DETERMINANTS OF EDUCATIONAL ATTAINMENT IN EGYPT AND MENA: A MICROECONOMETRIC APPROACH

   DETERMINANTSOF  EDUCATIONALATTAINMENTINEGYPTANDMENA: A MICROECONOMETRIC APPROACH   MENSHAWYGALALMOHAMEDBADR BSc(Hons),MSc    ThesissubmittedtotheUniversityofNottingham forthedegreeofDoctorofPhilosophy  July,2012  Abstract UsingTIMSSdatasetonMENA countries,thisstudyexaminesthedeterminantsof educational outcome and gender inequality of learning in eight selected countries. The complicated structure ofthedatahasbeen considered carefully duringall the stages of the analysis employing plausible values and jackknife standard error technique to accommodate the measurement error of the dependant variable and theclusteringofstudentsinclassesandschools. The education production functions provide broad evidence from mean and quantile analysis of very low returns to schooling; few school variables are significantandnonehaveeffectsacrosscountriesandquantiles.Ingeneral,student characteristicswerefarmoreimportantthanschoolfactorsinexplainingtestscores, buttherewasconsiderablevariabilityacrosscountriesinwhichspecificfactors were significant. Strikingly, computer usage was found to influence students’ performance negatively in six MENA countries. Only Turkey and  Iran had a significantpositiveeffectofcomputerusageonmathsachievements. Genderinequalityofacademicachievementhasbeeninvestigatedthoroughlyusing mean and quantile decomposition analysis. There is mixed picture of gender inequality across the eight countries with three pro‐boys, three pro‐girls and two gender‐neutral. This exercise gives no general pattern of gender inequality across MENA. A detailed analysis of Egyptian students’ achievements explains the differentialgapbetweenschooltypes,notablybeingsingleormixedsexandArabic or language schools.Single‐sex schools perform better than mixed schools especially for girls. The single ‐sex language schools are more effective than the Arabicsinglesexschool.Thisconfirmsthedominanceof thelanguageschoolsand isalsorelatedtothestyleandsocial‐economicstatusofenrolledstudents. TheUniversityofNottingham  ii Acknowledgements ʺAllpraiseisduetoAllah”and“whoeverdonotthankpeopledonotthankAllah” Workingonthisthesishasbeenalearningprocessthathasfarexceededanyofmy expectations.I  wouldliketoacknowledgethe peoplewhohavecontributedinthis regard. FirstandforemostIoffermysincerestgratitudetomysupervisorsOliverMorrissey and Simon Appleton whose knowledge and research experience gave both scope andfocustomyownresearch.Theyputmeontherighttrack,gavemethesupport andthetimetolearnandtobeproductive.Theyopenedtheirdoorstomewithout anylimitations.WhateverIwouldsayIwillneverfulfiltheirrightsonmyself. InmydailyworkIhavebeen  blessedwithafriendly  andcheerful groupof fellow students.Thankstomycolleaguesattheschoolofeconomics;specialthanksgoesto Paul Atherton, Festus Ebo Turkson, Emmanuel Ammisah and Zehang Wang. I wouldlikealso tothanktheUniversityofNottinghamfortheirhospitalityandthe  great facilities they offer to accommodate the different cultures and religions. I wouldliketothankSarahNolan,postgraduatesecretary,forherhelpwhichstarted evenbeforemyarrivaltotheUKandcontinuestillthisday. Iwouldalsoliketothankmyfamilyforthesupport theyprovidedmethroughmy entire life and in particular, I must acknowledge my wife and my son, Mohamed, without their love, encouragement and patience, I would not have finished this thesis. In conclusion, I would like to express my gratitude to my country Eg ypt  and I recognize that this research would not have been possible without the financial supportandScholarshipfundfrommylovelycountryEgypt.  TheUniversityofNottingham  iii Dedication   To my wife, my children Mohamed and Maryam, Also special dedication to my grandma, and my family I also dedicate this thesis to the brave youth of the 25 th of January revolution in Egypt. TheUniversityofNottingham  iv TableofContents Abstract ii Acknowledgements iii Dedication iv Chapter1INTRODUCTIONANDLITERATUREREVIEW 1 1.1Introduction 1 1.2LiteratureReview 3 1.2.1EstimationproblemsofEPFandpossiblesolutions 7 1.2.2Inequalityineducation 8 Chapter2OVERVIEWOFTHEDATA 14 2.1TheTIMSSstudentperformancedata 14 2.2TIMSSsampledesign 15 2.3TIMSSanalysisandcomplexityofthedata 16 2.3.1ComputingSamplingvarianceusingtheJRRtechnique 16 2.3.2PlausibleValues(PVs) 17 2.4MENAcharacteristics 19 2.5ComparativedescriptivestatisticsforMENAcountriesinTIMSS 23 2.5.1InternationalBenchmarks 26 Chapter3EDUCATIONALATTAINMENTDETERMINANTSINMENA 91 3.1Introduction 91 3.2Background 93 3.3LiteratureReview 96 3.4Empiricalmodel 99 3.4.1EducationProductionFunction(EPF) 100 TheUniversityofNottingham  v 3.4.2MetaRegressionAnalysis(MRA) 101 3.4.3Quantileregression 103 3.5Results 104 3.5.1Familybackgroundsandstudentperformance 104 3.5.2Schoolresources,teachercharacteristicsandperformance 110 3.5.2.1Schoolfixedeffects 112 3.5.3Meta‐Analysisresults 112 3.5.3.1Thehomeinfluenceonperformance: 113 3.5.3.2Computerusagereducesperformance 118 3.5.3.3Theschoolinfluenceonperformance 118 3.5.4 Quantile Regressions: Heterogeneity of Covariates Effects by Performance (ability) 119 3.6Conclusion 121 AppendixA‐3:QuantileEstimates 126 Chapter4GENDERDIFFERENTIALSINMATHSTESTSCORESINMENA 132 4.1Introduction 132 4.2GenderInequalityinEducation:ContextandMENA 136 4.2.1TestScorePerformanceinMENACountries 136 4.3Methods 141 4.3.1TheOaxaca‐BlinderDecompositionFramework 142 4.3.2Meandecomposition 144 4.3.3QuantileDecomposition 146 4.3.3.1RecenteredInfluenceFunctionRIF(unconditionalquantiles) 146 4.3.3.2RecenteredInfluenceFunctionRIFandReweighting 148 4.4Empiricalresults 150 TheUniversityofNottingham  vi 4.4.1Decompositionresultsofthemeangendergap 151 4.4.2Decompositionresultsalongtheeducationalachievementdistribution.154 4.4.3 Quantile decomposition results for Saudi Arabia and Iran (without teachers’variables) 161 4.5Conclusion 162 AppendixA‐4:MeanDecompositions 165 AppendixB‐4:QuantileDecompositions 174 AppendixC‐4:QuantileDecompositionDo‐file 193 Chapter5SCHOOLEFFECTSONSTUDENTSTESTSCORESINEGYPT 29 5.1Introduction 29 5.2Egypt’seducationsystem 30 5.3Dataanddescriptivestatistics 32 5.3.1EgyptinTIMSS2007 32 5.3.2Descriptivestatisticsonhomebackgroundandschoolresources 36 5.4TheEmpiricalmodel 42 5.5MainResults 43 5.5.1Studentsbackground 43 5.5.1.1Parentaleducation 43 5.5.1.2Homepossessionsandbooksathome:Socio‐EconomicStatus(SES).46 5.5.1.3Nationalityandhomespokenlanguage 47 5.5.1.4GenderDifferences 48 5.5.1.5TypeofcommunityandPovertyLevels 48 5.5.1.6Computerusageandgameconsoles 48 5.5.2TeachercharacteristicsandSchoolbackground 49 5.6Furtheranalysisusinginteractions 53 TheUniversityofNottingham  vii 5.6.1Genderinteractions 54 5.6.2ParentsʹEducationandhighSES 55 5.6.3ParentsʹeducationeffectandParentalsupport 56 5.6.4Parentaleducationinteractionwithcomputerusage 57 5.7SchoolEffectsandschooltypes 58 5.7.1Schoolfixedeffects 58 5.7.2ArabicandEnglishschools 61 5.7.2.1Splittingsampleusingtestlanguage 62 5.7.2.2Testlanguagedifferenteffectonmathsandscienceachievements 64 5.7.2.3Testlanguageandhomespokenlanguage 65 5.7.3Schoolstypebysexcomposition 66 5.8Extensions 69 5.8.1Testingforaccountabilityandautonomy 69 5.9Conclusions 70 AppendixA‐5:Descriptivestatisticsandfurtherestimations 73 AppendixB‐5:Principalcomponentforhomepossessions 88 Chapter6CONCLUSIONS 197 6.1Introduction 197 6.2Summaryoffindings 198 6.3Futureresearch 201 Bibliography 202   TheUniversityofNottingham  viii ListofFigures Figure 1‐1:LossintheHumanDevelopmentIndexduetoInequalitybyregions 10 Figure 2 ‐1:GrossEnrolmentRatesinMENA(1970‐2003)(%) 22 Figure 2 ‐2:MENAenrolmentratioofprimaryeducation 22 Figure 2 ‐3:PopulationPyramidinMENA,2007 28 Figure 3 ‐1:Distribut ionofstudentachievementsbysubject 33 Figure 3 ‐2:DistributionofstudentMathsachievementbyschoollanguage 34 Figure 3 ‐3:Distribut ionofstudentMathsachievementbygender 34 Figure 3 ‐4:Distribut ionofstudentScienceachievementbyschoollanguage 35 Figure 3 ‐5:Distribut ionofstudentscienceachievementbygender 35 Figure 4 ‐1: Hanushek and Woessmann estimates of the test scores relation to Growth 94 Figure 4 ‐2:MathstestscoresandGDPpercapitaforTIMSSselectedcountries 95 Figure 4 ‐3: Maths test scores and GDP per capita for TIMSS (without high income Araboilcountries) 95 Figure 4 ‐4: Forest plot displaying an inverse‐variance weighted fixed effect meta‐ analysisfortheeffectofeducationdeterminantsonstudentperformance 114 Figure 4 ‐5: Forest plot displaying an inverse‐variance weighted fixed effect meta‐ analysisfortheeffectofeducationdeterminantsonstudentperformance 115 Figure 4 ‐6: Forest plot displaying an inverse‐variance weighted fixed effect meta‐ analysisfortheeffectofeducationdeterminantsonstudentperformance 116 Figure 4 ‐7: Forest plot displaying an inverse‐variance weighted fixed effect meta‐ analysisfortheeffectofeducationdeterminantsonstudentperformance 117 Figure 5 ‐1:GenderInequalityIndex(GII),1995and2008 132 Figure 5 ‐2:TestscoresdistributionbygenderacrossMENAcountries 135 Figure 5 ‐3:TestscoresgapbetweenboysandgirlsinMENAacrossquantiles 139 TheUniversityofNottingham  ix Figure 5 ‐4:Relativedistribution ofmathstestscoresinMENAcountriesbygender (boysasreference) 140  TheUniversityofNottingham  x [...]... dataset,  first  conducted  in 1995  by  the  International  Association  for  the  Evaluation  of Educational Achievement  (IEA),  an  independent  international  cooperative  of national  research  institutions  and government agencies. Members of the IEA are top educational research institutions  from participating countries in Africa, Asia, Australia, Europe, Middle  East, North  Africa,  and ... fields  of study  at  higher  levels  of education  between  boys  and girls.  Streaming  based  on  girls’  advantage  in reading  and literacy  and boys’  perceived  advantage  in maths  can  affect choice and success in subjects and earnings after graduation.   Another reason for skill differences is related to gender combination of teachers and students. Parental and social prejudices about field of study and future occupations ... growth  in addition  to  having  adverse  social  implications  (Alderman  et  al.  1996;  Alderman and King 1998). Allowing for the impact of female education on fertility  and education of the next generation, girls have higher marginal (social) returns to  education  (Klasen  and Lamanna  2009).  Thus,  discrimination  against  female  education is socially costly and may be problem in MENA countries.  ... Testing language   Arabic  Arabic  Arabic  Arabic  Arabic  Farsi  Turkish  Arabic, English   A common factor among MENA countries is the low performance of its students in maths  and science  relative  to  international  peers.  Surprisingly,  MENA’s  lowest  performing  countries  are  among  the  highest  in per  capita  income.  Saudi  Arabia,  Qatar, Oman, Kuwait exhibit poor performance in maths and science. Qatar has the ... values and an imputation variance. The average sampling variance is computed by  estimating  the  sampling  variance  associated  with  each  plausible  value  and averaging them. The imputation variance is determined by estimating the variance  of the five estimates of using the normal method of calculating the variance:     Imputation variance = (1/ 4 ) ∑ (θ 5 PV =1 i −θ ) 2   (2.4)  The  sampling  variance  is  then  simply  the  average ... 1.1 Introduction   This  thesis  investigates  the  determinants of education  achievement  in Middle  East  and North  Africa  countries  with  special  focus  on  Egypt.     The  determinants of education achievement are key factors affecting the quality of education and hence  the human capital capacity in the developing countries. This thesis investigates the  main determinants of education analysing both the ... 2007 round namely; Algeria, Bahrain, Egypt,  Iran, Israel, Jordan,  Kuwait, Lebanon,  The University of Nottingham     23  Chapter 2. Overview of the Data  Morocco,  Oman,  Palestinian  National  Authority,  Qatar,  Saudi  Arabia,  Syria,  Tunisia, Turkey, United Arab Emirates (Dubai), and Yemen.   This study considers the eighth grade students at 8 countries: Algeria, Egypt,  Iran,  Jordan,  Saudi  Arabia,  Syria,  Tunisia,  and Turkey.  The  remaining  countries  are ... countries  of Bahrain,  Kuwait,  Oman,  Qatar,  United  Arab  Emirates,  Saudi  Arabia  and Libya.  Second,  middle  income  countries  are  some  large  oil  exporting  countries  (Algeria,  Iran  and Iraq)  as  well  as  Egypt,   Syria,  Jordan,  Lebanon,  Tunisia,  Morocco,  Palestine  and Turkey.  Third,  the  low  income  countries  include  Djibouti,  Sudan  and Yemen.  The  largest share of MENA’s population falls in the middle income category with more ... trials  to  measure  and quantify  the  effect  of educational attainment and distribution on economic and social outcomes (Barro and Lee 2010) but they mostly  focused on the quantity of education not on quality.   The University of Nottingham     8  Chapter 1. Introduction and Literature Review  Equal educational achievements for men and women have been regarded as one of the  main  drivers  of ... school attainment and other factors.  Johnes  (2006)  argued  that  growth  depends on  initial  income,  the  investment  to  GDP  ratio,  school  enrolment  rates,  schooling  quality,  schooling  distribution,  openness,  growth  amongst  trading  partners,  and a measure  of political  stability.  The  quantity,  quality  and distribution  of educational (inequality and discrimination) attainment have an impact on social outcomes, such  . s beingpublicorprivate,singlesexorcoeducationanddomesticlanguageorforeign languageschoolremainsambiguous.  1.2.1 EstimationproblemsofEPFandpossiblesolutions Estimating education production functions faces a number of practical difficulties: omitted. ii Acknowledgements ʺAllpraiseisduetoAllah”and“whoeverdonotthankpeopledonotthankAllah” Workingonthis thesis hasbeenalearningprocessthathasfarexceededanyofmy expectations.I  wouldliketoacknowledgethe. Mohamed, without their love, encouragement and patience, I would not have finished this thesis.  In conclusion, I would like to express my gratitude to my country Eg ypt 