acroeconoic variables which are usually available with a long-ter delay and whose values have to be often odified. MATERIAL AND METHODOLOGY The paper analyses the relationship between the stock arkets and the real econoic activity with focusing on the exaination of the direction of identified links. As the proxy of the econoic activity the Gross Doestic Product (GDP) at constant prices of the year 2005 are eployed. By this preise the ethodological approach of corresponding researchers (Adaapoulos, 2010; Goktas and Hepsag, 2011) is followed. Moreover, this assuption is in the line with the hypothesis indicating the relationship between the stock arkets and the real econoic activity. The stock arkets are represented by the values of stock indices at the daily closing values. For the stock indices the effect of nonsynchronous trading is not taken under the consideration. The paper eploys the data of seven countries, where the European Union (EU27) is considered as one country, further soe of the European Union Meber States (the Czech Republic, Gerany, Poland, Hungary) are individually analyses; lastly the data of the United States of Aerica (USA) and Japan are included. The selection of the countries was perfored with the ai to analyse the relationship between

3 Do Stock Markets Have Any Ipact on Real Econoic Activity? 285 the stock arkets and the econoy of the Czech Republic and its neighbouring countries and to confront the situation of the Czech Republic with the large worldwide stock arkets of the United States and Japan as well as in the context of the European Union Stock Market. As was already entioned the stock arkets are represented by the values of stock indices of the ost iportant stocks of these countries, naely by: Nikkey 225 Stock Average for Japan, Dow Jones Industrial Average for USA, Euro Stoxx 50 for EU27, Warsaw Stock Exchange Top 20 Index for Poland, Budapest Stock Exchange index for Hungary, PX for Czech Republic and Deutsche Borse AG Geran Stock Index for Gerany. The values of GDP at constant prices are available fro the database of the Organization for Econoic Co-operation and Developent (OECD), while the values of stock indices were obtained fro Blooberg database. The period under consideration covers the data of the first quarter of the year 2000 to the second quarter of The datasets of both variables are expressed on the quarterly basis, were the quarterly values of stock indices were calculated as the average of daily values of a respective quarter. As was already entioned, quarterly data have according to Faa (1990) or Binswanger (2000) better explanation power for the detection of exained relations in coparison with onthly, weekly or daily data. Moreover, the values of GDP are available only on quarterly basis. The analysis is based on the correlation analysis, which identifies and easures a relationship between two sets of variables. Although the correlation analysis researches the intensity of a respective relationship, does not exaine its direction in relation of cause and effect 1. For this purpose the arranged vector autoregressive odels (VAR odels) were exained by the Granger causality test 2. The proposed VAR odels were as follows: t 0 i t i i t j 1t i 1 j 1, (1) GDP GDP I u t 0 i t i i t j 2t i 1 j 1, (2) I I GDP u where in the first equation (1) the assuption that the current values of GDP (GDP t ) are explained be its values of previous period (GDP t i )and previous values of stock indices (I t j ) was analysed. And it the second equation (2) is assued that current values of stock indices (I t ) are explained by its values of previous period (I t i ) and previous values of GDP (GDP t i ). RESULTS AND DISCUSSION The paper analyses the relationships between the stock arkets and the real econoic activity of seven chosen countries with the research objective to identify these relationships in relation of cause and effect. In the first step, the values of stock indices were tested by X-12-ARIMA analysis for the identification of the significance of the seasonal coponent. With regard to the fact that the values of seasonally adjusted values significantly differ in coparison with unadjusted values in the further steps of analysis the seasonally adjusted values were eployed. The values of GDP were not tested for the identification of the seasonal coponent, since their values were directly obtained in seasonally adjusted fro OECD database. Further, with objective to avoid of the proble of the apparent correlation, which ay be caused by the developing trend or by the influence of the other variable, the trend coponent of both sets of data were eliinated by their transition on logarithic returns according to following relations: GDP t (l) = 100 (lngdp t lngdp t 1 ), (3) I t (l) = 100 (lni t lni t 1 ). (4) By this adjustent also the assuption of the Granger causality test requiring the stationarity of tie series were et. The fulfilling of this assuption were tested by the ACF graph and by ADF unit root test 3, while KPSS 4 unit root test was used of the verification of stationarity only in abiguous cases. The results of the ADF and KPSS test are available in Appendix 1. The second assuption of the Granger causality test, i.e. nonseries correlation of the error ter, was tested by the Lung-Box portanteaus statistic at the 5% significance level of the proposed VAR odels (1) and (2) as well as of the adjusted VAR odels as follows: t() 0 i t i() i t j() 1t i 1 j 1, (5) GDP l GDP l I l u t() 0 i t i() i t j() 2t i 1 j 1, (6) I l I l GDP l u where the values of GDP and stock indices (I) were transfored into the logarithic returns (according 1 For ore see Hindls (2007). 2 For ore see Gujarati (2003). 3 The Augented Dickey-Fuller test. 4 The Kwiatkowski-Phillips-Schidt-Shin tests.

4 286 Kateřina Krchnivá to above indicated forulas (3) and (4)), which were taken into consideration in the further research steps. With regard to the results of studies of Estrella and Mishkin (1996), indicating that the developent of the stock arkets affect the developent of the econoic with the advance of one up to three onths, and Goktas and Hepsag (2011) arguing that the econoic developent lead the stock arket prices with advance of six onths, the lag of one quarter or two were considered for the analysis. The verified research hypotheses were as follows: the developent on stock arkets is reflected by the developent of the econoy, indicated by GDP, with the lag of one or two quarter, i.e. with the lag of three or six onths. Also the reversal relationship indicating the ipact of the econoic activity on the developent of the stock arkets were analysed. Firstly, any possible relationships between both sets of variables, i.e. stock indices and GDP, were identified by the correlation analysis. The pair correlation coefficients were calculated for each pair of stock index and a respective GDP indicator without any lag as well as for the situation where one of the variables were lagged by one or two quarter. To be clear, the correlation analysis was applied on the adjusted data for the further analysis, i.e. on the transfored data into logarithic returns according to relation indicated by equations (3) and (4). The following table (Tab. I) shows the results of the correlation analysis. The list of abbreviation used for the presentation of the results in the paper is available in Appendix 2. The above stated table shows strong relations between the stocks arkets and the econoic activity for the situation, where the stock arkets are considered as the predictor of the econoic activity I: The pair correlation coefficients of logarithic returns of the quarterly values of GDP and the logarithic returns of quarterly average values of stock indices for the period of Q1:2000 Q2:2012, n = 49 BUX DJIA GDAX NKY225 PX SX5E WIG20 GDP_HU GDP_US GDP_GE GDP_JAP GDP_CR GDP_EU27 GDP_PL GDP (t 2) ; I (t) GDP (t 1) ; I (t) * GDP (t) ; I (t) ** *** *** *** *** *** *** GDP (t) ; I (t 1) *** *** *** *** *** *** *** GDP (t) ; I (t 2) *** ** *** *** *** in Gretl 1: The developent of logarithic returns of SX5E Index with the lag of one quarter and EU27 GDP in the period Q1:2000 Q2:2012

5 Do Stock Markets Have Any Ipact on Real Econoic Activity? 287 with advance of one or two quarters. The strongest relation is indicated for the Czech stock arket and econoy and the European Union. The below stated figure (Fig. 1) shows the progress of the relations between the European stock arket and econoy. The correlation analysis shows strong links between the developent perfored by the stock arkets and the econoic activity. In addition to the identified relations also the direction of these relations in ters of causal effects were analysed. The Granger causality test were applied on paraeters of proposed VAR odels in equations (5) and (6). The tested null hypotheses were as follows: the developent on the stock arkets affects the econoic developent in ters of the Granger causality (5) and the developent of the econoy is not influenced by perforance of the stock arkets in ters of the Granger causality (6). The Granger causality test was applied for proposed VAR odels with the lag of one or two quarter and further evaluated by p-value of Wald s F-statistic. The obtained results are available in below stated table (Tab. II). The direction of relations are indicated by direction of arrows. The above table shows, that there are statistically significant causal relation at 1% significance level in the direction fro the stock arkets to the econoic activity with the lag of one quarter for Hungary, Gerany, the Czech Republic and the European Union and at 5% significance level for Japan, the and Poland. In other words, the developent on the stock arkets of these countries deterines the developent in their econoy with advance of one quarter. The results correspond with the hypothesis that the stock arket leads the econoic activity. The siilar conclusion were also obtained by Kaplan (2008) and Adapolous (2010). The reversal relation were identified for the United States of Aerica. For the analysis of the Granger causality of VAR odels with the lag of two quarter the siilar results were identified, i.e. the direction of causal effect leading fro the arket to the econoy. In case of the Japanese data no relation between the stock arket and the econoy were detected at 10% significance level. The reversible relations were identified for Hungary, while no relations were identified for the United States of Aerica. Based on the obtained results the following question ay arise: Do the stock arkets lead the econoic activity with advance of one or two quarters? For identification, whether the lag of one or two quarter has ore conclusive for the considered saple, the evaluation based on the inforation criteria BIC, AIC and HQC 5 of the arrange VAR odels were used. The decision was derived fro the lower value of respective inforation criteria for each individual pair of variables, where the BIC and AIC inforation criteria were considered as the ain decisive, while the HQC inforation criterion as additional, i.e. decisive in the abiguous cases. The results are presented in below stated table (Tab. III), where only these pairs of variables, where the relation in the direction fro the stock arkets to econoic activity was identified, were taken into consideration. Fro the table (Tab. III) is observable, that the Geran, the European, the Czech and the Hungarian stock arkets indicate perforance of the econoy with advance of one quarter, while in case of Poland, the advance of two quarter is II: The results of Granger causality test for stock indices and GDP in the period Q1:2000 Q2:2012 with the lag of one and two quarter p-value Lag 1 p-value Lag 2 GDP_HU=>BUX ** BUX=>GDP_HU *** *** GDP_US=>DJIA ** DJIA=>GDP_US GDP_GE=>GDAX GDAX=>GDP_GE *** *** GDP_JAP=>NKY NKY225=>GDP_JAP ** * GDP_CR=>PX PX=>GDP_CR *** *** GDP_EU27=>SX5E SX5E=>GDP_EU *** *** GDP_PL=>WIG WIG20=>GDP_PL ** *** 5 Akeike inforation criterion, Bayesian inforation criterion and Hannan-Quinn inforation criterion.

6 288 Kateřina Krchnivá III: Analysis of the suitability of the length of lag VAR odel, Lag 1, AIC BIC HQC BUX=>GDP_HU GDAX=>GDP_GE PX=>GDP_CR SX5E=>GDP_EU WIG20=>GDP_PL VAR odel, Lag 2, AIC BIC HQC BUX=>GDP_HU GDAX=>GDP_GE PX=>GDP_CR SX5E=>GDP_EU WIG20=>GDP_PL ore appropriate. Based on the analysis of ore conclusive length of lag, it is possible stated, that for Geran, Hungary, the Czech Republic and the European Union are the results in the line with Estrella and Mishikn (1996). In case of the US stock arket, the reciprocal relation were identified, which is consistent with the conclusion Goktas and Hepsag (2011). Generally, based on the analysis perfored in the paper, ay be concluded that in five of seven considered countries the causal relation of stock arket and the econoic activity were detected. The results of the paper are in contrary with the conclusion of soe studies (Gjerde and Saette, 1999; Know and Shin, 1999) but on the other hand in accordance with any others (Schwert, 1990; Faa, 1990 or Adaopoulos 2010). This discrepancy ay be caused by the aturity or the size of a respective country and its stock arket. However, based on the results of the analysis carried out in the paper; in considered countries the stock indices ay be used as the iportant leading indicator of the econoic activity. This fact ay be beneficial within the forulation of an investent strategy; if stock prices grow in the long-ter, it is possible to assue that the investent will be profitable. The use of this theoretical approach ay be a very powerful tool at the prevention of loss-aking investent or the sudden failure of the businesses. Appendix 1 IV: Results of ADF tests, the seasonally adjusted tie series of stock indices and GDP at constant prices Variable Without constant With constant With constant and trend BUX DJIA * ** GDAX NKY SXE PX WIG GDP_CR GDP_EU GDP_GE * GDP_HU GDP_PL GDP_US GDP_JAP

7 Do Stock Markets Have Any Ipact on Real Econoic Activity? 289 V: Results of ADF and KPSS tests, the logarithic returns adjusted tie series of stock indices and GDP at constant prices ADF test KPSS test Without constant With constant With constant and trend t-stat BUX *** *** ** DJIA *** *** *** GDAX *** *** ** NKY *** *** *** SXE *** *** ** PX *** *** ** WIG *** *** ** GDP_CR ** ** * GDP_EU *** * *** GDP_GE *** *** *** GDP_HU * ** ** GDP_PL ** *** GDP_US *** *** *** GDP_JAP *** *** *** Appendix 2 VI: List of abbreviations The abbreviations for Stock Indices BUX Budapest Stock Exchange Index for Hungarian Stock Market DJIA Dow Jones Industrial Average for Aerican Stock Market GDAX Deutsche Borse AG Geran Stock Index for Geran Stock Market NKY225 Nikkei 225 Stock Average for Japanese Stock Market PX An official stock indices of Prague Stock Market SXE5 Dow Jones EURO STOXX 50 for the European Union Stock Market WIG20 Warsaw Stock Exchange Top 20 Index for Poland Stock Market The abbreviations for GDP GDP_CR Gross Doestic Product for the Czech Republic GDP_EU27 Gross Doestic Product for the European Union GDP_GE Gross Doestic Product for Gerany GDP_HU Gross Doestic Product for Hungary GDP_JAP Gross Doestic Product for Japan GDP_PL Gross Doestic Product for Poland GDP_US Gross Doestic Product for the United States CONCLUSION The paper analysed the relationship between the stock arkets and the real econoic activity. As a proxy of the econoic activity the Gross Doestic Product indicator at constant prices was considered, while the stock arkets were represented by the ain stock indices of the chosen countries (the United States of Aerica, Japan, Gerany, Poland, Hungary, the Czech Republic and the European Union). The hypothesis of the predictive ability of the stock arkets on the econoic activity was verified in the paper. However, based on the theoretical background the direction of this relation ay be reversal or even reciprocal. The analysis was based on the verification of the statistical significance of pair correlation coefficients. The direction of detected relation was tested by the Granger causality test of the proposed VAR odels. Based on the correlation analysis strong relations between the stock arkets and the econoic activity were identified. With regard to results of the Granger causality analysis ay be stock arkets represented by stock indices considered as the iportant leading indicator of the econoic developent.

8 290 Kateřina Krchnivá Fro the epirical results of correlation analysis is obvious that there are quite strong links between stock arket developent and econoic activity. Based on the analysis of Granger causality it is possible to consider stock arkets as an iportant leading indicator of econoic activity. The findings of the paper are in contradiction with the conclusion of soe studies but on the other hand in accordance with any others. The epirical findings of the research in the paper support the contention that the stock arkets leads the real econoic activity and there are no feedback relationships between these two indicators. REFERENCES ADAMOPOULOS, A Stock Market and Econoic Growth: An Epirical Analysis for Gerany. Business and Econoics Journal, BEJ-1. BINSWANGER, M Stock returns and real activity: Is there still a connection? Applied Financial Econoics, 10(4): BINSWANGER, M Stock returns and real activity in G-7 countries: Did the relationship change during the 1980 s? Quar. Rev. Econoic Finance, 44(2): BLOOMBERG.COM. [online] DAX:IND. Available at: < DAX:IND>. CHAKRABORTY, I Does financial developent cause econoic growth? The case of India. South Asia Econoic Journal, 1(9): DEMETRIADES, P., ADRIÁNOVÁ, S Finance and Growth: What We Know and What We Need to Know. In: GOODHART, C. (ed.), Financial Developent and Econoic Growth: Explaining the Links. Houndills: Palgrave Macillan. DOMAIN, D. L. and LOUTON, D. A A Threshold Autoregressive Analysis of Stock Returns and Real Econoic Activity. International Review of Econoics and Finance, 6: ESTRELLA, A., MISHKIN, F. S The Yield Curve as a Predictor of U.S. Recessions. Current Issues in Econoics and Finance, 2(7): 1 6. FAMA, E. F Stock Returns, Real Activity, Inflation, and Money. The Aerican Econoic Review, 71(4): FAMA, E. F Stock Returns, Expected Returns, and Real Activity. The Journal of Finance, 45(4): GJERDE, O., SAETTEM, F Causal relations aong stock returns and acroeconoic variables in a sall open econoy. J. Int. Financial Markets Inst. Money, 9: GOKTAS, O., HEPSAG, A Do stock returns lead real econoic activity? Evidence fro seasonal cointegration analysis. Econoics Bulletin, 31(3): GUJARATI, D. N Basic Econoetrics. 4 th ed. Singapore: McGraw-Hill. HINDLS, R., HRONOVÁ, S., SEGER, J Statics for Econoists [in Czech: Statistika pro ekonoy]. 8 th ed. Praha: Professional Publishing. IBRAHIM, M An Epirical Analysis of Real Activity and Stock Returns in an Eerging Market. Econoic Analysis & Policy, 40(2): KAPLAN, M The ipact of Stock Market on Real Econoic Activity: Evidence fro Turkey. Journal of Applied Sciences, 8(2): KNOW, C. S., SHIN, T. S Cointegration and causality between acroeconoic variables and stock arket returns. Global Finance Journal, 10: KRCHNIVÁ, K Study of interdependence of stock arkets and their ipact on the gross doestic product developent of chosen countries. Diploa Thesis. Brno: Mendel University in Brno. SCHWERT, W Stock returns and real activity: A century of evidence. The Journal of Finance, 45(4): MAURO, P Corruption and Growth. Quarterly Journal of Econoics, CX: MAO, Y., WU, R Does the stock arket act as a signal for real activity? Evidence fro Australia. Econ. Papers, 26: MORCK, R., SHLEIFER, A., VISHNY, Do anagerial objectives drive bad acquisitions? Journal of Finance, 45: OECD.ORG. [online] Quarterly National Accounts. Available at: < index.aspx?queryid=26674>. SILIVERSTOVS, B., DUONG, M. H., 2006: On the role of stock arket for real econoic activity: Evidence for Europe. DIW-Discussion paper. Contact inforation Kateřina Krchnivá: xkrchniv@node.endelu.cz, k.krchniva@gail.co