GENDER AND CENTRAL BANKING

Transcription

1 GENDER AND CENTRAL BANKING Ibrahima Diouf, Dominique Pépin To cie his version: Ibrahima Diouf, Dominique Pépin. GENDER AND CENTRAL BANKING. Economic Modelling, Elsevier, 2017, 61, pp < /j.econmod >. <hal v3> HAL Id: hal hps://hal.archives-ouveres.fr/hal v3 Submied on 12 Apr 2017 HAL is a muli-disciplinary open access archive for he deposi and disseminaion of scienific research documens, wheher hey are published or no. The documens may come from eaching and research insiuions in France or abroad, or from public or privae research ceners. L archive ouvere pluridisciplinaire HAL, es desinée au dépô e à la diffusion de documens scienifiques de niveau recherche, publiés ou non, émanan des éablissemens d enseignemen e de recherche français ou érangers, des laboraoires publics ou privés.

2 Gender and Cenral Banking Ibrahima Diouf, Dominique Pépin ABSTRACT Female Cenral Bank chairs represen bu a iny minoriy. To undersand why, his aricle analyzes socio-economic and socio-poliical characerisics of he counries where women have chaired Cenral Banks. Then, i suggess ha gender differences in preferences as regards moneary policy goals may have some influence. An innovaive economeric mehodology, which does no require esimaing all he srucural and preference parameers of a moneary model, bu only how men and women s parameers differ, is developed o es his hypohesis. The resuls show ha female Cenral Bank chairs focus more han heir male counerpars on achieving he price sabiliy goal. This means, hen, ha women are more resisan han men o poliical pressures. Finally, i concludes ha gender differences in degree of conservaism may be an explanaory facor in female underrepresenaion in he Cenral Bank chairs. Keywords: Cenral Bank, conservaism, female, gender gap, moneary policy, preference parameers. JEL Codes: E52, E58, J16, Z10 1. Inroducion «Mr. Summers and Mr. Yellen... euh, Ms. Yellen». This slip of he ongue by Presiden Obama speaks volumes abou women s absence in he world of cenral bankers (Perlberg, 2013). As known, Cenral Banks are among he larges insiuions in modern economies. In charge of defining and implemening of moneary policy, hey deermine he sock of money in circulaion, se he level of shor raes and perform he essenial public funcions (managemen of foreign exchange reserves, banking supervision). Cenral Bankers acions have an immense impac on he funcioning of our economy. Bu, unil recenly, women were deeply underrepresened in Cenral Banks. I seems ha women have greaer access o posiions of responsibiliy in poliics han in Cenral Banks where he glass ceiling is ougher. For insance, in Europe, among he 23-member policy board and he six-member Execuive Board 1

3 of he European Cenral Bank (ECB), here are no women a all. Since is creaion, he ECB s Execuive Board has included only wo. The lieraure has jus begun o worry abou women s absence in moneary policy making (Chappell and McGregor, 2000). Mos of he exising sudies focus on he composiion of he Moneary Policy Commiees (MPC). They are relaed o he Federal Open Marke Commiee (FOMC) of he US Federal Reserve and he Board of he European Cenral Bank (ECB). The main idea is ha he diversiy in he MPC, in erms of gender, bu also of age, of professional, secoral and academic backgrounds, can influence he moneary policy decisions (Farvaque e al., 2014, Masciandaro e al., 2016). In paricular, i poins ou ha women s presence in he MPC could have an imporan effec on policy oucomes (Chappell e al., 2005; Farvaque e al., 2011; Bennani e al., 2015; Masciandaro e al., 2016). The findings obained by he empirical lieraure highlighed a higher share of women in he MPC is associaed posiively wih price sabiliy. In ohers words, in heir decisions, female moneary policymakers are raher inclined o figh agains inflaion. To explain his aiude, i is observed ha, from some level of responsibiliy, women seek o prove ha hey can be ougher han men. In paricular, as cenral bankers, hey need o make or o acquire a repuaion and credibiliy. Hence hey are invied o have hawkish raher han dovish aiude (Wilson, 2014; Hix e al., 2010; Farvaque e al., 2011, 2014; Eijffinger e al., 2015; Masciandaro e al., 2016). Thus, higher women s presence in MPC is associaed wih beer performance in erms of price sabiliy. In his paper, we ry o idenify he facors which may explain he absence of females Cenral Bank chairs. Indeed, in many counries, women have access o he highes poliical posiions as Head of Sae, Prime minisers. For insance, in Germany, France, England, Norway, women have been Prime minisers, bu, have never been appoined Cenral Bank chairwoman. Only a minoriy of women have managed o break he glass ceiling and reached he Cenral Bank chair. This presigious funcion remains monopolized by men. Nowadays, here is every indicaion Cenral Bank chairs are imporan people. In heir duies, hey have he masering of he moneary policy agenda (Riboni and Ruge-Murcia, 2008). This ends o show ha hey are powerful and influenial: powerful, because he Cenral Bank s independence proec hem from all kinds of poliical pressures; influenial, hrough he credi ha markes and governmens aribue o heir words and he consequences of heir decisions on he economy. This is why Ms. Jane Yellen is considered as he second mos powerful person afer he US Presiden (Gergen, 2013). Her appoinmen, in 2014, as he 2

4 firs female o chair he Federal Reserve Bank, i.e. he world's larges Cenral Bank, is definiely an essenial milesone for female represenaion in he world of cenral bankers, i.e. people who acually make moneary policy (Jones, 2013). So wha facors can accoun for he women s absence as Cenral Bank chairs? Obviously, here is no one specific facor. Formal or informal barriers are ofen poined ou, for example: A) The relucance of women for economics and finance sudies (Hale and Regev, 2001); B) The women s absence in he closed neworks of influence from which Cenral Bank chairs are chosen. The firs argumen is no consisen wih he facs. As noed by he websie Worldwide Guide o Women in Leadership, beween 2000 and 2015, over 200 women headed he Minisry of economy, finance and/or budge in heir counry or erriory. In early 2015, five are sill in office, in: Eas Timor; Lesoho; Republic of Bosnia-Herzegovina; Somaliland; he Unied Kingdom. Holding hese posiions requires commimen, availabiliy and skills, a leas equal wih hose of a Cenral Bank chair. The second argumen is, finally, misleading. Tha is wha appears from J.K. Galbraih s ironic words when he observed In Cenral Banks, as in diplomacy, he look, well-cu convenional clohing and an ease o keep company wih very rich people couns for a lo wih much lower resuls (1977, p. 217). However, i should no be excluded from discussion. The idea is ha here are differences in preferences for he moneary policy goals (Krause and Méndez, 2008). These differences are explained simply because moneary policy decisions will affec, as he case may be, price levels, economic growh, employmen and income wih consequences on individual and collecive well-being (Epsein and Yeldan, 2010). Hence, he aversion o inflaion will change according o individuals or individual generaions (Farvaque and Mihailov, 2009, Scheve, 2004). Thus, considering he independence ha Cenral Banks enjoy, governmens wan someone wih preferences compaible wih heirs o chair Cenral Banks. Commercial banks wish o know he Cenral Bank chair appoined o ensure ha he preferences are no in conflic wih heir own ineress. Indeed, according o he economic heory of bureaucracy, people are driven by self-ineres, and appoining hem as cenral banker does no change his naure. In ohers words, Cenral Banks decisions can be poliically expensive (Toma and Toma, 1986). Hence, he raional choice(s) made by governmens when appoining Cenral Bank chair wih paricular profiles (Chang, 2006). This raionaliy can be exended o commercial banks concerned o preserve 3

5 heir ineress. In he ligh of he effecs of cenral bankers decisions, we assume ha he women s absence from Cenral Bank chairs can be accouned for by heir version o inflaion. In his respec, his paper examines female Cenral Bank chairs degree of conservaism, as defined by Rogoff (1985), wih respec o heir male counerpars. For ha purpose, an innovaive economeric mehodology, which does no require esimaing all he srucural and preference parameers of a moneary model, bu only how men and women s parameers differ, is developed o measure he influence of gender s role in cenral banking. Bu firs, i is worh o noe ha only a minoriy of women have managed o break he glass ceiling o reach he chair of he Cenral Bank. 2. A minoriy of female Cenral Bank chairs A close scruiny of Cenral Banks organizaion chars reveals ha hey are overwhelmingly chaired by men. The oldes among hem, he Swedish Riksbank (founded in 1668), has never been chaired by a female 1. Since he 1950s, only 56 women have chaired a Cenral Bank. In 2015, on a full-ime basis, only 16 women chaired a Cenral Bank ou of a oal of 184 worldwide i.e. less han 10% (8.69%). Among he 34 Organizaion for Economic Cooperaion and Developmen (OECD) counries, i.e. economically influenial, 27 have never appoined female Cenral Bank chairs; he seven excepions are Ausria, Denmark, Finland, former Eas Germany, Israel, Poland and he Unied Saes. By crossing available daa in periodicals and newspapers (Cenral Banking; Financial Times) and several websies (CenralBankNews.info; The Group of Thiry; Council for Pariy Democracy; Worldwide Guide o Women in Leadership; BloombergBusiness; CenralBanking.com; Connecing Women in Sovereign Eniies Globally (WSE)), we have idenified 56 female chairs, since 1949, for 46 Cenral Banks 2. The resuls obained are summarized in able Tha even wih he highes proporion of females members in he MPC (50% since 2003) (see Farvaque e al., 2011) Some websies, such as he Worldwide Guide o Women in Leadership, wrongly include female Cenral Bank chairs: Ms Ellen Johnson Sirleaf (he curren presiden of Liberia); Violea Asfura in Venezuela; Felisa Miceli Jsefina (former Miniser of Economy and Producion of Argenina, ); Ana Dias Lourenco (former Minisry of Planning in Angola). On Guaemala, here is confusion beween Ms. Lilly Zapaa and Mr. Willy Zapaa Waldemar Sagasume (Presiden of he Banco Cenral, ). 4

6 Table 1 Cenral Banks wih female chairs. Cenral Banks Year of Number of Dae of Female Chairs creaion Governors appoinmen Naional Bank of Ukraine Valeriya Gonareva Since 06/2014 Cenral Bank of Cyprus Chrysalla Georghadji Since 04/2014 Maldives Moneary Auhoriy Azeema Adam Since 04/2014 Federal Reserve Bank Jane Yellen Since 02/2014 Cenral Bank of Nigeria Sarah Alade 02/ /2014 Cenral Bank of Madagascar Vonimanira Razafimbelo 10/ /2014 Cenral Bank of Somalia 1960 na Yussur Abrar 09/ /2013 Reserve Bank of Tonga Siosi Cocker Mafi 07/ /2013 Bank of Israel Karni Flug Since 11/2013 Bank of Russia Banco Cenral de El Salvador Cenral Bank of Venezuela Naional Bank of Serbia Elvira Sakh. Nabiullina Since 06/2013 Taiana V Paramonova 10/ /1995 Mara Evelyn A. de Rivera 04/ /2014 Luz Maria S. de Porillo 08/2002 5/2009 Edmee Beancour 04/ /2013 Ruh de Krivoy 04/ /1994 Jorgovanka Tabakovic 08/2012 Kori Udovicki 07/ /2004 Cenral Bank of Seychelles Caroline Abel Since 03/2012 Cenral Bank of Lesoho Reselisisoe A. Malanyane Since 01/2012 Cenral Bank of Samoa Aalina Ainuu Enari Since 08/2011 Naional Bank of Republic of Belarus Nadejda Ermakova 07/ /2014 Tamara D Vinnikova 02/ /1997 Naional Bank of Kyrgyzsan Zina Asankojoieva 06/ /2014 Banco cenral de la Republica de Argenina Jeenbaeva Bakygyl 07/ /2011 Mercedes Marcó del Pon 02/ /2013 Maria Elena Mondragon de Villar 01/ / Banco Cenral de Honduras Sandra Regnia de Midence Gabriela Nunez de Reyes Vicoria Asfura de Diaz /1999 Bank of he Marshall Islands 1982 na Ann Marie Muller Since Souh Africa Reserve Bank Gill Marcus 11/ /2014 Cenral Bank Van Aruba Jeanee R. Semeleer Since 09/2008 Bank of Thailand Tarisa Wananagase 11/ /2010 Banco de Guaemala María A. Del Cid Navas de Bonilla 10/ /2010 Cenral Bank of Kenya Jancina Mwaela (acing) 03/ /2007 Sae Bank of Pakisan Shamshad Akhar 01/ /2009 Cenral Bank of he Bahamas Wendy Craigg Since 06/2005- Banco Cenral del Paraguay Monica L. Perez dos Sanos 05/ /2007 Cenral Bank of Turkmenisan 1991 na Mukhammedova Shakersolan Since 2003 Bank Negara Malaysia Zei Akhar Aziz 05/ Since 04/2011 Bank of Boswana Linah Mohohlo Since 10/1999 Cenral Bank of Barbados Marion Williams 11/ /2009 Cenral Bank of Sao Tome e Principe Maria Do Carmo Trovoada Silveira /2005 Since 03/2011 Moneary Auhoriy of Bermuda Cheryl Ann Liser /2006 Cayman Islands Currency Board 1971 na Cindy Scoland Bush Since 06/ Bank of Guyana Dolly Sursaie Singh 04/ /2014 Denmark Naionalbank Bodil Nyboe Andersen 11/ /2005 Banco Cenral del Ecuador Ana Lucia Armijos 07/ /1996 Naional Bank of Poland Hanna Gronkiewicz-Walz 03/ /2000 Bank of he Lao PDR (Laos) Pany Yahoou Cenral Bank of Finland Sirkka Hämäläinen 4/1992-5/1998 Naional bank of Ausria Maria Schaumayer 05/ /1995 People's Bank of China Chen Muhua 03/ /1988 Banco Cenral de Bolivia Tamara Sánchez Peña 1985/1985 Margaree Wikowski Naional Bank of Eas Germany Margareha Kuckhoff Noe: We coun he number of governors in charge since 1949, excep for Ecuador (since 1986) and Ausria (since 1992). The majoriy of female Cenral Bank chairs were appoined in emerging counries. Among he 46 Cenral Banks lised, only seven (Belarus, Eas Germany, Russia, Serbia, Honduras, 5

7 Kyrgyzsan) have been chaired by women more han once 3 since Similarly, hose wih a high urnover rae have only one nominaion. Thus, he Banco Cenral de Bolivia has only one female (less han 2%) among is 53 successive governors; Banco Cenral de la Republica de Argenina, one ou of 36 (2.7%); Banco Cenral del Ecuador, one ou of 35, (2.8%); Cenral Bank of Finland, one ou of 32 (3%). Many argumens have been pu forward o explain his under-represenaion, including: he conservaive and reacionary ideology of Cenral Banks; he sereoypes in children's educaion relaed o cusoms, religion; he differeniaed educaional guidance received by boys and girls in erms of curriculum preparing hem for he world of work; he absence, in some counries, of gender-posiive policies; he way Cenral Bank chairs are appoined wih a sor of naural bias in favor of men 4. These argumens pose he exisence of circumsances ha are more or less favorable o he promoion of female as Cenral Bank chairs. Wha are hose circumsances? Wha are heir characerisics? To highligh hem, we will refer o he socio-economic and socio-poliical indicaors esablished by he Unied Naions Developmen Program (UNDP) The socio-economic indicaors They concern he Human Developmen Index (HDI) and he Gender Inequaliy Index (GII). The HDI measures he average level of human developmen achieved in a counry in hree basic dimensions: healh and life expecancy, access o educaion and decen sandards of living. Is value is beween 0 and 1. The higher he index, he higher he number of people in he counry enjoying significan well-being. Depending on he HDI value, UNDP ranks 195 counries or erriories in four caegories: very High human developmen (49 counries); High human developmen (52 counries); Medium human developmen (41 counries); Low human developmen (42 counries). Eigh counries are no classified. From he available daa, able 2 shows counries wih female Cenral Bank chairs. By crossing daa from ables 1 and 2, among he 56 female Cenral Bank chairs: Ten are from nine of he 49 counries classified as Very High human developmen; seveneen, from hireen counries of he 52 ranked in High human developmen; nineeen, from foureen counries of he 41 ranked in Medium human developmen; five, from five of he 42 counries classified as low human developmen; wo, from eigh unclassified counries. 3 4 Of he 56 female Cenral Bank chairs, 4 were named wice a differen periods (Ms. María Elena Mondragón Villar in Honduras; Ms Zei Akhar Aziz in Malaysia; Ms Yahoou Pany in Lao PDR; Ms Maria Do Carmo Trovoada in Sao Tome principle). In oal, here are hus 60 nominaions. See inerviews of female Cenral Bank chairs in Cenral Banking Journal, Vol. XI, No. 3, Feb 16,

8 Table 2 Indicaors of Human Developmen Very high human developmen (HDI > 0,80) Counry HDI rank Human Developmen value Gender Inequaliy Index value GII rank Share of seas in parliamen Number of female Cenral Bank chairs Unied Saes Germany Denmark Israel Ausria Finland Cyprus Poland Argenina High human developmen (0,80 > HDI > 0,70) Medium human developmen (0,70 > HDI > 0,55) Low human developmen (HDI < 0,550) Bahamas Belarus Russia Barbados Malaysia Venezuela Seychelles na na Serbia na na Ukraine Thailand China Ecuador Tonga Maldives Turkmenisan Na na Samoa Boswana Paraguay Bolivia El Salvador Souh Africa Guyana Guaemala Kyrgyzsan Honduras Lao PDR Sao Tome & Principe na na Pakisan ,5 1 Kenya Nigeria na na Madagascar na na Lesoho Ohers counries Marshall Islands na na na na or erriories Somalia na na na na Noe: Daa in he ables are hose available o he Human Developmen Repor Office as of 15 November, 2013, UNDP repor. Daa on erriories of Bermuda, Cayman Islands and Van Aruba do no appear in he UNDP ranking. Female Cenral Bank chairs are, hence, mosly o be found in Medium human developmen counries (33.9% of he oal). Then respecively in: High human developmen counries (30.3%); Very high human developmen counries (17.8%); Low human developmen counries (8.9%). This disribuion shows here is no obvious relaionship beween he counries level of human developmen and female Cenral Bank chair appoinmens. 7

9 2.2. Inequaliy gaps beween men and women Those are measured by he Gender Inequaliy Index (GII), which provides a measure of he losses in erms of human developmen caused by inequaliies beween women and men in he hree dimensions of reproducive healh, empowermen and he labor marke. Is value varies from 0 (siuaion in which women have a salary equal o men) o 1 (siuaion in which he saus of women is as bad as possible). By crossing available daa from ables 1 and 2, among he 56 female Cenral Bank chairs: Five (8.9% of he oal) are from 4 counries, ou of he 15 (0 GII < 0.1), where he reamen of women is almos idenical o ha of men; nine (16.0%) from 8 counries, ou of he 35 (0.1 GII < 0.3), where he reamen of women is similar o ha of men; weny-wo (39.3%) from 15, of he 52 counries (0.3 GII < 0.5), where he reamen of women is quie remoe from ha of men; seven (12.5%) from 7 counries, ou of he 85 (0.5 GII), where he reamen of women is very remoe from ha of men. These differen disribuions indicae ha he socio-economic indicaors of female/male inequaliy gaps, as measured by he Gender Inequaliy Index (GII), have more influence han he level of human developmen, measured by he Human Developmen Index (HDI) for he access of women o Cenral Bank chairmanship Socio-poliical characerisics They are observed from hree ypes of indicaors: he imporance of he females poliical represenaion; he poliical sysem; he dominan religion. The imporance of he females poliical represenaion is measured by heir weigh in Parliamen. In he UNDP ranking, only Rwanda has a female majoriy in Parliamen. I has no, however, appoined a female Cenral Bank chair. The analysis of daa from ables 1 and 2 shows where women have in he Parliamen: A) Under 10% of he seas, here are seven female Cenral Bank chairs (12.5% of he oal) in 6 counries ou of a oal of 33; B) Beween 10% and 30% of he seas, here are hiry-four female Cenral Bank chairs (60.7% of oal) in 17 counries ou of 153; C) Beween 30% and 50% of he seas, welve women have succeeded o one anoher a he Cenral Bank chair (17.8%) in 10 counries ou of a oal of 32. This disribuion helps esablish a relaionship beween he relaive imporance of he females poliical represenaion and heir appoinmen as Cenral Bank chairs. 8

10 As for he naure of he organizaion of he poliical sysem, counries can be divided ino four subses of poliical sysems 5 (wesern-syle democracies, former popular democracies, monarchies, ohers): Table 3 The poliical sysems. Wesern-syle democracies Former popular democracies Ohers Consiuional monarchies Argenina; Ausria; Bolivia; Boswana; Cyprus; Ecuador; El Salvador; Finland; Guaemala; Guyana; Honduras; Kenya; Maldives; Marshall Islands; Paraguay; Israel; Madagascar; Nigeria; Pakisan; Sao Tome e Principe; Seychelles; Souh Africa; Unied Saes; Venezuela Belarus; China; Eas Germany; Lao PDR; Kyrgyzsan; Poland; Russia; Serbia; Turkmenisan; Ukraine Somalia Bahamas; Barbados; Bermuda; Cayman Islands; Denmark; Lesoho; Malaysia; Samoa; Thailand; Tonga; Van Aruba 24 counries, weny nine female Cenral Bank chairs 10 counries, fifeen female Cenral Bank chairs 1 counry, one female Cenral Bank chair 11 counries, eleven female Cenral Bank chairs Crossing daa from ables 1 and 3 shows he earlies female Cenral Bank chair appoinmens were made in counries claiming progressive ideology, namely: Eas Germany (1950, 1967); China (1985); Lao PDR (1988); Poland (1992); Russia (1994, 2013); Belarus (1996, 2011); Kyrgyzsan (2010, 2011); Serbia (2003, 2012); Turkmenisan (2003); Ukraine (2014). These counries have quickly promoed he emancipaion of women hrough he implemenaion of gender-friendly policies so as o ensure equaliy beween men and women in all spheres of sociey: poliical, economic, culural, educaional and family life. For example, in China, in 1954, he consiuion sipulaed equaliy beween men and women and, in 1956, a law was passed o proec he righs and ineress of women (Cox-Fill, 1996). The 56 female Cenral Bank chairs are divided as follows: 24 in he Wesern-ype democraic counries; 10 in he former people's democracies; 11 in he consiuional monarchies; one in a counry wih anoher poliical sysem. Thus, even hough he former people's democracies did make he firs appoinmens of female Cenral Bank chairs, heir number is wice as imporan in counries whose organizaion is based on he model of Wesern democracy; hey accoun for 51.7% of appoinmens compared o 26.7% for former people's democracies agains 19.6% for consiuional monarchies. Finally, he weigh of radiion and/or he characerisics of he majoriy religion may more or less influence he role and saus of women in sociey. Indeed, some facors relaed o radiions influence economic choices and he economic developmen 5 Wih he excepion of Denmark, Finland and he Unied Sae, all counries are in democraic ransiion: free elecions are held periodically. Bu in mos, here is no social compromise and no really horough legal sysem. 9

11 process (Alesina and Giuliano, 2013). By and large, hese facors can be approached in reference o he geographical locaion of counries. Table 4 Number of female Cenral Bank chairs disribued by Coninen and Counry. Europe America Africa and Indian Ocean Ausria (1); Belarus (2); Cyprus (1); Denmark (1); Eas Germany (2); Finland (1); Poland (1); Russia (2); Serbia (2); Ukraine (1) Argenina (1); Bahamas (1); Barbados (1); Bermuda (1); Bolivia (1); Cayman Islands (1); Ecuador (1); El Salvador (2); Guaemala (1); Guyana (1); Honduras (4); Marshall Islands (1); Paraguay (1); Samoa (1); Unied Saes (1); Van Aruba (1); Venezuela (2) Boswana (1); Kenya (1); Lesoho (1); Madagascar (1); Nigeria (1); Sao Tome e Principe (1); Seychelles (1); Somalia (1); Souh Africa (1) Asia, Indian-subconinen and he Pacific China (1); Israel (1); Kyrgyzsan (2); Lao PDR (1); Malaysia (1); Maldives (1); Pakisan (1); Thailand; Tonga (1); Turkmenisan (1) 10 counries, 14 female Cenral Bank chairs 16 counries, 22 female Cenral Bank chairs 9 counries, 9 female Cenral Bank chairs 10 counries, 11 female Cenral Bank chairs The 56 female Cenral Bank chairs are disribued in 46 counries belonging, geographically, o he mainland: America, 17 counries, including 11 in Souh America; Europe, 10 counries; Asian, Pacific and he Indian subconinen, 10 counries; Africa and he Indian Ocean, 9 counries. Thus, he American coninen has he highes proporion of female Cenral Bank chairs (39.2% of he oal), followed by: Europe (25%); Asia, he Indian subconinen and he Pacific (19.6%); Africa and he Indian Ocean (16.07%). The weigh of religion impacs he counries' developmen processes (Barro and McCleary, 2003). In paricular, according o he imporance given o he family, he women s saus in he sociey, i can encourage more or less he female professional careers. If we ake ino accoun religion, counries wih female Cenral Bank chairs are disribued as follows: Table 5 Disribuion of counries by majoriy religion (a percenage). Chrisian Jewish Buddhis/Taois Muslim Ohers Argenina Ausria 90 Bahamas 81 Barbados 67 Belarus 55 Bermuda 90 Bolivia 76 Boswana 70 China 87,5 Cayman Islands 81 Cyprus 78 Denmark 77,8 Ecuador 80 El Salvador 83 Finland 75 Eas Germany 85 Guaemala 87 Guyana 57 Honduras 87 Israel 75,4 10

12 Kenya 82,6 Kyrgyzsan 80 Lao PDR 67 Lesoho 90 Madagascar 52 Malaysia 62 Maldives 99 Marshall Islands 97 Nigeria 50,8 Pakisan 98 Paraguay 96 Poland 87,5 Russian Federaion 41 Samoa 98 Sao Tome & Principe 82 Serbia 85 Seychelles 90 Somalia 99,8 Souh Africa 80 Thailand 94 Tonga 83 Turkmenisan 89 Ukraine 72 Unied Saes 70 Van Aruba 81 Venezuela 92 Sources The daa for each counry are obained from websies: Wikipedia; Le Guide du Rouard; Encyclopedia of he Naions. When classifying religions ino hree subses (Chrisian-Jewish, Muslim-ohers Buddhis- Taois), and crossing daa from ables 1 and 5, i appears ha 56 female Cenral Bank chairs are locaed in: 36 predominanly Chrisian (Caholic, Orhodox, Luheran) and Jewish counries; 7 counries where he majoriy is Muslim-ohers; 3 counries where Buddhis-Taois is predominan. Counries wih a majoriy of Chrisian-Jewish herefore accoun for 80.0% of female Cenral Bank chairs agains 14.3% in he counries where Muslim-ohers is predominan, and 5.35% wih a majoriy of Buddhis/Taois. Thus, he observaion of counries wih female Cenral Bank chairs, according o socioeconomic crieria (level of human developmen (HDI), gap of inequaliy of reamen beween men and women (GII)), socio-poliical crieria (naure of he poliical organizaion, weigh of culural radiions, majoriy religion) shows ha: A) Female Cenral Bank chair nominaions seem correlaed o he relaive weigh of radiion, he majoriy religion and he gap in reamen beween men and women; B) There is no obvious relaionship beween he counries level of human developmen and he female Cenral Bank chairs named. As shown by able 1, here is a small minoriy of women in he closed world of Cenral Banks. This siuaion is parly due o he fac ha appoining Cenral Bank chairs is sricly he governmen or parliamen prerogaive. Hence, i would appear imporan o focus on he profile of appoined female Cenral Bank chairs. 11

13 2.4. Female Cenral Bank chairs: academic and career backgrounds To analyze he profile of female Cenral Bank chairs, we colleced biographical daa available from he websies of naional Cenral Banks and supplemened hem wih hose available from oher websies. Daa are summarized in able 6. Cenral Bank chair preferences, on moneary policy goals, are influenced by heir career background. Tha is he case when he is encouraged o express his loyaly o some lobbies, an indusry, a poliical pary or a bureaucracy. In paricular, a pas career as Cenral Bank saff may encourage execuives o idenify heir own ineress wih he Cenral Bank s goals. Indeed, as an insiuion, he Cenral Bank seeks o build up a repuaion and mainain credibiliy. Those repuaion and credibiliy are based on is abiliy o fulfill is mission, mainly summarized by mainaining price sabiliy. Compared o a previous career as a poliician or official (including inernaional civil servans), he Cenral Bank chair is promped o a sronger preference for price sabiliy 6. Those differences sem from moivaions ha are linked wih redisribuion policies, real sae deb reducion, simulus policy. A pas career in he privae secor (in companies, financial and banking secors) will influence he Cenral Bank chair s preferences in he direcion of business saisfacion. All in all, officers wih a background as Cenral Bank saff members, as business execuives (insurance companies, commercial banks, managerial secor) or as public servan have, on average, a sronger preference for price sabiliy han hose wih a background as poliicians or rade-unioniss. A scruiny of female Cenral Bank chairs career background daa reveals ha: 71.4% have held various responsibiliies wihin he Cenral Bank as vice-chair, board member, economis or deparmen manager; 40.8% have had experience in banking and/or finance and insurance; 26.5% have augh as universiy professors or assisans; 51.0% have worked in he public secor; agains 39.6% for he privae secor; 39.6% have been very acive in poliics as members of parliamen, minisers, prime minisers; 33.3% have worked for inernaional insiuions such as IMF, World Bank, UNPD. 6 From he Public Choice perspecive, Cenral Bank chairs preferences have influence on he moneary policy goals. Hence, i is in he service of heir own ineress or special ineres groups. For example, a pas of commercial banker will lead o he capure of he Cenral Bank o he banking secor. 12

14 Cenral Bank Banks/ Insurance Academia Public secor Privae secor Poliics Inernaional funcions PhD of Economics PhD of Business managemen Masers of Economics MBA/finance s/commerce Bachelor Ohers diploma Table 6 Daa on he academic and career backgrounds. Career background Academic background Cenral Banks Governors Ukraine Valeriya Gonareva X X X X Cyprus Chrysalla Georghadji X X Maldives Azeema Adam X X X US Jane Yellen X X X X X Nigeria Sarah Alade X X X X Madagascar Vonimanira Razafimbelo X X Somalia Yussur Abrar X X X Tonga Siosi Cocker Mafi X X X Israel Karni Flug X X X X Russia Elvira Sakhipzadovna Nabiullina X X X X X Taiana V Paramonova X X X X El Salvador Mara Evelyn Arévalo de Rivera X X X X Luz Maria Serpas de Porillo X X X Venezuela Edmee Beancour X X X X X Serbia Ruh de Krivoy X X X X X X X X Jorgovanka Tabakovic X X Kori Udovicki X X X X X Seychelles Caroline Abel X X Lesoho Res elisisoe Adelaide Malanyane X X X Samoa Aalina Ainuu Enari X X Belarus Nadejda Ermakova X X X X Tamara D Vinnikova X Kyrgyzsan Zina Asankojoieva X X X Jeenbaeva Bakygyl X X X X X X Argenina Mercedes Marcó del Pon X X X X X X X Honduras María Elena Mondragón de Villar X X X X Sandra Regnia de Midence Gabriela Nunez de Reyes X X X X X X X X Vicoria Asfura de Diaz X X Marshall Islands Ann Marie Muller Van Aruba Jeanee R. Semeleer X X X Souh Africa Gill Marcus X X X X X X Thailand Tarisa Wananagase X X X Guaemala María A. Del Cid Navas de Bonilla X X X X X X Kenya Jancina Mwaela X X X Pakisan Shamshad Akhar X X X X X Bahamas Wendy Craigg X X X Paraguay Monica Lujan Perez dos Sanos X X X Turkmenisan Mukhammedova, Shakersolan Na na na na na na Malaysia Zei Akhar Aziz X X X Boswana Linah Mohohlo X X X X Barbados Marion Williams X X X Sao Tome Principe Maria Do Carmo Trovoada Silveira X X Bermuda Cheryl Ann Liser X X Cayman Islands Cindy Scoland Bush X X X Guyana Dolly Sursaie Singh Denmark Bodil Nyboe Andersen X X X X X X Ecuador Ana Lucia Armijos X X X X Poland Hanna Gronkiewicz-Walz X X X X X Laos Pany Yahoou X X X Finland Sirkka Hämäläinen X X X X Ausria Maria Schaumayer X X X X X China Chen Muhua X X X Bolivia Tamara Sánchez Peña na Eas Germany Margaree Kuckhoff X X X X Margareha Wikowski X X X X Sources: websies: Bank for Inernaional Selemens; The Group of Thiry; Women in Sovereign Eniies; periodic reviews: World Who s who Marquis Who s who; The Banker; Europa Year Book. 13

15 Thus, a vas majoriy of female Cenral Bank chairs held various posiions in he Cenral Bank before being appoined. Many worked in he public and privae secors and hey had experience in he banking and financial secors. A significan proporion was very acive in poliics. I would hus seem ha he profiles of hese female Cenral Bank chairs seem o il he preferences in favor of price sabiliy. The abiliy o undersand economic and inflaion mechanisms depends on heir academic background. From a Public Choice perspecive, he Cenral Bank chairs are driven by heir own narrow professional perspecives and ambiions. Wih an economis s profile will normally have an advanage enabling him o influence policy decisions in favor of his own preferences or specific arges (Acheson and Chan, 1973). A more business-oriened raining (managers, engineers) will predispose he Cenral Bank chair o greaer aversion o inflaion. Indeed, price sabiliy is a major asse for he economy and business developmen (Mishkin, 2007). The academic background daa for female Cenral Bank chairs reveal (see able 6) ha 40.8% of hem hold a PhD or equivalen (in economics for 31.2% for hem agains 10.4% in business and managemen); 48.98% hold a Maser s degree or equivalen (including 31.25% specialized in economics and 18.7% in finance or business adminisraion or commerce); 6.12% a specialized Bachelor or equivalen in economics, finance, commerce or accouning echniques; 4.08% anoher diploma or cerificae (accouning, engineering). Thus, female Cenral Bank chairs mosly possess echnical skills o influence moneary policy decisions in he direcion of heir preferences. Even if a significan proporion holds a business-relaed diploma, i would be premaure o infer heir preferences beween various poenial moneary policy goals. Overall, he analysis of daa relaed o he female Cenral Bank chairs career background leads us o he conclusion ha, concerning he moneary policy goals, heir preferences mosly go owards price sabiliy. The examinaion of heir academic raining shows ha hey have he abiliy o asser heir preferences. However, only he resuls of he empirical analysis can robusly esablish he exac naure of heir preferences wih respec o price sabiliy, in comparison wih male Cenral Bank chairs. To his aim, he model we are presening hereafer has been designed o esimae he policy parameer of female Cenral Bank chairs compared o heir male counerpars. 14

16 3. Gender and he inflaion/oupu rade-off In he lieraure abou he opimal moneary policy, i is sandard o assume ha moneary auhoriies operae by following a argeing rule, as defined by Svensson (1999). The Cenral Bank chair is supposed o minimize a quadraic loss funcion L, which penalizes deviaions of he objecive variables from heir arge. Le us assume ha hese objecive variables are he oupu gap he inflaion rae x (he deviaion of he acual GDP from is poenial value) and, which he Cenral Bank chair wans as close as possible o zero and he inflaion arge. The quadraic loss funcion is specified as 2 i 2 E x 1 i1 2 2 L x E, 0 1, 0, (1) i 1 i where is an ineremporal discoun facor and represens preferences of he Cenral Bank chair regarding he sabilizaion of he oupu, relaively o he sabilizaion of he inflaion rae around is arge. This quadraic loss funcion is a slighly modified version of he loss funcion frequenly used in he lieraure, wih he presen quadraic inflaion deviaion 2 2 replaced wih he expeced value E. This is because we suppose ha does no depend on he curren moneary policy. Due o delay of he moneary ransmission mechanism, he moneary policy canno influence he conemporaneous value of he inflaion rae, an i operaes only on one-year forward inflaion rae anicipaion. Apar from ha, Eq. (1) is convenionally inerpreed; he higher, he less conservaive he Cenral Bank chair is in he sense of Rogoff (1985). Our aim is o esimae he value of for female Cenral Bank chairs comparaively o heir male counerpars. Are female Cenral Bank chairs more, less or as conservaive as men? The Cenral Bank chair is supposed o conrol he oupu gap 1 x. A firs sigh i may seem oo srong a hypohesis, bu i is in fac a common aci assumpion in moneary models, he policy-maker seing he ineres rae o conrol he oupu gap (see for example Clarida, Galí and Gerler, 1999). Wih his assumpion, we will no need o specify a demand or IS curve. This is clearly an advanage as our resuls are independen of he demand equaion specificaion. The only srucural equaion we need o specify is a New-Keynesian Phillips equaion, which is supposed o be of he form 1 E2 u 1 x, 0, (2) 15

17 where E is he expecaion operaor, condiional on informaion available a dae, and u 1 is a supply shock, evenually auocorrelaed. In his forward-looking formulaion of he Phillips equaion, oupu gap affecs he inflaion rae wih one-year lag, which is consisen wih annual daa according o Svensson (1997). Eq. (2) is grounded on dynamic general equilibrium heory, which keeps he Lucas criic away, and apar from he fac we suppose ha here is a conrol lag of one year, i is of he same form as he Phillips curve considered in Clarida, Galí and Gerler (1999). The policy problem is o choose a ime pah for x which minimizes he loss funcion (1), 2 subjec o he consrain (2). This problem reduces o minimizing x E, he conrollable par of (1), subjec o (2). The firs order condiion of his problem is x 2E 0 2 1, which can be wrien E 1 x. (3) When he expeced inflaion rae is above (below) arge, he Cenral Bank reduces (raises) he oupu gap. The aggressiveness of his response depends on parameers and. The response is more aggressive if he influence of oupu gap on inflaion is high ( is high), and 1 if he conservaism of he Cenral Bank chair is more pronounced ( is high). This las parameer is likely o depend on he personaliy of he policy maker. I may also depend on he economic, poliical conex in which he Cenral Bank chair is doing his or her job. Le denoe he inflaion forecas error. Eq. (3) can be rearranged 1 1 E 1 1 x 1 0, 0 E 1 1. (4) Eq. (4) can be esimaed by OLS. According o condiion 0 E 1, errors of his model canno be serially correlaed. Eq. (4) shows ha for a counry whose moneary policy is consisen wih he model defined by (1) and (2), he regression of 1 on x mus produce a negaive slope and mus be characerized by a serially uncorrelaed error erm. Bu no every counry s moneary policy is consisen wih his model. Counries which are no consisen wih i, paricularly hose which are no following a argeing rule described by he loss funcion (1), should no be included in our empirical analysis. Eq. (4) suggess ha he way o selec hose which are consisen wih he model is o look a he esimaed slope and a he residuals of he regression equaion (4). 2 16

18 In case he Cenral Bank policy is no consan, depending on he curren Cenral Bank chair, we have o make allowance for is variabiliy. In paricular, if female Cenral Bank chairs operae differenly from heir male counerpars, i should be he case ha, and hus, differ depending on he gender of he policymaker. The loss funcion (1) would be differen for men and women, in ha he consan parameer would be replaced wih a variable parameer. In order o generalize he model in his way, le us consider he exogenous gender indicaor d which can eiher be d 0 (he Cenral Bank chair of period is a male) or d 1 (he Cenral Bank chair of period is a female). According o he gender of he Cenral Bank chair, he relaive weigh for sabilizaion of he oupu gap in he loss funcion can be eiher M (for d 0 ) or F (for 1 d ). So he slope of he regression of 1 on x can vary wih he value of he gender indicaor; M F M F i can be eiher for ( d 0 ) or (for d 1). The generalized model of he inflaion/oupu rade-off is hen x d x, 0. (5) F 1 For / x d, E 1 E so M, whereas for 1 d, / x 1 E so. A posiive (negaive) value for indicaes ha he female Cenral Bank chairs are less (more) conservaive han heir male counerpars. More precisely, as F M F M, we ge ha F M M. The raio / measures in percenage how much female Cenral Bank chairs are more cauious abou oupu sabilizaion han men. Esimaing he policy preferences by he way of running he regression equaion (5) is a much more simple mehodology han hose used in he lieraure (see Cecchei and Ehrmann, 2002, Favero and Rovelli, 2003, Ozlale 2003, Caselnuovo and Surico, 2003, Tachibana 2004, Krause and Méndez, 2005, Dennis, 2006). In hese aforemenioned sudies abou he esimaion of Cenral Banks preferences, an idenificaion problem arises, due o he naure of he objecive. In effec, he common objecive o all hese sudies is o ge an esimae of he preference parameers. I requires esimaing all he srucural and preference parameers in order o recover he preference parameers from he reduced form of he model. On he conrary, our mehodology is simpler (and presumable more robus o error specificaion) because we do no ry o esimae he preference coefficiens 17 F and M bu only o esimae

19 how hey differ. To do so, we don need o esimae he srucural parameer ; i is no even necessary o esimae an aggregae demand curve. The basic principle of his mehodology is borrowed from Diouf and Pépin (2010). Esimaion of policy preferences are usually based on esimaion of all policy and srucural parameers. These radiional esimaion procedures require an esimaion of he srucural macroeconomic model and an ineres rae rule, in order o recover all he parameers (Favero and Rovelli, 2003, Ozlale, 2003, Caselnuovo and Surico, 2003, Dennis, 2006), or of he srucural macroeconomic model and he firs-order equaion of he loss funcion s minimizaion (Favero and Rovelli, 2003, Tachibana, 2004), using an approach close o ours. Anoher approach is o esimae he srucural macroeconomic model and o use an esimaion of he variances of inflaion and oupu gap o recover he preference parameers (Cecchei and Ehrmann, 2002, Krause and Méndez, 2005). More simply, we focus on he firs-order condiion, which solves he opimizaion problem faced by he Cenral Bank chair. Our mehodology is also disincive in preferring a forward-looking model o a backward-looking one, and using annual daa (insead of quarerly daa). Esimaing Eq. (4) and esing he significance of ˆ (he OLS esimaor of ) is hen sufficien o find ou if female and male Cenral Bank chairs behave similarly way or no, and in case hey don, o infer abou heir differences. Theoreically, we can run such a regression for every counry o assess he differences beween women and men. For every counry, we can esimae he counry parameers and. The drawback of running individual (counry) regressions is ha he individual esimaors ˆ will be imprecisely esimaed if he number of observaions characerized by d 1 is low. And i is in fac he case for almos every counry ha his number is very low, because of he overrepresenaion of men in he Cenral Banks managemen. If we esimae counry by counry, he variance of ˆ would be high, and he esimaed parameer would no be saisically significan, even if he rue parameer is non zero. Thus we prefer o adop an alernaive esimaion and es mehodology, based on a panel daa analysis. On he one hand, he drawback of such a mehodology is o impose a resricion on counry regressions ha implies he exisence of a common parameer; bu on he oher hand, using he panel sample produces a beer esimae of his common parameer. The panel model can be wrien x d x, 0 (6) j,1 j j j, j j, j, j,1 E j, 1 18

20 where j is inroduced o idenify he counry j (j = 1 N). All he variables,x, d and j,1 j, j, j,1 are counry-specific, jus like he parameers j all counries included in he panel. By generalizing our previous noaions, we can wrie and F So M F j 1 j or in an equivalen manner M j 1 j j. The parameer is common o M M j j j and j F F j j (1 ) j.. The common parameer represens he deviaion of he weighs given o sabilizaion of oupu by women, agains he weighs given by men, in percenage of he weighs given by men. For example, if 0. 10, hen female Cenral Bank chairs are 10% more inclined o sabilize oupu gap han men. The panel model resrics his difference beween female and male Cenral Bank chairs o be he same, regardless of he counry considered. Eq. (6) is a non-linear panel model, which has o be esimaed by numeric mehods (like Gauss-Newon). The sysem of N equaions (6) can be esimaed by non-linear Leas Squares. Bu as he errors j,1 are conemporaneously correlaed beween counries, we implemen a non-linear Seemingly Unrelaed Regression (SUR) mehod o esimae his sysem, exploiing his addiional informaion. j 4. A panel daa analysis The macroeconomic daa used is from he FMI web sie. A he ime we picked up he daa he las updae was 03/2015. We use annual GDP daa (valued a consan prices in naional currency) and annual end of period consumer prices index 7 relaive o hiry-hree counries (Argenina, Ausria, Bahamas, Barbados, Belarus, Bolivia, Boswana, China, Denmark, Ecuaor, Salvador, Finland, Guaemala, Guyana, Honduras, Kenya, Kyrgyzsan, Lao, Lesoho, Malaysia, Marshall Islands, Pakisan, Paraguay, Poland, Russia, Samoa, Sao Tome, Serbia, Seychelles, Souh Africa, Tonga, Turkmenisan and Venezuela). Daa are colleced on he period or on a shorer period for some counries 8. We calculae he inflaion rae for every counry by applying he formula lnp / P 100 1, wih P designaing he price 7 8 For Argenina and China, we use annual average consumer prices index. For Belarus, Russia and Turkmenisan, he GDP dae are available from 1992 and he inflaion daa from 1993; for China, he inflaion daa are available from 1987; for Kyrgyzsan, he GDP and inflaion daa are available from 1993; for Marshall Islands, he GDP daa are available from 1997 and he inflaion dae from 2004; and for Serbia, he GDP and inflaion daa are available from

21 index. The poenial GDP is esimaed by regressing he acual GDP on a quadraic rend, and * he oupu gap is hen defined by x 100 ln(y / Y ), wih Y and acual and poenial GDP. For every counry j and every year we se he dummy variable * Y for respecively he d j o 1 if a female was he Cenral Bank chairman for a leas six monhs. Oherwise he dummy is se o 0. In he iniial daa se, seven counries have a unique d = 1 in 2014 (Cyprus, Israel, Madagascar, Maldives, Nigeria, Ukraine and he US). As inflaion depends on moneary policy wih a delay of one year, and because we have no he 2015 daa ye, hese counries don bring any informaion abou he parameer of ineres in his sudy ( ). This is why hey are no included in he previous lis of hiry-hree counries Individual regressions In order o selec counries o be included in he panel analysis, individual regressions were run. For every counry, Eq. (5) is esimaed by OLS. Some adjusmens have been made for some counries, abou he inercep of he regression and abou he sample period. These adjusmens were necessary because he counries examined in he sudy are no similar, paricularly concerning inflaion raes. Firsly, some counries have experienced disinflaionis processes. So he implici inflaion arge, which equals he long erm inflaion rae if he Cenral Bank his he arge in he long run, canno be supposed consan. So he consan inercep was replaced wih a quadraic rend a 2 b c. This adjusmen concerns Bahamas, Boswana, Denmark, Finland, Guyana, Honduras, Lesoho, Paraguay, Poland, Russia, Sao Tome, Salvador and Serbia. To decide wheher o include he quadraic erm or no, we looked a he serial correlaion of he residual erm, by means of he Breusch-Godfrey Serial Correlaion LM Tes. A significan auocorrelaion of residuals deeced in he model wihou quadraic rend may indicae he exisence of a rend in he inflaion process. So, if we deeced a significan auocorrelaion of residuals, and if he inclusion of quadraic rend makes his auocorrelaion disappear, we include he rend in he regression. I is worh o noe ha he inclusion of he quadraic rend has lile impac on he esimaed parameers ˆ and ˆ, because he oupu gap is a de-rended series by consrucion. So he inclusion of quadraic rends has no consequences on he resuls; i only makes he residuals auocorrelaion disappear. Secondly, adjusmens have also been made on he sample period. The heoreical model described by equaions (1) and 20

22 (2) may be a good descripion of he moneary policy for some counries on a more recen period, bu no since he early 1980s. The concep of a argeing rule is relaively recen, and few Cenral Banks in he world have followed such a rule since he early 1980s. For some counries included in he sample, i is obvious ha he apparen inflaion insabiliy is no compaible wih he inflaion/oupu rade-off. In hese counries, moneary policy was no inended o conrol inflaion, even parially, bu o fund public spending, his policy creaing hyperinflaion. Some of hese counries have more recenly regained conrol of inflaion, and we do ake hese periods of sabiliy ino accoun when running he regressions. Argenina experienced insabiliy of inflaion process up o Before 1992, inflaion measured wih he logarihmic formulae, which underesimae inflaion for high raes relaively o he index price growh rae, was over 60% (on he period ). The inflaion index fell from 1992 and was fairly sable since hen. In Belarus, inflaion has sabilized since Beween 1993 and 1995, inflaion was over 200% before dropping. In Kyrgyzsan inflaion rae was over 200% in 1993 and 100% in 1994, before dropping in In Turkmenisan, inflaion sabilized below 25% from 1998 onwards. The Polish inflaion rae was unsable up o In addiion, in China and in Ecuador inflaion sabilized respecively from 1995 and 2002 onwards, bu as here was no more female Cenral Bank chair from hese years, we didn shoren he sample period. Moreover, here is a lack of reliabiliy wih Paraguay s inflaion daa before The repored inflaion rae was exacly he same ( ) every year from 1980 o 1989; so we excluded his period of he sample. And finally, inflaion was analyzed in Ausria up o 1998, as he counry enered he Euro zone in Table 7 presens he resuls of he individual esimaions of Eq. (5). We es he hypohesis of nulliy of parameers ˆ,â,bˆ, ĉ and ˆ, and we es he hypohesis of negaiviy of ˆ (in ohers words we look if ˆ is significanly posiive). The resuls show ha few counries have adoped a moneary policy compaible wih he model of inflaion/oupu rade-off. In order for he resuls o be consisen wih he model, we have o keep in mind ha ˆ has o be significanly posiive (and he residuals have o be non-serially correlaed). Only eigh counries ou of he hiry-hree considered seem consisen wih he model. Moreover, he Breusch-Godfrey es shows ha he null hypohesis of non-auocorrelaion of residuals of he regressions for hese eigh counries is no rejeced for conservaive significance level (probabiliies are over 1%). The counries concerned are Argenina, Belarus, Guaemala, Guyana, Kyrgyzsan, Paraguay, Salvador and Turkmenisan. According o indicaors of 21