The statistical analysis of the relationship between Religion and macroeconomic indicators

Transcription

1 011 Internatonal Conference on Socalty and Economcs Development IPEDR vol.10 (011) (011) IACSIT Press, Sngapore The statstcal analyss of the relatonshp between Relgon and macroeconomc ndcators Aurelan Petruş Plopeanu 1 and Chrstana Bălan 1 PhD, Postdoctoral Researcher at the Faculty of Economcs and Busness Admnstraton, Alexandru Ioan Cuza Unversty of Ias, Romana, emal:aplopeanu@gmal.com PhD, Assstant Professor at the Faculty of Economcs and Busness Admnstraton, Alexandru Ioan Cuza Unversty of Ias, Romana, emal: chrstana.balan@uac.ro Abstract. In ths paper we am to verfy f, n the present context, t s verfed the Weber s hypothess concernng the relatonshp between relgous denomnaton and economc development. Usng a study conducted on 36 countres, the sample s observed by countres adherence to major relgons. The varance of the macroeconomc ndcators was syntheszed wth PCA method nto two latent varables. The dstrbuton of the scores for the two prncpal components was represented graphcally. The results have valdated the ntal hypothess, by the groups of countres dentfed usng the cluster analyss. Keywords: relgous denomnaton, female employment, drect nvestment abroad, actve populaton, Prncpal Component Analyss. 1. Introducton Ths paper ams to verfy f, n present context, one verfes the Weber s hypothess concernng the relatonshp between relgous denomnaton and economc development. Weber s theory consders that protestant countres decsvely and strongly nfluence the confguraton and the dynamc of labour market due to the paradgm assocated wth hard and legal work. The only way of ganng the lfe after death s workng hard and sparng no effort. Thus, the feature of the Calvnst ethc system s professonal work n communty servce, whch demonstd the superorty of ths socal organzaton. Regardng the effects of varous relgous denomnatons on the structure and dynamcs of the labour market, partcularly on female workforce, there are few studes n the lterature. Inglehart and Baker (000) stress that known relgous tradtons have hstorcally shaped the natonal culture of dfferent socetes. The fact that a socety has been shaped by Protestantsm or Confucansm or Islam leaves a cultural legacy wth lastng effects that nfluence future developments. Compton (1991) conducts a study about the nfluence of Catholc and Protestant relgous denomnatons on labour market, partcularly on unemployment. The results showed that, under specfc condtons, unemployment n Catholc communtes, whch follow the structure of Protestants characterstcs, remans hgher by 17%, and the reason s gven by endogenous factors that explan approxmately 80% of ths gap. Ewng (000) emprcally demonsts the effects of ndvduals growth and educaton n the sprt of Catholc prncples on future wages. Feldmann (007) founds that female labour force and female employment are hgher n Protestant countres than n countres wth other predomnant relgon.. Data and Methodology For a sample of 36 countres, whch range from major Orthodox relgon to Protestantsm, Catholcsm, Islam and Judasm, we observed a seres of macroeconomc ndcators used n the lterature for the analyss of economc growth (Barro, Sala--Martn, 003): Gross Domestc Product per capta, expressed n 48

2 Purchasng Power Party (current nternatonal dollars per capta), Employment of actve populaton, Unemployment, Actve populaton, Drect Investment Abroad (bllons of US dollars). The data we have used are taken from dfferent databases, lke Internatonal Monetary Fund, The World Bank, LABORSTA, OECD Employment Outlook, EUROSTAT (European Commsson Statstcs) and dfferent natonal government agences and nsttutons. The CIA database offered nformaton on populaton structure by relgous denomnatons. To dentfy the man determnants of economc growth, we appled Prncpal Components Analyss for summarzng the macroeconomc ndcators. Its objectve s to reduce a large set of varables to a small set contanng the most nformaton from the ntal set. (Escofer, B., Pages, J., 1998). P.C.A. requres the dentfcaton of the man k components (factoral axes) whch are lnear and uncorrelated combnatons of the orgnal varables. F k = a1 k X1 + ak X a pk X p, where X j are the ntal varables, wth j = 1, p. In order to dentfy homogenous groups of countres based on the observed varables, we used the herarchcal cluster analyss. The method starts wth clusters formed by one case that merge together untl they form one bg cluster. The frst step n the cluster analyss s the proxmty matrx (smlarty matrx) or the dstance matrx. The dstance between two cases s measured usng the Squared Eucldean dstance. It represents the geometrcal dstance n a multdmensonal space (the sum of the squared dfferences between all the varables on two cases): d Euc ( x, y ) = ( x y ) The dstances are calculated n order to group the cases nto clusters. The rule used for clusterng the cases s the Ward s method: the cluster membershp s based on the sum of the squared devatons from the cluster mean. The optmal number of clusters s decded on subjectve reasons: as the cases form groups, the clusters that merge are bgger and bgger and they are composed of more and more dssmlar cases. Moreover, n order to determne the dfference n employment by relgous denomnatons, we bult an ANOVA regresson model wth the followng expresson (Jaba, 00): Y = β + β D 0 + ε, where: Y s the dependent quanttatve varable (employment ); D s the ndependent dummy varables wth two possble values (1 - yes, 0 - no). The number of dummy varables s equal to the number of denomnatons mnus 1 (the denomnaton consdered for reference); β the regresson coeffcents; ε the resdual varable wth mean equal to zero and normally dstrbuted. The sgnfcance of the β stresses the mportance of the dfferences among confessons concernng the mean level for the output varable. 3. Man Fndngs The selected macroeconomc ndcators are correlated wth dfferent sgn and ntensty (Table 1). GDP per capta (PPP) suggests a postve causalty wth employment of actve populaton, Drect Investment abroad, employment of female populaton and a negatve dependence wth Unemployment. Drect Investment abroad (bllons of U.S. dollars) s postvely correlated only wth Actve populaton and GDP (PPP) per capta. Annual unemployment has no drect nfluence on other factors, but has strong and negatve relatonshp wth GDP (PPP) per capta, total employment and female employment. Actve populaton has a strong drect relatonshp drect nvestment abroad. Employment of actve populaton postvely nfluence female employment and GDP per capta n PPP; t s negatvely correlated wth the unemployment. Female employment s postvely nfluencng the employment of actve populaton and GDP per capta n PPP, whle t s nverse correlated wth the unemployment. GDP per capta PPP Drect nvestment Unemployment Actve populaton Employment Correlaton abroad GDP per capta PPP Drect nvestment abroad Unemployment Actve populaton Employment Female employment 49

3 Female employment Table 1: Correlaton matrx between the macroeconomc ndcators Orgnal varables are correlated sepaly wth the two man components (see table ): Varables GDP (PPP) per capta, employment and female employment are postvely and strongly correlated wth the frst factoral axs; The unemployment has a negatve and relatvely strong and correlaton wth the frst factoral axs; Actve populaton varables and Drect Investment abroad are postvely and closely correlated wth the second factoral axs. On the bass of the correlaton coeffcents value, showed n the obtaned correlaton matrx after the rotaton of axes, we defne two components: demographc and economc output (Component 1) and demographc and economc nput (Component ). Component 1 Employment Female employment GDP per capta PPP Unemployment Drect nvestment abroad Actve populaton Table. Rotated Component Matrx The countres are represented on the factoral map of the two prncpal components (see Fg. 1), thus allowng to dentfy clusters of countres accordng to two crtera, demographc and economc nput (DEI) and demographc and economc output (DEO). The frst component (DEO) dstngushes between two groups of countres, namely: a group of countres postvely correlated wth the frst factoral axs, the demographc-economc output, conssts of hghly developed countres (Unted Kngdom, Canada, Sweden, Iceland, Unted States of Amerca, Denmark, Fnland, New Zealand); a group of countres negatvely correlated wth the frst factoral axs s composed of countres that recorded low economc growth s (Mal, Iran, Egypt, Morocco, Kyrgyzstan, Turkey). Malaysa performs the best n ths cluster, beng an atypcal case. The two groups of countres dstngushed by the demographc-economc output factors are dfferent n terms of relgous confesson. The developed countres wthn the consdered group are Protestant (Unted Kngdom, Sweden, Iceland, Unted States of Amerca, Denmark, Fnland and New Zealand), except Canada, whch s Catholc, whle the poorest states, n terms of our hypotheses, defne almost all the Islamc countres. Also, most Catholc countres are postoned on the factoral map showng postve scores for the frst component, as these states have a hgher than average level of economc development. The Orthodox countres have a lower level of economc development than the Catholcs, but superor to Islamc states, wth most of the scores n the negatve sde of the frst axs. Fg.1. Dstrbuton of countres accordng to labor market performance and major relgon 50

4 The performance of the economc and labor market ndcators s dfferent accordng to countres relgous denomnaton (Fg. ) Regardng the demographc-economc nput factors, we dentfy a cluster of countres wth a sgnfcant volume of human captal resources (Unted States of Amerca, Indonesa, Brazl, Russa). These states are heterogeneous n terms of relgous afflaton of the populaton: the USA s preponderant Protestants (about 51.3% of populaton), Indonesa s an Islamc state (86.1% of populaton), Brazl s predomnantly Catholc (73.6% of the populaton) and Russa s manly Orthodox (71.8% of the populaton). A sepa cluster of countres s composed by the states wth a level of populaton above the average, namely Egypt, Iran, Turkey, Ukrane, Italy, France and Unted Kngdom. Also ths cluster shows a strong heterogenety n terms of relgous denomnaton. We analyze the dstrbuton of labor market ndcators by groups of countres dstngushed by a domnant relgon. Snce the studed varables are expressed n dfferent unts, n order to observe the dfferences between the consdered macroeconomc ndcators, whch are low compared wth the GDP varable, t s useful to standardze the varables. Among the studed varables, GDP dffers most between the fve clusters of countres determned by the relgous (Catholc, Orthodox, Protestant, Islam, and Judasm).Thus, the Protestant states record sgnfcantly hgher GDP s than the Orthodox and Islamc states and hgher than the Catholc ones. Israel s characterzed by a GDP rankng below the Protestant average and above the average of the Catholc states. Also, t s noted that the Catholc states present the greatest varaton of the GDP, compared to other clusters. The average GDP for the Catholc states s hgher than the Orthodox and Muslm states. There are mportant dfferences between the four clusters also due to the employment of the actve populaton. The hghest average level of ths ndcator s observed for the protestant cluster, whch s the most homogeneous, and the lowest level s recorded for the Islamc one, the most heterogeneous of the consdered clusters. The Catholc, Orthodox and Protestant countres record average values for the employment of the actve populaton above the average of actve populaton for the 36 countres sample, whle the average for Israel and Muslm countres s lower than that. Fg.. The dstrbuton of macroeconomc ndcators by relgon In the Protestant countres, the female employment records levels above the average of the Catholc, Orthodox, Protestant and Jewsh clusters. Moreover they are the most homogeneous countres. These results confrm the research hypothess, accordng to whch the female employment s the hghest for Protestant than all of the analyzed denomnatons. Moreover, we try to determne by how much the employment dffers among the man groups of relgons. To check ths hypothess, we bult an ANOVA regresson model: Y = β0 + β1*catholc + β*orthodox + β3*islam + β4*judac, wth Employment - dependent varable, Relgon - ndependent varable. Relgon s defned n the model by four dummy varables; Protestant denomnaton s consdered the reference category. 51

5 The estmated model s: Employment Rate = *Catholc *Orthodox 17.15*Islam *Judac (Sg=0.000) (Sg=0,03) (Sg=0,075) (Sg=0,000) (Sg=0,094) The dummy varable coeffcents are statstcally sgnfcant; therefore, the employment on the total populaton s sgnfcantly hgher n the protestant states than n predomnantly Catholc or Muslm ones. The ANOVA regresson model for female employment s: Female Employment = *Catholc 7.478*Orthodox 6.94*Islam *Judac (Sg=0.000) (Sg=0,05) (Sg=0,050) (Sg=0,000) (Sg=0,176) We observe a better performance of the protestant countres labor market, n terms of female employment. The Protestant states female employment s hgher by about 8% than n the Catholc countres and about 6.9% from the Islamc countres. 4. Conclusons For the labour market analyss, accordng to the domnant relgon of a state, we observed a sample of countres coverng fve groups where the predomnant relgon rangng from Catholcsm to Protestantsm and Orthodox, Islam and Judasm. Macroeconomc ndcators used n the analyss were dvded nto two categores: demographc-economc nput ndcators and demographc-economc output ndcators. The frst category of ndcators do not dstngush between the countres wth dfferent relgons; however, the ndcators of the second category of states dstngush two man groups: one group wth hgher performance n the labor market, the vast majorty of states beng Protestant and another wth underperformances of the labor market whch conssts manly of Muslm states. The best results recorded on the labor market by the Protestant countres compared wth the Catholc and Islamc states are confrmed both by analyzng the average level of demographc-economc ndcators for each group of countres dstngushed by the predomnant relgon; also, by testng the sgnfcance of dfferences between relgons on the employment for total and female populaton. 5. Acknowledgement Ths paper appeared wth the fnancal support under the project POSDRU/89/1.5/S/49944 Dezvoltarea capactă de novare creterea mpactulu cercetăr prn programe post-doctorale (Developng the Innovaton Capacty and Improvng the Impact of Research through Post-doctoral Programmes), funded by Sectoral Operatonal Program for Human Resources Development, n afflaton wth Alexandru Ioan Cuza Unversty of Ia. 6. References [1] Barro, R., Sala--Martn, X., Economc Growth, Second edton, The MIT Press, 003. [] Escofer, B., Pages, J., Analyses factorelles smple et multples, Dunod, [3] Feldmann, H., Protestantsm, Labor Force Partcpaton, and Employment Across Countres, Amercan Journal of Economcs and Socology, 007; [4] Guso, L., Sapenza P., Zngales L., Does Culture affect Economc Outcome?, Journal of Economc Perspectves, Vol. 0, No. 1, 006. [5] Inglehart, R., Baker, W. E., Modernzaton, Cultural Change, and the Persstence of Tradtonal Values, Amercan Socologcal Revew, Vol. 65, No. 1, 000. [6] Jaba, E., Statstcs. 3rd edton, Economc Publshng House, Bucharest, 00. [7] Weber, M., The Protestant Ethc and the Sprt of Captalsm, New York: Charles Scrbner s Sons;