R.portNo. EC-173 FILE COpy This report may not be published nor may it be quot~d -~s representing the view of the Bank and its affiliated organizations. They do not accept responsibility for its accuracy or completeness. INTERNATIONAL BANK FOR RECONSTRUCTION AND DEVELOPMENT INTERNATIONAL DEVELOPMENT ASSOCIATION COST-BENEFIT ANALYSIS IN EDUCATION A CASE STUDY ON KENYA • November 1969 I' Economics Department Prepared by: Hans Heinrich Thias Martin Carnoy (Consultant) ».::onomics Department Studies on Pro b1ems of Sector and Pro ject Analysis * EC-ll1, July 15, 1963, Herman G. van dar Talc, The ~onomic Conpariso~ Hydroelectric Projects with Alternative Developments of Thermal Elec'line Power EC-128, May 1, 1964, Herman G. van der Tak, The EValuation of Agricultural Projects: A study of Some Economic and Financial Aspects ' ... EC-130, October 26, i964, Lee Charles Nehrt, A Pre-Investment stuqy of the Flat Glass Indust!Z EC-132, January 22, 1965, Herman G. van der Tak and Jochen K. Schmedtje, Economic Aspects of Water Utilization in Irrigation Projects EC-138, October 21, 1965, Jochen K. Schmedtje, On Estimating the Economic Cost of Capital * EC-140a, December 16, 1965, Jan de Weille, Quantification of Road User Savings * EC-141 ,~1;'eptember 26, 1966, Hennan G. van der Tak: and Jan de Weille, An Economic Re7)p raisal of a Road Project: The First Iranian Road Loan of 1959 (IRN-221 EC-151, December 20, 1961, Mark. Blaug, A Cost-Benefit Approach to Educational Planning in Developing Countries * EC-1S8, January 11, 1968" Alan A. Walters,The Economics of Road User Charges EC-160, March 18, 1968, Herman G. van der Tak and Anandarup Ray, The Economic Benefits of Road Transport Projects - * Eqr161 , Mq 2~, 1968, Ayhan Cilingiroglu, Manufacture' Ol: Heavy Electrical, Eg,Uipnent in Leve10ping Countries * EC-162, May 31, 1968, Jack Baranson, Automotive Industries in reveloping Countries EC-164, August 21, 1968, Shl8mo Reutlinger, Techniques for Project • Appraisal under Uncertainty . - EC-169~ May 1969, Helen Hughes, Problems of Food Processing Industries in .Developing Countz1ies i~ Reissued in World. Bank Staff Occasional Papers, distributed by Johns Hopkj,ns Press. PREFACE • This study is the second Economics Department report dealing with problems of the economics of education. It was preceded by a paper prepared for the Bank by Mr. M. Blaug of the University of London Institute ot Education and the London School of Economics, A Cost-Benefit Approach t,o Educational Planning in Developing Countries (EC-157, December 20, 1967). Both stUdies originated from a growing awareness that the methods currently used in planning expenditure for education in developing countries fail to take into account important links between the educational system and 'the economy. On the one hand, current methods take the economic value of more education largely for granted; they do not attempt to J!1easure the ben- efits of the various types of education in monetary terms, thus precluding systematic economic analysis of the benefits as well as the costs of pro- viding addi tiona! education by type and level. On the other hand , little if any attention is being paid to the role of earnings in the demand for and supply of educated people in a country; both demand and supply are affected by earnings and are dependent on each other. Whereas the earlier study provided a theoretical framework for a cost-benefit analysis of educational expenditure, the present report is an attempt to implement those suggestions in a case study on one country: Kenya. Although ~Qst of Mr. Blaug's proposals have been followed up successfully and som~J.J;1)nes of thought have been developed well beyond the original sug- gestions, the authors did not expect the study to yield a once-and-tor-&ll answer to the question of how to approach educational planning in developing countries. They consider the fact that it revealed almost as many new prob- lems as the well-known ones it helped to &nswer to be a true reflection of its experimental nature. The authors wish to express their appreciation tor the great help they received during the various stages of the study. While it would be im- possible to list all those who supported the field work in Kenya, the authors would like to mention Messrs. P. R. C. Williams, Ministry ot Education; A. T. Brough, Ministry ot Economic Planning and Development; D. Higgs, Ministry of Local Government, and K. Wilson, Ministry of Education, without whose con- tributions the study could hardly have been undertaken. Mr. E. Rado, ot the Institute for Development Studies, University College, Nairobi, participated as a. consultant in the early stages ot the • "study. His assistance and wise advice, along with the knowledge and experi- ~nce of Mr. D. C. Rogers, then at the Institute for Development Studies, significantly helped the authors to overcome the initial difficulties and to understand the country's economy and educational system. The hospitality the authors enjoyed trom Mr. D. L. Gordon and his colleagues at the World Bank's Permanent Mission for Eastern Africa is very much appreciated. Me.srs. J. Gitua, O. C. Ntapengerva, C. Okemo, D. Otieno, G. Owino, J. Ouma, E. K. Rotich, and J. B. Rudasingwa were intelligent and hard-working interviewer-s, on vhose judgment the author's could alW8\YS rely. The organi- zational. skills otMiss V. Henderson, the authors' secretary in Nairobi, deserYe the greatest praise. Last but not least t the authors gratefully acknowle.dge the cooper- ation ot the management and employees ot the firms participating in the Kenya Labor Force Survey, on which the success of the study depended more than on anything else. The ensuing analysis in Washington benefi.ted greatly from the thor- ough work ot, Miss N. Abudabbeh, then a, research assistant in. the Economics Department • Mr. H. Schneider did most of the programming work efficiently and with unfailing patience... In the final stages of the study, Mr. R. C. Manning made aignificant contributions to the presentation o~ its theoretical buis; Mi.s S. Snell proved to be a competent and cODsiderat:e::editor. Miss J. Naaaire, Mil" C. Morales, and Mrs. D. Drabwell did & splendid. job ot typing the long and ditf1cul1;. manuscript and many tables. From it .. inception, the authors' work profited trom the guidance ot Mr. H. G. van der'Tak, whose expert advice is reflected in the study's present scope and fora, and whose impartial support was on more than one occaeion a genuine encouragement. While discus.ion. with all the above-mentioned and with colleagues in the Economics and Education Projects Department helped to clarity many difticult points, the responsibility for any remaining shortccmings and er- rors, ot course, remains entirely with the authors. This study ia part ot con'tinuing work in the EconOllica Department OIlproble. . ot sector' and p~pject analysis. Other studies in this field which haTe been given circulation ollttside the Bank are listed on the inside front cover. Andrew M. Kamarck Director Economics Depart.ent TABLE OF CONTENTS Page No. i} \( PREFACE I. INTRODUCTI ON • • • • • • • • • • • • • • • • • • • ...• • 1 1. Purpose of the study • • • • • • • • • • • • • • • • • • 1 • 2. The Cost-Benefi t Method Applied to Education • • • • • • 1 a. The Objections to its Use • • • • • • • • • • • •• 2 b. Re~onse to the Objections • • • • • • • • • • • •• 3 c. Education Process Analysis • • • • • • • • • • • •• 6 3. Outline of the study • • • ..· ..• • • • • • • • • • 7 II. KENYA ~ A BRIEF INtRODUCTION • • • • • • • • • • • • • • • • 8 1. The People and Their Environment • • • • • • • • •• 8 2• The Economy • • • • • • • • • • • • 0 • • • • • •• 9 a. General Remarks • • • • • • • • • • • • • • • • • • 9 b. Agricul ture • • • • • • • • • • • • • • • • • • •• 11 c. Other Economic Activities • • • • • • • • • • • • • 13 III. KFllYA I S EDUCATIONAL SYSTEM ·. ..· ..• • • • • • • • • • 19 1. A Historical Survey of Education in Kenya. ••••••• 19 a. b. From 1900 to 1945 • • • • • • • • ,x; • • • • • From 1945 to Independence • • • • • • • • • • • • • ·.. 19 20 2. The Present Educational System • • • • • • • • • • • • • 22 a. b. Primary Education • • • • • General Secondar,y Education • ·• • • • • • • • • • • • • • • • • • • • • • 22 29 c. Technical Education • • • • • • • • • • • • • • • • 40 d. e. Teacher Training • • • · • • Higher Education • • • • • • • • • • • • • • • • • • • • • • • · • • • • • • 40 42 Fducation Cost and Finance • • • • • • • • • • • • • • • 42 a. The Costs of Primary and Secondary Education • • • • 42 b. The F.i.nancing of Primary and Secondary Education • • 49 IV. THE URBAN LABOR FO ROE • • • • • • • • • • • • • • • • • • • 53 1. Major Socia-Economic Characteristics ·. .. • • • • • 53 Ci II - ii - Page No. a. The Age structure and Fami.ly Status of Urban Employees • • • • • • • • • • • • • • • • '" 53 b. The Ethnic and Tribal Composition of the c. Work Force • • • e • • • • • • • • • • • .. • • The Changing Character of African Emp1oymerr~ • • • • .. 58 d. Parents' Literacy ~~d. Father's Occupation • • • • • 59 e. 61 The Occupational Distribution • • • • • • • • • • • 65 f. The Social lbbili ty of Urban Employees • • • • • • • g. 69 Concluding Remarks • • • • • • • • • • • • • • • • • 75 V. EARNINGS AND SCHOOLING, SOCIO-ECONOMIC AND OCCUPATIONAL CHARACTERI STICS • • • • •• • ••• • • • • • • • • • 77 1. 2. The Approach • • · · · ·· • ···• • ·· ······ ·· • • • • • 77 3. The D9.ta Basi s The M:>del • • • · .. • • · · · ·· • • • • • • • • • • 79 4. • .. · · · .' · • • • • • • The Relationships between Education, Age, and • • • 79 Earnings • • • • • • · · • ··· · · • • • • • • • " • • 81 a. The Case of Urban African Males w.i th 11 or Fewer Years of Schooling • • • • • • • • • • • • 81 b. '!he Case of African Females and Non-Africans: Age-Adjusted Education-Earnings Profiles • • • • • • 89 c. The Case of Persons wi th Higher Secondary Schooling and University Education: Age-Adjusted Education- Ea.rnings Profiles • • • • •• • • • • • • • • • • " 89 5. The Significance of Non-Schooling Variables __ ......... A 1li.gression • • •• . . . . . . . . . . . . . . 93 6. Rural Incomes and Scmoling • • • • • • • 95 VI. THE RATES OF RETURN TO SCHOOLING • • • • • • • • • • • • • • 103 1. 2. The Types of Rates and Adjustments • • • • • The Calculation of the Rates and the Sources · ...• • 103 of Ia.ta • • • • • • • • • • • • • •• • • • • 106 • • • • • • a. Benefits • • • • • • • • • • • • • • • • • • • • • 107 b. Costs •• • • • • • • • • • • • • • • • • • • • • • 107 3. The unadjusted Urban Rates • • • • • • • • • • • • • • • llO 4• Adjusting for Tax and Mcrtall ty • • • • • • • • • • • • 113 5. Adjusting for Socio-Economic Differences • • • • • • • • 114 6. Adjusting for Occupational Variables • • • • • • .. • • • 115 o - iii - PaBa No. a. Union Membership • • • • • • • • • • • • • • • •• 115 b. Public Sector Employment • • • • • • • • • • • •• 116 7. The Question of Ability • • • • • • • • • • • • • •• 117 a. b. Examination Success and Rates of Return " • • • Biases in the Ability Effect Estimate • • • " " · . 117 120 8. Rates of Return to Education in the Rural Sector • • • • • • • • • • • • • " • • '0 • • •• 122 9. ~lral and. Urban unemp1oym~Lt • • • • • • • ~ • " • ". 125 10. Finsl H.'tt.es, .. Adjus~iments Accumulated •• " II • • • •• 133 VII •. ." • II • • • • · " ~ . 139 1. The Quality of School Output •••••••••••• 139 2• Primary Educa ti on • • • • • • •• • • • • • • • • • •• 143 a. ,. b. The Kenya Preliminary Eicam • • • • • The Data and Their Limitations • • • • • · • • • • • • • • • · 143 144 c. d. Regression Analysis Results • • • • • The Rates of Return to Additional Inputs · • • • • • • • • • • 145 148 3. Secondary Education • • • • •• • • • • • • • " • • •• 151 a. b. The CSC and HSC Exams • • •• • • • • • Data and Assumptions • • • • • • • • • • . • • • • • 151 • • • • • 151 c. Regr.ession Analysi s Results • • • • • • • • • • • 154 d. The Rates of Return to Additional Inputs • • • • • 160 4. S1lllDlla.:ry'. • • • • • • .; • • • • • • • • • • • • • • • 164 VIII. THE FUTURE DEMAND FOR SCHOOLING AND THE SUPPLY OF EDUCATED LABOR • • • • • • • • • • • • • • • • • • • • • • 167 1. The Demand for Primary Education • • • • • • • • • •• 168 2. The Demand for Secondary and Post-Secondar,y Educ,ation •••••••••••••••••••••• 17 3 IX. THE FUTURE DEMAND FOR EDUCATED LABOR • • • • • • • • • • • 177 1. Introduction: The Projection Techniques • • • • • •• 177 2. The Demand for Labor Estimated DLrectly from \W1ges and Employment Data • • • • • • • • " • • • • • 0 • •• 180 - iv - Pase No. 3. The.Demand for Labor Derived from Production Function Estima. -OOS •• ..... ... • • " 186 a. Simple Labor Component • • • • • • • • • • • • • 187 b. \\e:ighted Labor Component • • • • 0 • • • • ., • • .', 187 4. The Future Labor Market for African Urban ~..ales with Primary Schooling • • • • • • • • .. 191 a. ~uilibrium Supply of Labor, Rate .of Return to Schooling Equal to 10 Percent • • • • • • • • • •• 192 b. &Dployment and Wage Changes, Labor Supply ,Exogenously Determined • • • • • •• • • • • • • •• 197 5. The Future Labor Market for African Urban Males 'With Secondary Schooling • • • • • ... • • • • 200 a. ~uilibrium SUpply of Labor, Rate of Return b. to Schooling Equal to 10 Percent. • • • • Employment and Wage Changes, Labor SUpply .. . · . 200 Exogenously Determined • • • • • • • • • • '-' .. • • :202 6. The Future, Labor Market for African Ul"ban Males "Wi th Uni ver si ty Training • • • • • • • • • • • • 203 a. Pl:luilibrium Supply of Labor, Rate of Return to Schoo>ling Equal to 10 Percent • • • • • • • • • • • 2103 b. Employment and W~e Changes, Labor Supply EXogenously Detl!erntined • ••• • • • • • • • .f . • • 2()4 7. Results anei Concl'aSions ~ •• • • . . · . 206 4! • • • • • • • ' 1. The ~ecific Results .. . . 21() . . . . . . . . . . . . . . . . 210 t' .,.;' .e .; t ~ • '. ~ 'f··"· c- • a. The Profi tabili ty of Education in Kenya • .' • • • • 210 b. A Detour Into EdUcation Production Functions • • • 215 c. Future Al temat1 ves: The Fmployment Outlook for Educated Manpower in Kenya •• • • • • . .' . . 216 2. The General Lesson • • • • • • • • • • • • • • • • • 217 a. Manpo1lJ8r Requirements Approach: The AJ.temati ve tD Cost-Benefit Analysis • • • • • 217 - v - Page No. b. Conceptual Difficulties with Cost-Benefit Analysi s • • • • • • • • • • • • • • • • • • • • •• 218 c. Practical Difficulties • • • • • • • • • • • • • •• 223 APPENmCES A. Distortions in Age Distribution Due to lll..gital Preference: 1962 Census versus 1968 Labor Force SUrvey B. ,Scope and Organization of the Labor Force Sanple Survey C. Regression Analysis, Income and Educat.i.on of African Males: Schooling as a Variable D. Derivation of Procedures for Estimation of Absolute Earnings from Regression Data E. Correction\\of Rural Landholder Incomes for Acreage and Family Siz~\ Differentials between Age Groups and between Education Levels for a Given Age GroUp F. Solution for Change in Wage Level Two from Discount Formula (9.35) G. Teacher Requirements in Kenya in 1974 TABLES AND FIGURES (See attached list) J:!!!.> Kenya: Provinces and Counties, Tribal Groups o - vi - LIST OF TABLES Page No. Gross D:nnestic Product, 1963-67 10 Table 2.2: Gross Domestic Product, Percentage Rates of Growth, 1963-67 14 Table 2.3: DLstrlbution of Non-African Urban Employees in the Private Sector by Ci. tizenship Status and Education, 1968 (in perQ~.Jlt) 17 Table 3.1: Primar,y School Examination Results, 1926-67 (African pupils only) 21 Table 3.21 Enrollments in Primar.y and Secondar,r Education,. 1959-68 23 Table 3.3: Number of Primary School Teachers by Qualifi cation,. 1960;..68 25 . Table 3.4: Primary Education Ehrollments and Prima;ry Schoo!, Teachers by Qualification in Five Major Towns, 1967 28 Table 3.5: Junior Secondary School Exanrination Results, 19.;26-1955 and 1966-67 (African pupils only) 30 Table j~~~: Cambridgt!, Sq'pool Certificate Examlnation (CSC) Results, 1944-1967 31 Tabla J. 7: Breakdown (in percent) of Higher School Certificate (HSC) ,~~.~~::::::-=:::===---=-~~;nnation Results, 196~-1967, African and Non~African School. Candidates " 33 Table 3.8:' Share ot. .A£rican Pupils in Total Enrollments in Secondary Schools "'n;' Form (Grade) and Sex, 1957-67 (Se1ec$.t.d\ Years) 34 Table 3.91 Distribution of Secondary School Teachers by Tm,$i of School (Aided-Unaided), Qualification, Sex and EbJployment/ Ci tizerl~hip status, 1961 37 Table 3.10: \~ Number \- Secon~ary School Teachers by QualificlJtion, 1960-61 0\ 38 \. If 11 Table 3.11: Range of Annuaf~:lees in Harambee Schools 41 Table 3.121. Students from Kenya at the East African Univer~ty Colleges, 1960/1-1966/7 43 Table 3.13: Estimated Expenditure on Primary Education in Kenya, 1967 45 - vii - . LI ST OF TABLES Table 3.14: Economies of Scale in Primary Education, 1967 47 Table 3.15: Level of Fees in Primary Schools in Kenya, 1966 50 Table 3.16: Fees in Secondary Schools as "of Februazy 15, 1966 51 Table 3.17: Estimated Education Expenditure, from lbmestic Sources, by Main Components, 1966 52 Table 4.1: Employment Ratios, by Ethnic Groups, Sex, and Age Groups, 1968 (in percent) 55 Table 4.2: 1li.stributinn of African Male Urban Employees, by Age and Education, 1968 56 Table 4.3: Mean Number of Live Births to African Women by .Age Group and Education Group (EKcluding Northern Province), 1962 57 Table 4.4: Distribution of African Male Urban Employees, by Tribe and Educational Achievement, 1968 60 Table 4.5: Average Educational Achievement (in Years) of Urban Employees, by Ethnic Group, Sex, and Literacy of Parents, 1968 64 Table 4.6: Average Educational Achievement (in Years) of Urban Employees, by Ethnic Group, Sex, and Occupation of Father, 1968 66 Tablet..7: Average Educational Achievement (in Years) of African Male Urban Employees, by Occupation and Economic Sector, 1968 68 Table 4.8: llistribution of Male African Urban Fmployees by Occupation and Tribe 70 Table 4.9: Average Earr.n.ngs of Urban Employees, All Ethnic Groups , by Age and Occupation (in Ksh per month) 71 Table 4.10: Education and Vertical Mobility: Average Educational Achievement (in Years) of .African Male Urban Employees, by Present and Previous Occupations, 1968 72 - viii - LIST OF TABLES Page No. Table 4.11: DLstribution of African Urban Employees (Males onl-y) by Own Occupation and Father',s Occupation) 7L Table 5.1: Kenya: Unadjusted Age-Earnings Profiles, Afri'Can Males, by Years of Scoooling, 1968 82 Table 5.2: . Kenla: Age-Earnings Profiles of African Males, by Ye~trs of Schooling 1968, Adjusted for Parents I Literacy, Tribe, and Father's Occupation (Ksh/month) 86 Table 5.3: Kenya.: Age-Earnings Profiles, African Male, by Years of Schooling, 1968 Adjusted for all Variables (Ksh/month) 87 Table 5.4: Kenya: Unadjusted Age-Earnings Profiles, AfriO.J ASian and European Females, by Years of Schooling, 1968 (Ksh/month) 90 Kenya: Unadjusted Age-Earnings Profiles for Af~can Males, Alone and as Variable, 11, 13 and l5-l7Years of Schooling, 1968 92 Table 5.6: Kenya: Mean Annual Income:, Farm Acreage, and Age of HOusehold Heads by Education Level of Household Head, 1963-64, Rural Central Province 97 Table 5.7: Kenya: Mean Annual Income of Rural Landholding Heads of Household, by Age and Education, 1963-6L, UnadJusted for Acreage and Family Size Differen.ces between Age Groups anci between ltiucation Levels 100 Table 5.8: K~nya: Total Iricome ,~;t Rural Head of Household, hy Age 1lDd Education, 1963-64, Aqjusted for Acreage and. Fanri.ly Size Differences between !1ge Groups and between ,Education Levels 101 Table 6.1: " Kenya a: Annual Eamings Foregone and Other Pri vgte Costs, by Sex,> Age, and Years of Schooling, African, 1968 (Ksh) 109 Table 6.2: Kenya: Annual Costs Per student Used in Calculating Public R$tes of Return III Table 6.31 Kenya I Average Social and Private Rates of Return to Schooling for Africans by Years of Schooling, Adjusted tor Age Only, tor Age, Taxes and ltlrtality Only,- and. for Age and Socio-Economi.c Variable Only. 112 - ix - LI ST OF TABLES Page No. Table 6.4: Kenya: Average Rates of Return to Different Exam Scores and Schooling Periods, Adjusted for Socio- Econonti.c Variables, 1968 119 Table 6.5:: Kenya: Distribution of African Landomers by Age, "I 1963-64 124 Table 6.6: Kenya: Rates of Return to Schooling, Lsndholding Household Heads in Rural Central Province, 1963-64, Corrected for p'ro babili ty of heing Landholder 126 Table 6.7: Kenya: Rates of Return to Scin?ling, Landholding Household Heads in Rural Central Province, 1966 Costs and 1963-64 Benefits, Corrected for Probability of Being Landholder 127 Table 6.8: Kenya: Estimated Numbers of Afri can Male Primary School Leavers and Estimated Urban Employment Probabilities 1960-1966 130 Table 6.9: Kenya: Urban and Rural, \veighted and Combined Private Rates of Retum to Primary Education, 1960-1966 131 Table 6.10: Kenya: Urban and Rural, Weighted and Combined Social Rates of Return to Primary Education, 1960-1966 132 Table 6.11: Kenya: Combined Urban and Rural Rates of Return to Primary Education, Adjusted for Socio-Economic Back- ground, 1960-1966 134 . Table 6.12: Kenya:: Average Priv.ate Rates of Return to Schooling, All Adjustments by Years of Schooling, 1968 135 Table 6.::'3: Kenya: Average Social Rates of Return to Schooling, All Adjustments, by Years of Schooling, 1968 136 Table 6.14: Social Internal Rates of Return to Schooling for Males in J Eight Developing Countries, Based on Urban Samples: Kenya, Northern Niger.ia, Uganda, India, Mexico, Chile 1 Colombia and Venezuela. 138 ,~- Table 7.1: Average Value (Mean) of Main Variables Used in Exam Performance Regressions, CSC and HSC Cycles, 1966 158 Table 1.2: Kenya: Rates of Return to Additional per Pupil Expenditure 166 - x - LIST OF TABLES Page No. Table 8.1: Four Estimates for Total Primary Education Enrollments, 1969-75 (in 1,000) 169 Table 8.2: Projected Enrollment Figures in Primary Education and Pupil Outflow, 1969-75 (in 1,000) 170 Table 8.3: Kenya: Enrollment s and Outflow of Secondary and Post-Secondar,y Schools by Form, 1969-75 175 Table 9.1: Kenya: Values of Regression .Coefficients in Demand for Labor Equations Based on Employre9nt, wage, and Output, 1957-1966 183 'Tab1e 9.2: Kenya: Gmwth Al terna ti ves, 1966-1974: So cia1 .::Ba~e of Return, Wages and Employment for Two Labor For.c:e Levels 208 Table 9.3: Kenya: Growth Alternatives, 1966-1974: Wages an#d Employment for Afri cans with Primary and Secondaw School and University Training 209 Table 10.1: Kenya: Rates or Return to Investment in Physical ,Capital 213 FIGURES Figure 3.1: Kenya: Distribution of P'ri'llary School Teachers 'by Qualification, 1967 27 Figure 3.2: Kenya: Economies of Scale in Primary Education, ,1967 46 Figure 4.1: Kenya: Dlstribution of Urban Employees by Length .1)£ Service, 1953 and 1968 62 figure 6.1: Variants of Rates of Return to Education Developed. in Kenya Case Study 104 ANNEX TABLES Annex Table 2.1: Kenya: Annual Rainfall and Population Denatty, 1962 Annex Table 2.2: Ratio of Large-Farm to Small-Farm Revenues from Specific Products, 1963-67 - xi - .ANNEX TABLES Annex Table 2.3: Percent Distribution of Non-African Urban Employees by Oi t izenship Status, 1968 Annex Table 2.4: Distribution of Non-African Urban Employees in the Private Sector, by Age and Citizenship Status, 1968 (in percent) Annex Table 2.5: Distribution of Non-African Urban Employees in the Private Sector, by Citizenship Status and Occupation, 1968 (in percent) Armex Table 2.6: Distribution of Non-African Urban Employees in the Private Sector, by Citizenship Status and Size of &nploying Firm, 1968 (in percent) Annex Table 3.1: Average School-Leaving .Age of Urban Employees, by Ethnic Group, Sex, and Educational Achievement, 1968 Annex Table 3.2: llistribution of Primary Schools by Lowest and Highest "Standard Taught, and by Number of Streams, 1967 Annex Table 3.3: Distribution of Primary School Teachers by Qualification am Employment/Ci tizensh1.p status, 1967 Annex Table 3.4: Percent Distribution of Weekly Hours by Subject Group, Primary and Secondary Education, 1968 Annex Table 3.5: Cambridge School Certificat,8 Examinatio.n(CSC) Results, 1944-1967 Annex Table 3.6: Breakdown of Cambridge School Certificate (CSC) ExamLnation Results, African School Candidates only, 1956-1967 (percent) Annex Table 3.7: Performance of Individual Schools in the HSC Examination, 1961-1966 Annex Table 3.8: Students in Teacher Education Colleges, 1960-1967 Annex Table 3.9: mstrlbution of Primary School Teachers over the Salary Scale, 1967 (Kiambu and Nyeri Oounties) Annex Table 4.1: Distribution of Urban Employees bY' Ethnic Group, Sex, and Age, 1968 (in percent) - xii - ANNEX TABLES Annex Table 4.21 Age Composition of Urban Population by Ethnic Group and Sex, 1962 (in percent) Annex Table 4•.3: Shares of Married Persons Among African Urban Employees by Age Group, Sex and Education, 1968, and Among Total African Population by Age Group and Sex, 1962 Annex Table 4.4. Average Number of Children of Urban African »nployees, bY' Sex, Age, and. Educational Attainment Annex Table 4.5: Tribal D1~tribution of African Male Urban Labor Force, 1960 and 1968, (Average Length of stay in Surve.y City (in percent) Annex Table 4.6: DLstribution by Sex and Age Groups of Urban and Total African Population, ~~962 (pe.rcent) Annex Table 4.7: Average Length" of SerVice (in Years) in the 'Employing F.i.rm of Ai'"rican Male Urban Employees, by Age of Interviewee and Siz e of »rrploying Firm, 1968 Annex Table 4.8: Average Length of Service (in Years) in EmploYing Firm ot Urb.nIl Employees, by Ethnic Group, Sex, aJ'ld Age, 1968 Annex Table 4.9: Average Educational Achievement (in Years) of African HUe Urban Employees, by Age and Length of Service in the Employing Firm, 1968 Annex Table 4.l0t Average Educational Attainment (in Years) of Parents of Urban :&uployees with Given Levels of Education, by Ethnic Group and Sex, 1968 Annex Table 4.ll. lli.str1.bution of Urban Employees by Ethnic GrOUp, Sex and Occupation, 1968 Annex Table 4.12r Average Educational Atta.:i.ll.ment (in Years) of African Male Urban Employees, by Age and Occupation, 1968 Annex Table 4.1.3' Average Education of Male African Urban EqlloY'ees, by Occupation and Size of Firm Annex Table 4.14: Average Educational Achievement (in Years) of Non-Af:rican Urban Employees and fhat of Their Fathers, by Ethni.c Qroup and Occupa tion, 19'68 - xiii - ANNEX TABLES Annex Table 4.15: Distribution of Urban Asian and European 1Ynployees by Own Occupation and Father's Occupation Annex Table 4.16: Cumulative Distribution of Urban _loyees by Length of Servi ce, 1953 and 1968 Armex Table 5.1t Key for Variables Used Annex Table 5. 2: Kenya: Regression of Coefficients of Independent ll.nmrtY' Variables; Dependent Variablet Monthly Earning., African Males, 1968 Annex Table 5.3: Ke~a: Regression Coefficients of Independent Dummy Variables, Socio-Economic Variables Held Constant; Dependent Variable: Monthly Earnings, African Males, 1968 Annex Table 5.4: Kenya: Means of Independent Variables, African Males, by Years of Schooling, 1968 Annex Table 5.5: Kany,at Coefficients of Independent Variables, All Variables Held Constant, Dependent Variable Monthly Earnings, African Males, 1968 (Ksh/month) Annex Table 5.6: Kenya:: Unadjusted Age-Earnings Profiles, Asian Males, by Years of SchoOling, 1968 (Ksh/month) Annex Table 5.7: Kenya: Regression Coefficients of Independent Dummy Variables; Dependent Variable: Monthly Earnings, All Sex-ethnic Groups and Means of Variables for African Males with 13 Years of Schooling, 1968 Annex Table 5.8: Kenya: Regression Coefficients of Independent Dummy Variables; Dependent Variable: Monthly Earni:ngs, .All Sex-ethnic Groups with 17 Years of Schooling, 1968, ~ and Means of Variables for African Males Annex Table 5.9: Kenya: Unadjusted Ageo",Earnings Profiles by Years of Schooling, .All Sex-Ethnic Groups, 1968 (Ksh/month) Annex Table 5.10: Kenya: Wage Employment and Average Earnings by Major Sectolli 1967 Annex Table 5.11: Kenya: Mean Annual Gross Rural Family Income by Age and Schooling of Household HE;!a.d, 1963 - :xi v - ANNEX TABLES Annex Table 6.1: Kenya.: Average Delay (in Years) Between Leaving School and starting 'tebrk Qf African Male Urban Employees, by Level of Education and Year of Leaving School (Education in Years) Annex Table 6.21 KeI\V'ar IncoJOO Tax, Surtax and Gra.duated Personal Tax in Rela'tion to Income and Family Size, 1968 (Ksh/month) Annex Table 6.3: Kenya: Average Number of Children for Urban African Males, 1968 Annex Table 6.4: Kenya: Average Taxes for African Males by Age and Yeal's of Schooling, 1968 (Ksh/month) Mex Table 6.5: Kenya: Average Taxes for African Males by . .\ge, 13 and 17 Years of Schooling, 1968 (in Kenyan shillings per month) Annex Table 6.6: Kenya: Average Rates of Return to Higher Secondary and Universi ty Education, African Males, 1968 Annex Table 6.7: Kenya: Model Life Table Functions for the African Population of Kenya Annex Table 6.8: Kenya: Average Differential Private Rates of Return to Schooling, by Years of Schooling, African Males, 1968 Annex Table 6.9: K~: AveraE§e Illfferential Social Rates of Return to Schooling, by Years of Schooling, African Males, 1968 Annex Table 7.1: Kenyaa: Salary Scales for the Teaching Service (IG. per year) .Annex Table 9.1: Kenya: Average Annual Earnings, 1957-1966 Annex Table 9.2t lnployment bT Etlmic Groups and Major Econondc Sectors, 1957-1966 (in 1,000) Annex Table 9.3: Gross Domestic Product at Factor Cost, by Major EconolTIi.c Sectors, 1957-1966 (Ia. Mill.) and Cost of Id. ving Index, Wa:trobi, 1957-66 Annex Table 10.1: Attitudes of East African Pupils towards Family Planning, 1965-66 Annex Table 10.2: Experts' Opinions on Importance of ~eci:t'ied Local Conditions for Sucoess of Development Projects, Disaggregated by Major Regions - xv - ANNEX TABLES Annex Table 10.3: EJeperts I Opinion as to which Specified Groups are Antagonistic to Development Projects, by Major Regions ANNEX FI GURES Annex Figure 5.1: Unadjusted Age-Earnings Profiles, by Years of Schooling, African Male s, 1968 Annex Figure 5.2: Age-Earn:L"lgs Profiles Adjusted for Socio-Econonti.c Vari ables, by Years of Schooling,- African Males, 1968 Annex Figure 5.3: Age-Earnings Profiles, Adjusted for All Variables, by Years of Schooling, African Males, 1968 Annex Figure 5.4: Unadjusted Age-Earnings Profiles, African" Asian, and European Females by Years of Schooling, 1968 Annex F.i.gure 5.5: Unadjusted Age-Earnings Profiles, Asian Males, by Years of Schooling, 1968 Annex Figure 5.6: Unadjusted Age-Earnings Profiles, African Males by Years of Schooling, 1968 I. INTRODUCTION 1. Purpose of the Study; 1. This study describes the evaluation of the educational system of Kenya with the use of. a cost benefit approach to education investment pro- posed by Mr. M. Blaug in a 1967 Bank paper. 1/ It is based on data collected in 4,742 interviews with employed Kenyans 9 conducted in January and February ot 1968 wder our supervision, and other data collected from KeDy'a. We sought to derive from these data several kinds of information which are valu- able in assessing an educational system and planning for it, and which so far have not been combined in cost-beneti t studies ot education. The most impor- tant are: (i ) private and social rates of return to investment gj in education adjusted for differences in socio- economic background and other factors; (ii) rates of return to increasing different kinds of per pupil expenditures , with benetits related to improvement in exam performance; and (iii) wage/employment/gross domestic product alternatives tor Kenya in 1974. 2. While the data from· the study apply to the Kenyan case only, the conceptual framework and analytical methods employed in this exercise have wide applicability; they are under examination rather than Kenya. 2. The Cost-Benerit Method Applied to Education 3. The study tries to meet the usual objections raised to applying the cost-benefit approach to expenditure on a social product like education. Conceptually, the effort to satisty these objections is successfUl though data shortcomings make it impossible in some cases to test the empirical relevance of the tramework developed. Nevertheless, information and con- clusions derived :from this particular case study m.ay seem rather obvious to those acquainted with Kenya. It could be argued theJt a simplified analysis would probably yield the same general conclusion: in terms ot maximizing short-run and possibly long-run growth ot GNP, some parts of the educational system have been given too high a priority in the use of both private and public fUnds. Because some of the problems ot Kenya' a educational. ay8tem ~e untypically easy to identi ry, this case provides a lesa powertul dem-· onstration ot the approach than would be possible in a country where demand tor and supply ot educated manpower were more balanced than 1n Kenya. JJ EC-157, A Cost-Benefit A roach to Educational Plannin in Develo in Countries, December 19 7. Y Throughout the study, investment is understood to mean expenditure, whether by individuals (school fees, income foregone) or by the govern- ment, and is not restricted to capital or development expendi ture • - 2 - 4. Even sOs a crude analysis of the problem would not produce figures which estimate th~ future consequences of present education policy in Kenya, nor would it be able to build the empirical foundations for alternative pol- icies, as -this study does. Therefore, it should be made clear from the out- set that any shortcomings of the present study are not a reflection of the usefulness ot the cost-benefit approach, but are rather the result of dat a insufficiencies and of the fact that the case chosen did not tully test the power ot the approach. In less clear cases, the method can provide infor- mation not available from unsophisticated analyses, and it can be of great value in the kind of educational investment decisions made by the Bank. Even in a situation similar to that of present-day Kenya, the resul.ts of the stuqy demonstrate the fundamentals for much more detailed educational planning than has been previously possible. a. The Objections to its Use 5. Major objections to cost-benefit analysis o~ investment in edu- cation fall into five classes. First, this kind of analysis usually defines the difference between the earnings of less educated and more educated per- sons as the benefits to the additional tormal schooling taken. However, these incremental. earnings are due not only to additional schooling, but also to differential education outside the school resulting trom differences in socio-economic background, and f'rom other "ability" differences not re- lated to schooling that exist between more and less educated persons. There- tore, using additional earnings as a measure of the benef!ts to investment in education exaggerates these benefits. 6. Second, the use of earnings or wage differentials .as the beneti ts to additional schooling overlooks the impact of unemplqy.ment in the labor market. Significant unemployment m8\Y make wages or salaries invalid as the sole measure of b~nefi ts, both from the indi·vidual 's point ot view -- the probability ot unemployment enters into his calculations -- and from the society's point ot view -- a certain percentage of graduates at each level do not earn measured wages. 7. Third, eost-benefit analysis assumes that measured wages equal productivity. Even if there i13 no unemployment, distortions i:n the labor market Dl8\V' create gaps between wages and producti vi ty • In Kenya and many other developing countries, public sector wages are higher than private sector wages, and union pressure and national manpower policies set wage and hiring guidelines that may have nothing to do with productivity. To correct for these distortions, we should estimate "shadow wages" which would prevail in a purely competitive, distortion-free labor market. 8. Fourth" the rates ot return to investment in schooling derived trom cost-beneti t analysis are based on the bene~:l.ts and costs at the time the data were collected. The rates may measure the return to investments already made, but they m~ well not remain valid for further investment under- , taken now or in the future, since both wage differentials and costs are likely "to change over time. The;retore, the method should ofter some theoretically sound w~ ot calculating future rates ot return which accounts for changes in the benetits and costs of investment in various levels of schooling • - 3 - 9. Fifth, cost-benefit analysis can perhaps measure the direct eco- nomic return to education investment. But education is Justifiable for many other reasons: for its value as a consumer good, enjoyable for both parents and children and worthwhile for its own sake; for its value as a poli·tical good, and as a means of changing attitudes towards, for example, family planning or national identity, or other socio-political iSBues. Therefore, the analysis should take these potential effects ot education into account. b. Response to the Objections 10. In response to the first objection, we have successfully adjusted the income streams of urban Africans for socio-econanic background, ability t and occupational variables which may or may not be directly related to in- creased education. Chapter V is devoted to the calculation and adjustment of age-income profiles for urban Africans (see Tables 5.1-5.5), from which income streams are derived in Chapter VI. Rates of return calculated from the adjusted streams are therefore automatically corrected for these factors. 11. The procedure used in Chapter V to arrive at the adjusted age- income profiles for urban Africans is as follows. Broadly speaking, regres- sion analysis is used to find the differences between the average incomes of wage-earners due to educational. variables, socio-et .""'''''~'1''''''.'' ".~ -P- - .. - ..... .~ ! SQur..cel I Tabl~ 3~~4J . ~.- - ' .. I .. i . . - -; ..' i . - -.::: :- !-: 1 ·1 I~ J I ,-1-... -j !-; --..J , .. ~ ~!:t"~ ... !" • ~~F~~~ ..... 'I "."1, OBLO 9t' H::>l\i( 3h.l.. O.L 01 X 01 • Table .3i..14 1 Economies of Scale in Primary Education, 1967 County I (Meru) _ County II ~uranla) Both Counties Percent Average Tea.cher Perq~nt Averageeacher Percent Average Teacher ; ot Salaries of Salaries of Salal~ies Size of '~chOols per SchOols --per SChools per School (pupils) PUPij {XL tdY 1KEY <70 16.9 7.48 4.9 7.60 11.7 7.51 70 -104 15.2 1.01 4.9 6.36 10.8 6.88 105-139 8.1 7.34 3.4 8.04 6.1 7.51 140-174 12.6 6.60 11.2 1.57 12.0 7.00 175-209 13.2 6.03 9.4 1.28 11.6 6.47 210-244 7.9 5.67 11.6 6.67 9.5 6.20 245-279 11.5 5.19 10.9 6.20 11.2 5.61 280-314 5.3 4.62 10.5 5.69 7.5 5.25 315-349 5.6 4.14 7.1 5.96 6.3 4.67 350-384 2.0 4.88 9.0 6.08 5.0 5.81 385-414 0.8 3.86 4.5 5.99 2.4 5.57 410-454 0.6 4.80 4.1 6.08 2.1 5.91 455-489 0.3 4.35 2.2 6.16 1.1 5.91 490-525 1.9 6.35 0.8 6.3.5 525-559 1.9 5.62 0.8 5.62 ;,;:;60-594 1. 5 5.16 0.6 5.16 ~95-629 0.7 4.96 0.3 4.96 630 0.4 5.17 0.2 5.17 'rotal Number of Schools 356 267 623 Source: Ministry of Local Government. - 48 - " 39. The capital costs of primary schools can be supposed. to be in the neighborhood ot KZl2~20 ~er pupil place for permanent structures and KE9-l5 for semi-permanent strucftures a 1/ For the tradi tiona! "mud-and-wattle n struc- tures prevailing in most of rural. Kenya, a separate estimate. is made, arri T- ing at ~l per pupil. place. This refers to a standard three-room building housing 100-120 pupils. 2/ It the duration ot the various structures is assumed to be 40, 30, aDd 10 years, respectively, the annual undiscounted amortizatioD charges come to Ksh 6-10, Ksh 6-10 and Kah 2 respectively, . i.e., the three types ot buildings show lesser cost differences than might have appeared at first. sight. Secon~ Education 40. The cost cOJ('Ponents of secOll\dary schools are analyzed in detail in Chapter VII, Section. 2 and 3 (for a SUlllll&ry of the main ~ipres, see Table 7.1). (! The DJeaa costs are about KZ65 for dq schools aa41m95 for boarding ~;~=hools, but in the case ot the high-cost schools a figure ot KE200 or more ma¥ apply. 3/ Schools wi th Form V and Form VI classes are much more expensive than lover secondar,y schools, not only because of the larger share of boarding pupils, particularly outside, the Nairobi area, but also because of a better av~rage qualification of teachers (the upper classes are taught by graduates only) anei a lover pupil/teapher ratio (the legal maximum. is 30, as compared to 40 in Forms I - IV). !I 41. Initial cQP.atruction costs tor lower secondary schools range trom Kr120 to ~OO per pupil/place, depending on whether the school haa boarding facilities or not and on its location (Iairobi/up-countr,y). These figures refer to schools vi th three streams and thus reflect certain econcaies of- scale. The construction of senior secondar,y schools is likely to be more expensive, particularly' if' a science stream is to be established. 5/ 1/ Adapted from: JaB. Knight, The Costing and Financing of Educational DeTelop!ent in Tanzania, UNESCO: International Institute ot EdUcational Planning, Paris 1966. The upper figures aasUI'1e 30 pupils: per ciassroom, the lover 50. gj ASSuming expenses of KZ20 for wattle poles, ~50 for corregated iron sheets, and an imputed. vage of KE50 for 1,000 hours of labor. The annual maintenance costs of these constructioDS mq be in the neighborhood of Kah 50. Thele figures emerged from discu~sions with Mr. D. S. Higgs, Senior Local Government Officer, and other officials from'the Ministries of Loc&1. Govermaent and Cooperatives and Social Services. 3/ Our figures yield a mean of 087 (of which Kr.67 were spent on teactiers' salariea) tor schools with Forms I - IV only (reg8.l dless, of whether they i he boarding tae!Ii ties or not), and of Km.75 for senior secondary schools (0144 on teacheJ!" salaries). I,' !!I See the Education (Education Standards) Regulations 1968" para 3 (April 4, 1968) • Due to CODsiderable dropout in Form V, the actual. average size ot Fol'll VI cluses is on17 about 25 pupils a· iI Source: Ministry ot Education. - 49 - b. The Financing of Primary and SecondarY Education 42. The responsibility tor the management and financing of primary schools is delegated to the local authorities (county councils and munici- palities ). Y They meet recurring expenditure out of their regular revenue, school fees contributing a sizeable share (35 percent" Dot counting fee re- missions which on average amount to 5 percent of the totai fee revenue). The central government lend$ som:.~· support in the form of grants channelled r through the Ministry of Local Government. The level of tees varies consid- erably among counties and wi thin counties between Standards (see Table 3.15). The national averages for fees in Standards I - VII range from KSh 54 - Ksh 70. 43. Fees in the secondary school system differ widely according to the status of the individual school (maintained, assisted, or unaided) and to whether they refer to day or boarding pupils. Senior secondary c;lasses in most schools were exempted from fees in 1966 (see Table 3.16). 2/ In Harambee, schools the fee level is considerably higher than in the majority of aided schools, fees being the major source of income for the school committee. An estimate of fee ranges and the extent of regional variations is given in Table 3.11. 44. A global estimate of &11 educational expenditures financed from do- m~stic sources concludes this chapter (Table 3.17). The figures referring to the private sector contribution contain a rather large element of uncer- tainty. While the estimates may not be completely free of double-counting, they are still probably on the low side. It is natural to ask whether this enormous investment in human capital fOlmstion will pay oft. We turn our attention to this question in Parts Two 'i}l.D.d Three of the study. ------------ ---~:',';;.-~, 11 Only three aided primary schools are exempt from this rule. 21 In th~ so-called high cost schools, fees remained subatantial. In 1966 fee exemption applied to 1,168 out of 2,079 students in Forms V and VI; 505 pup~.ls paid KElO/l4, 72 paid ~33/l0, and 334 pupils up to 1(~140. It can bJ assumed, however, that for the great majority of African pupils fee exemPtion or coverage of fees by scholarships was the rule. The figures cited in this footnote were computed by Mr. P. R. c. Williams, Ministry of Education, and communicated by Drfl Daniel C. Rogers. - 50 - Table 3.151 Level of Fees in Primarl Schools in Kenla,z 1966 (in Ksh. per year) Standard 1/ Local Authoritl I II III IV V VI VII VIII Nairobi City 54 54 54 54. 54 54 54 Kiambu County 60 60 70 10 70 10 10 Muranga County 65 65 65 65 65 65 65 Nyeri County 60 60 60 60 65 65 6.5 Thika Municipality 50 50 64 64 64 64 64 Kirinyaga County 60 60 60 60 60 60 60 Nyandarua County 63 63 63' 63 63 63 63 Kakamega County 40 40 50 50 80 80 80 Bungoma County 40 40 50 50 80 80 80 Busia County 40 40 50 50 80 80 80 Taita-Taveta County 50 50 50 70 70 10 10 Ki1iti County 50 50 50 10 70 70 10 'rana River C.ounty 45 45 45 10 10 10 10 Lamu County 45 45 45 10 70 10 10 Kwale County 55 55 55 60 60 60 60 Mombasa Municipality 60 60 60 60 60 60 60 Meru County 50 50 50. 60 60 60 60 Masaku' County 60 60 60 60 60 60 60 Kitui County $0 50 50 50 60 60 60 Fmbu County 50 50 50 50 60 60 60 Isiolo/Marsab1t Counties 50 50 50 50 60 60 60 Laikipia County 60 60 60 60 60 60 60 Narok County 50 50 50 50 60 60 60 60 Samburu County 50 50 50 50 50 50 50 01kejuado County 40 40 40 40 50 50 50 50 Turkana County 20 20 20 1002/ 100Y lOO!/ 100Y Kipsigis County 60' 60 15 15- . 85 85 85 Nakuru Municipality \60 60 60 60 60 60 60 Central Rift County 69 69 69 69 69 69 69 , Eldoret' Municipality 60 60 60 60 60 60 60 Kltale Municipality 45 4, 60 60 60 60 60 5irikwa County 55 55 55 72 72 72 72 \.,,- Ki.sii Oounty 50 50 50 50 80 80 80 Kisump County 50 50 50 50 80 80 80 Kis~~Municipality n.a. n.a. D.a. n.a. n.a. n.a. n.a. South Nyanza County 50 50 50 50 80 80 80 Garissa County )0 30 30 65 65 65 65 Manders. County 30 30 30 65 65 65' 65 Wajir Oounty 30 30 30. 65 65 65 65 1/ Only Harck and OO-kejQ.dQ Co1lllt1es had Standard· VIII classes in 1966. gl Inclu.d1ng boardilll tees. SOUl'oce: Ministry t»t Bdllcation. - 51 - Table 3.16: Fees in Secondary Schools as of February 15, 1966 Category Annual Fees (in Ksh) of School Tuition and Books ~arding A. Maintained Schools General Secondary Schools and Four-Year Courses in Technical Schools Forms I - IV Boys 200 250 Girls 200 100 Forms V - VI Free Free Technical Schools (two-year trade courses) First year 150 Second year 180 High-Cost Maintained Schools Duke of Glouces ter School 414 1,620 Duchess of Gloucester School 414 Highway Secondary School 414 Ngara Secondary School 414 Eastleigh Secondary School 414 Technical High School, Nairobi 414 Allidina Visram High School 414 Coast Girls' High School 414 M.P. Shah Central High School 414 Uasfn Gishu Secol1dary School 41h Menengai High School 414 Kisumu Girls' Hieh School 414 Kisumu Boys' High School 414 Technical High School, Mombasa 414 B. Assisted Schools Nairobi School 1,170 2,010 Duke of York School 1,170 2,010 Kenya High School 1,170 2,010 Highlands School 1,170 2,070 Delamere Girls' High School 870 Delamere Boys' High School 870 Limuru Girls' School 1,050 3,450 strathmore College 840 2,100 All other assisted schools 414 SOUI'ce: r~nistry of Education, Gazette Notice No. 527, February 15, 1966. - 52 - Table 3.17: Sources, Central GovernIll4!nt: recurrent expenditure 7.92 Municipal Counc:L1s: recurrent and capital expenditure 1.46 Count,- Councils ,: recurrent expenditure 5.70 East African COl~ni ty: General Fund Services 0.26 General Government: capital expenditure 0.91 Private Sector Contribution!! 2.23 Misce11.neou~ 0.14. Total 18.62 Share in 1966 C~p at market prices: --..------------------------------------------------------------ !I Secondar,-education fees and all expenditure on unaided secbndal'1 schools. 2/ Expansion, :replacellent, and maintenance of temporary prima·r.7. school 1 - constructions. Source: Republic of Kenya, Economic Survey, 1968, and own estimates. ... 53 - IV • THE URBAN LABOR FORCE 1. The Labor Force Survey was conducted in January/February- 1968. While we were primarily concerned with the relationships between age, edu- cation, and earnings, our survey &1.80 provided a we&l.th of information on the nature of urban eJlployment in Kenya. Many of' theBe obserYatiou .uQ' apply to the wbllll labor forc~ of other African countries &8 vells 2. To sUIIDIlarize briefly the more important observations, our data indicate that in Kenya, the average effective work life of Africans is shorter than tor the Europeans and Asians, that employees below the age of 30 are better educated than their elders, and that wanen with more education tend to have fewer children. While members of a tribe in general tend to find work in the towns neighboring their trib&1. areas, persons with more education b~come .ore mobile. Census and seniority data suggest that the tempor&ry' "target workers" who made up much ot the Kenyan urban labor force before independence, have largely been replaced by permanent urban wage earners. Parents' literacy is positively correlated with the child's level of education.. Occupational data show that in general the 'lather'. social level hu less effect on the child's educational opportunities than is the ease in most Eur 0pean countries. Level ot education seems not to vary in 1 any systematic w~ between difterent sectors and firm sizes, and larger firms do not seem to hire proportionately more highlY qualified employees than smaller firms do. Most ot the employee~ in the survey had changed occupation categories during their careers; and those who moved upward had on the whole received more education than those who stqed in the SUle category or moved downward. 1. Major Socio-Economic Characteristics a. The Age Structure and Faaily Status ot Urban EDpl9Yeea 3. Our crude estimates of employment profiles show distinct differ- ences between Kenya's three ethnic groups. On the whole , it appears that the average length of the work life tor African males and females is rela- tively shorter, whereas the age-employment profiles tor the ethnic minor- i ties, particularly for the Europeans, resemble those in aemi-indu.trialized end industrialized countries. Annex Table 4.1 gi vee the age breakdown by ethnic group and sex ot the private industry andL cOJlllllerce segaent ot the .. survey. The figures are compared with the parallel cross-classification ot the 1962 Census results in Annex Table 4.2. 1/ On the assUllPtion that the ,age composition ot the urban population haa not undergone major changes since !/ Irregulari ties in the age distribution caused by indi vidu&1.e' preterences for numbers ending in certain digits when g1 ving their age ( "digital preference") were found to be much less important in the Labor Survey ,fiata than in the 1962 Census data; see Appendix A. - 54 - 1962 t ve obtain a veipted 88e distribution ot urban employee.. By multiply- ing each/ot the six. sex/ethnic segments by the overall eaployment coeffi- cient, 1. ve arrive at estimates ot the age-apecitic eaploylle~t rates for these six sepents ot the total population, presented in Tabl~ 4.1. 4. The ace-educati_ coaposi tion (Table 4.2) clearly reflects the country' 8 recent educational expan8ion. The aTerase educational attainment ot Atricaaa in the older age-groups appears to be high in the light of avail- able enrol.lllent recorda. This suggests an occasional over.tatelent of edu- cational. achlne_nt bY' the inteninees:, in spite ot scme siJlple checking procedure. which. vere uaed whenever major incODsistencies were 8uspected during an intemew. 5. The shares o'f: married and UDDarried (single, widowea::., d1 'YOrced, separated) persons by qe and educational achievement is givea:.in Annex Table 4.3 U1d compared with. the corresponding figures of the 1962 Census. The larger shares ot UDJD&rried females in the Survey reflect the greater likelihood ot this group to find an urban job. 6. Almex Table 4.4 points to another character.1atic ditterence be- tveen urbaa employee. 8Dd the rural maJority ot the African population, naaely' the tendency ot people to have less children 88 their ,dueatioD in- creues. S1lI11ar obsen:utions in the 1962 Census are compiled in Table 4.3. In this latter CU., hovenr t the reductiOll in tertili ty i8 not as larse .. in the surYey population, which suggests that both education aDd urbanization contribute to the change'. It is interesting to note that thi,> reduction in the nuaber' ot birtha hu cCllle about spontueousl.y', i .• e. in the' abaence ot an otticial tadly pl&DDing policy. This s11fJgesta that such a progrUl. might be quite succes.tul, given the large number~ ot young people who h&~ recentlY received :fUll pri...r;y education ad are thus potentially read7 to, abandon traditional concepts ot desirable tamilJ size. 7• The obsened negative &asociation ot a persQIl' a education and his (or her) nuaber ot children and the advant&8es resultiD@ from'it U\Y be in- terpreted &8 indirect benefits ot education vhich would have to be taken into account when &8aessing the total benetits accruing to society as. a re- sult of invest_nt in education. Ho attempt haa been made in:: this study to quantity" theae benet! ts • One obvious tora ot savings, n_e17 in medical expeD8es related. to childbirth, is estimated by the KeDy'aD authorities to be of the order of ISh ~O in 1963. 2/ Other avoided costs vould be in the 1/ This h .. been calcu.lated in the tollowing wlQ": the lfairobi, Molibua and Nakuru. em.plo7llent figures tor the six groups as publisheeL in. the Stat- istical Abstract 1967 have been multiplied by the 1966/6r growth rate tor nOll-asricultural. -.ployment.. The populations of the three cities have been e.tUiated. tor 1967 on the buia ot the. obsened. inter-cen.us growth (1948-62), ...u.1ng constant ah are s. ot males and temales and constant absolute nUlibera ot! non-Atrl.cUls. Y Source: M1nist17 of BconClDi.c Plannil18' and Development t FDiU Planning in lCe.a, 1967 t ;,p. 6. Table u.l: Emp1o~ent Ratios 2 bl Ethnic GrouE2 Sex 2 and A~e GrouEsz 1968 (in percent) European European ~ African Males African Females Asian Males Asian Females Males Females 15 - 19 16.7 18.1 25.0 14.7 22.2 35.7 20 - 29 79.,0 20.5 79.6 27.3 36.7 70.0 30 - 39 98.5 21.0 94.4 12.0 96.2 41.7 40 - 49 78.4 21.6 70.9 5.8 77.3 36.8 1/ 50 - 59 65.2 5.6 83.3 (100)11 (63.6)- Averages (15 - 59): 79.0 19.9 72.8 16.0 79.0 50.0 ~ 1f Small absolute numbers of observations. Sources: Tables 3.l and 3.2jassumptions as explained in the text. Table u.2: Distribution of African Male Urban Employees J by Age and Education J 1968 Years ot Education Age (Years) - 0-2 ~ 6-1 8-9 10-11 12-13 1h and more Totals - 14 and under 0 0 1 0 0 0 0 1 15-16 0 0 1 0 0 0 0 1 17-19 4 6 32 12 j 0 0 57 20-24 11 69 280 67 105 18 3 553 25-29 17 173 315 41 19 6 14 645 30-34 30 231 228 16 9 2 13 529 I V1 Q'\ j5.. j9 32 160 123 15 10 3 11 359 40-4h 29 87 48 8 3 1 2 118 45-49 19 48 19 4 0 1 0 91 50 and over 30 10 2 1 -11 - 2 1 2L Totals 153 8~ 1051 165 211 32 44 2466 Source: Labor Force Sample Survey, january/February 1968. Table 4.3: Mean Number of Live Births to African Women b,___AKe (}~ol.lP and FJducation Groue (Excluding Northern Province Number of Years of Schoolini Less than 1 Year ~ or not stated 1 - 4 5 - 8 9 - 12 13 and more 15 - 19 0.4 0.3 0.2 0.1 0.1 20 - 24 1.7 1.7 1.3 0.5 2.5 25 - 29 3.0 3.2 2.9 2.4 2.6 )0 - 34 4.1 4.8 4.5 3.8 2.0 35- 39 5.0 5.9 5.4 4.6 3.0 40 - 44 5.5 6.8 6.0 6.6 3.0 \n. -..J 45 - 49 5.8 7.4 6.6 3.9 4.7 50 and over 5.2 6.9 6.1 3.1 0.6 Source: Kenya Population Census, 1962. Vol. III: African Population, Nairobi 1966. - 58 - field of education. AssUldng Wli veraal primary education a generation from nov and disregarding the probable changes in both infant mortality and per pupil coats of primary education (which may or m~ not offset each other), the undiscounted savings on future primary schooling alone would already compensate tor a good deal. ot the costs of schooling incurred in the present generation. 1/ other benefits, such as the improved health of mothers and children as a result of fewer and more widely spaced pregnancies, are also obvious but less easy to &Ssess. These considerations reinforce rather than weaken the argument that primary education, while not being judged very protitable from the income creating point of view, produces important in- direct benetits which must not be left out of accolmt. b. The Ethnic and Tribal Composition of' the Work Force 8. The share of the three main ethnic groups in the Labor Force Survey population ditters slightly from the 1966 total urban employment figures. There are various possible reasons for these discrepancies: first, o~ fig- ures refer to the private sector only, the ethnic composition of which is clearly difterent trODl that of the government sector. Y Secondly, the Arab population, Which Is an important component ot the work force of Mombasa, has been lumped together with the Asian population in our Survey, whereas they ~ have been included in the African population. in the 1966 employment figures. 3/ 9. The tribal composition of the African labor force is given in Annex Table 4.5. A comparison wi t~ that of 1960 reveals a number of shifts in the shares of the main tribes. ~ ~~e last line of the table shows the !I The Labor Force Survey data actu&l.ly do Dot permit such a statement to be made vithout qualification. One would have to analy'ze differences between urban and rural. population with the same educational achievement in order to isolate pure urbanization effects, and also, which parent IS education contributes most to the decision to keep the family ~ize small. However, referring to orders ot magnitude the above statement is tenable. y The 1966 shares of the three main ethnic groups (Africans, Asians, and Europeans) was 81.8, 13. 7, and It.. 5 percent , respectively, tor employment in the public service, and 93.7, 4.0, and 2.3, respectivelYt for employ- ment in private industry aDd cODlllerce (Source: Statistical Abstract, 1967) • 'JJ A later checking of individual questionnaires suggested that the Arab employees in our lample had more &tfinity to the African than the Asian population in the narrower sense (Indians, Pakistanis, Goans ) with regard /) to the main social. characteristics; this mlQ' have biased - albeit margj.n- ally - our findings tor the Asian minority. '!!I The large share ot "Other East Atricans ff , particularly in the Mombasa sample, is partly explained by an incorrect coding of minor coastal tribes; However, the Mambasa employment does include, in comparison to 1960, an increased share of other East AfricaDs (i.e. Tanzanians), as was dilc10sed by a checking of the questionnaires. - 59 - average length of stay in the survey cities by members of the main tribal groups. As was to be expected, the nearby tribes have a clear time-lead over the more distant ones. The results can be. summed up in the trj.vlal. statement that the neighboring tribes are more than proportionately repre- sented in the respect! ve urban lTork forces, whereas the more remote tribes contribute less than proportionately. 10. This simple observation does not apply when educational achievement is taken into consideration. Table 4.4 shows that it is more relevant for the less educated and vice versa. This finding touches on an effect of edu- cation not mentioned so far, namely its role as a catalyst of migratory movements. From an individual's. point of view, this effect is no doubt a. beneficial one; whether this holds for society as a whole is doubt~ul: since migratory flows are usually from areas of low economic potential to the more favored ones, a substantial out-migra·tion of educated persons means tha't the poorer areas lose much of what may be their only asset, and tha.t in the long run the regional. differences in wealth will widen rather than dis appeal" • 1/ A policy of equal educational opportunities which has as its ultimate goal a more equitable distribution of incomes m~ thus well be self-defeBting if not accompanied by appropriate economic and fiscal measures. c. The Changing Chare.cter of African Employment 11. Until recently, the great majority of Africans in wage employment could not be considered as a permanent work force comparable "&0 that of in- dustrialized countries. They had come to town for a limited span of time only and hence in most cases lef"li their families behind in the villages to work their small farms. In most cases, these workers st~ed only long enough to make savings sufficient to meet certain extraordinary expenses such as taxes, Bchool fees, the bride price, the costs of additional livestock or of land improvement, etc. This system, which has been widespread not only in Keny~ but in other African countries as well, has been analyzed so frequently that we content ourselves with thls short description. 12. The result of this temporary migration to the towns is shQlm in Annex Table 4.6. At the time of the 1962 Census, only 38 percent of 'the urban African population were females and 34 percent children, as compared to 51 and 52 percent , respectively, for the l~al areas. In the main regions of out-migration, a substantial percentage of households was hea.ded by women (18 percent in the case of Embu dis'trict). However, when cODlparing the 1962 with the 1948 Census figures, it can be seen that the employment pattern described above was already changing at the eve of independence. 13. This impression is confirmed by survey data on the seniority of African erA7.)loyeea. The median number ot years African male employees had 1/ However, if one adopts the traditional. opt imal i ty viewpoint, then mi- gration of labor in response to wage differentials, even on a large scale, would not be objectionable. This point was made by Dr. De.niel C. Rogers, Teachers College, Columbia University, New York. TableU.4: Distribution of African Male Urban &n~1ol!es.! b~ Tribe and Educational Achievement, 19 B Years of Education Never went Tribe 0-2 3-5 6-7 8-9 10-11 12..13 14 and more to School Totals Kiku)'U 119 240 391 64 120 15 14 172 1135 ;Wip- 107 206 176 11 28 4 2 188 728 Hera 58 186 190 41 32 3 3 90 603 Hijikenda 10 45 67 7 6 5 0 99 239 Luo 26 127 159 23 27 6 11 86 465 I ~~enjin- '" 0 ~~aking Tribes 3 8 21 1 .5 0 2 8 48 Other Kenya Tribes 1 4- 6 1 3 0 0 2 17 Other-East African Tribes 25 60 71 15 12 2 1 12 258 Not applicable, answ&? --2 ~3 28 ~o ~T9ta18 358 889 1109 1 ---2 2 -0 _2 --1! 170 242 37 33 126 3564 Source: Labor Force Sample Survey, January/February 1968. - 61 - been with their firms in 1968 was 4.4 in Nairobi and 3.7 in Mombesa, as com- pared to 1.1 and 1.6 in 1953. This means that the turnover 't'ry on the whole very limited. The breakdown by size of firm suggests the turnover to be smallest in firms with 15 - 49 employees, whereas the 50 - 99 employees group appear to have the greatest turnover, the large firms with 100 and more em- ployees being in a middle position (Annex Table 4.7). A breakdown of the African employees by age and seniority gives results which implied a somewhat greater propens i ty of more educ ated persons (e. g ., esc certi fi cate holders) to change employers than among less educated or illiterate employees (Annex Table 4.9). The main explanation for this difference ~ be the ls,rge pool of persons with little or DO schooling competing tor urban jobs as opposed to the ,hortage of more qualified labor ,,,hich only now is beginning to sub- side. 1 14. Figure 4.1 summarizes our arguments by showing the seniority pro- tiles of' African employees in private employment in Nairobi and Mombasa in 1953 and 1968, respectively, and that for non-Africans in the -three survey cities in 1968. There appear to remain only marginal seniority differences between African and non-African employees (obviously the proportion of employ- ees with 10 or more years of service is still an exception), which suggests that the transition process from the "target worker" to the permanent urban wage earner has be~n largely completed (see also Annex Table 408). Another facet of this process should show in future censuses through an increased share of females and children in the urban population as the now settled employees bring their families to the cities. "d. P~nts' Literacy and Father's Occupation 15. It is often assumed in enrollment projections and wastage estimates that parental education is positively related to both the demand for education and school performance. While this assumption is examined in Chapter VII, and the impact of parental education on children's earnings assessed in Chapter V, no use has been made of parental education as predictor of the demand for education, largely because ot unsatisfactory test results. 2/ !I An observation which points in the srune direction was recorded in Business Safari Digest (Vol. IX, No.1, January/February 1968, p. 9). A leading banker deplored that If • • • young ICenyans entering t,he banking profession are abandoning their posts rather than ~ontinuing to study past the early stages. During the past three years 1678 Kenyans were employed locally. Of these, 1027 left their employment It .. This 61 per- cent turnover meant a heavy loss to the employing b·anks, since n ••• over the same period, 1300 employees wel'"e trained ,locally at a coat of E126,650 in tive local bank training schools ~nd at on-the-job training., More than 50 were sent overseas at a cost ot Z«18, 250 • " gj Reliable time-series exist only for cert1ticatesissued, not for years of education (primary and lower secondary level, see Tables 3.1 and 3.6). Furthermore, fluctuations in the annual numbers are so strong that even the introduction of gliding averages does not yield results which would lead to a satisfactory fit in a regression of past on prescn"& education. I~{~~E ;0, ~}O,:~~/HE INCH ~~~ ,?780 l\t:t'FFF.~ ... rIO C'SSt..~ CO~ --~ r :, -~.~~--~-l- ;--=--~ -'- ., .. .~--".-- " -t~~:~~~~f:' - --+- - + - - - - •. -. "r-- .---'--' I ~ I\) ... ,. __ .- f !- -----1-----1-----; --I - l' -,- .' 1 _--, ---~ .,- ! .. i ,r.: .- .._. ···L~~··- ; . ~.~-~ t I _ _ _ _ _ _ _ _d _ _ _ _ __ F ~-----" -- j -.'- - --!j - , - ~ . .. .. . ~ -...... . . I ~, ~ ~ I i. I 1 2' 3 I 4 5' 6 . 7-- 8 9 ~: I I ~.- ; , r .. ---1 I i I --r-- - 1 i r-- I ! -, I i ~~th of service. .(Year, Sourc~: Annex Table 4.16. _I -! ,----,-.J . • . I - 63 - This does not imply, however, that parental education and children's edu- cation are unrelated. On the contrary, even a very crude distinction (be- tween literate I illiterate fathers and mothers) reveals the advan·tages or disadvantages children of these two groups are likely to have - not only &8 regards the likelihood to be sent to school at all but also to complete it successfully (Table 4.5). 16. Annex Table 4.10 pursues this argument further ; it ShOlTS for the six ethnic/sex groups the average parental education of interviewees with a g1 ven level of education. A number of features are common for the six groups: (i) both father's ana, mother t s education increase with the interviewee's education, in general. from the lowest through the highest level; (i1) an important exception is the level of paren'tal education of African males, which drops for university graduates below that for persons with senior secondary schooling (HSC holders); 1/ this indicates that exceptional talent is either well identified and taken care of in Kenya or by itself strong enough to overcome environmental ob- stacles such as low parental. education; 27 (iii) the females in the sample had - regardless of the ethnic group they belonged to - consistently better educated parents than the maJ.es; :furthermore, the educational achievements of their fathers and mothers were more similar than in the case of the male interviewees. The obvious conclusion is that the socio-economic conditions in Kenya are, or have been, 'Working against female edu- cation. Only the better educated parents agreed on the de8irabili ty of their daughters going to school; appar- ently this decision was strongly influenced by the mothers. 3/ Y The high values for the parental. education ot African males, Asian fe- males, and European fem$les with incomplete university education are not considered here because ot the negligible number of observations (one, one, and two cases, respectively). 2/ The same conclusion can be derived from the data on the parental edu- catic:a of African males with 7 years of education broken down by the riesult obtained in the Kenya Primary Exam: those who passed had better educated parents than those who failed, but the parents of those who r.!~~itied for secondary education (a high pue) had received only about 'tile same amount of education as the parente of the .failers. 3/ \rheae conclusions probably have validity for Dlore developed~ountries and .,' higher levels of education, respectively • Table 4.2_: Average Educ~tional Achievem~tlt (in Years) of Urban !nployees, biEthnic Group, Sex, and Li tersey dor Parents,· 968 (in brackets numbers of ob~ervat~ons) Father Illiterate Father Literate Fg-ther 1111 terata Father Literate Hother~lliterate Mother 1111 terata Moth~r Literate Mother Literate AfriQ@ ~es 5.1 (1725) 6.7 (624) 6.5 (24) 8.1 (372) ;miean F~JDales 4.6 (l07) 6.4 (63) 8.0 (2) 8.0 (101) ~ian Hales 6.4 (72) 8.6 (132) 10.8 (5) 10.9 (176) ABianFemales 11.5 (h.) 10.7 (18) 11.0 (1) 11.4 (82) I E1l1'Ppt!an Male~ 6.3 (4) 12.0 (2) 12.9 (84) ~ +=- E\WP~apFeJnale~ 11.8 (87) SOl.ll'qe: Labor Force Sample Survey, January/February 1968. - 65 - 17. We turn now to the relationship between the education or inter- viewees and the occupation of their fathers (see Table 4.6). It is fre- quently maintained that ther~ are certain social groups -- uainly defined along occupational lines -- who show a greater propensity to have their children educated~ The data in Table 4.6 suggest that for male employees at least, fathers in vhi te-collar occupa'tions show auch an attitude, but in general, that there is less association between tat);ler' s occupation and edu- cation level than in European countries, for example. 18. The expected. association between 1'ather's occupation and educa- tional level is most clearly seen in the case of European males; the only majol- deviation in this group is the low level ot education in the farmers group, though Highlands farmers are supposed to be both keell on educating their children and wealthy enough to aftord it. For temales, inter-occupa-' tional differences; t:"tre far less p:ronounced, as we expe·cted. But some re- lationships indicated by Table 4.6 are rather unexpected. Africans whose f'athers tell in the 11 administrati ve, executive, and manageria.l workers II category achieved a comparatively lower, and tho8e with fathers who vere skilled workers, a higher educational level than a priori may have been expected. Since the former category comprises owners of businesses, it ~ well be that the majority of those observations referred to petty traders and similar occupations. In the case of Asian males, the SODS of skilled workers have a surprie ingl.y low average education level as compared to the sons of semi-Bkilled and unskilled workers. It is pos8ible that many of the former were village artisans whereas '~he latter were 'preponderantly urban induatriaJ. workers. This points to the limited usefUlness of standard occu- pational definitions f'or developing countriea and to the fact that urbani- zation and parental education have a stronger impact on parents' attitude towards education than occupational affiliation. 19. F'or all six groups, 'the children of professional workers in scien- tific and technical fields seem to be ~t a disadvantage compared to the children of other profeasoL.onal workers. This sallewhat surprising result ~ in part be explained by the preponderance in the latter group of teachers (whose children are aJ.m.ost by definition "education-prone") and partly by an incorrect recording of' occupations in the :former (the most typical error being to call anybody in a technical occupation an "engineer"). 20. On the whole, the figures in Table 4.6 show that the effect of. social stratification on educa'cional opportunities is not nearly as strong in Kenya as it is in BlOst European countries, and is to a large extent super- seded by other environmental. influences. e. The Occupational Distribution 21_ The disaggregation o~ the survey population by major occupational categories l / reveals characteristic differences among the six ethnic/sex 1/ The occupations of interrlewees were first coded on a three-digit basia, tollowing mostly the I.S.C.O. (International Standard Classification ot Occupations) • For the analysis of the Survey results, these occupations are aggreg~;~ed into the 10 groups found in the tables. Table be" : Average Educational Achievement (in Years ~ of Urban F4E1olees, bZ Ethnic Gro~~ S8X Z and OccuEation of Father, 1 68 - OccuE!tion of Father VIII Ethnic Group/Sex -I II III - IV V -VI VII - II - X Urican "Male~ 6.0 8.0 8.5 7.0 6.9 7.4 1.2 6.9 6.5 "'.3 African F..uea 6.1 7.0 7.7 3.0 8.0 7.3 7.1 5.1 6.5 Asian Kales 9.1 9.6 12.8 9.0 10.6 10.6 10.0 7.6 9.1 8.8 Asian ~p..males 11.0 11.0 14.6 11.5 11.8 10.9 11.1 10.8 10.8 EurQP8a1l Hales 13.4 14.0 14.6 11.0 15.7 13.3 12.8 11.2 9.5 12.1 I Eur,opean 'Females 11.4 11.9 1l.9 11.0 11.0 11.6 12.4 11.7 11.7 11.5 '" '" I - Farmers VI = Administrative, Executive and Managerial Workers. II= Professional Scientific and Technical Workers VII = Clerical and Sales Workers. III- Other Professional Workers VIII = Skilled Workers IV - Technicians IX c Semi-8killed and Unskilled Workers V • Foremen and Supervisors x = No answer. . Source: Labor Force Sample Survey, January/February 1968. - 67 - groups as shown in Annex Table 4.11. Annex Table 4.12 gives the average education for an occupation/age cross-classification of the African male employees. Y It is interesting to observe the gradual. upgrading of occu- pations in terms of "acceptable" educational qualifications as the educa- tional system provides increasing numbers of primary and secondary school Ie avers • This leads to the question of the equi vfilence, in qual! tati ve tenns, of a given amount of education over time. While the present figures do not permit any conclusion, the authors during their field work wer.e fre- quently offered the opinion that over the past years the ability and knowl- edge of recruitments 1lith a given nominal. qualification (eq, the KPE) has sharply decli,ned. Although this would hardly come as a aurp'rise, given the unprecedented. expansion at all levels of the educational system and the in- evi table temporary deterioration in the qual! ty of teachers that "VTent with it, the authors hesitate to regard this argument as ~ established fact in the absence of any supporting sts'tistical evidence. 2 22. In these tables, the rather low average level of education of pro- fessional workers in scientific and technical occupations is probably due to the mistaken class.ification mentioned in the previous section. The low fig- ure for the two persons classified as "other professional yorkers" is purely accidental, the intel~ewees being a clergyman and instructor, respectively; this again puts the blame on the rigidi'ty of standard occupational classi- fications. 23. Table 4.7 and Annex Table 4.13 give the average education of urban employees, ~ross-classified by occupation and economic sector on the one hand and by occupation and size of employing firm on the other. Whereas the shares of the various occupational categories in total. employment differ considerably from one sector to another, the average education varies to a lesser extent and not in any systematic w~ (with the possible exception of manufacturing and commerce which have some.hat better qualified employees in most, occupational ~ategories). 'll}}e public sector observations which have been included in Table 4.7 stem from a non-random sample and hence are not strictly comparable with the other figures. 24. Annex '.rlable 4.13 shows the average education of African male em- ployees by occupation and size of the employing firm. The results fail to produce evidence for the otten-heard argument that large firms frequently hire "overqualified" personnel for certain jobs and p~ them higher salaries. This kind of behavior, which invalidates tIle basic assuaption of the rate-of- return's approach to education, namely that a person's earnings reflect his or her' productivity, does not seem to occur in Kenya. 1/ For leek of apace and in view of the much smaJ.ler nuabers of obserTationa, the results for the, other five' sex/ethnic groups are not given. Y The 'expansion in enrollments does not by necessity imply that the addi- tional. pupils are less able (even if the selection system is strictly _rit-qrient~d). However, it is JIlOre than likely that. a co.bination ot &bi11 ty and enTiroDJlleDtal factors w'orks to the disadvantage ot the new- COIlers. More school iXlputs would be neces8ary to OTerC(8! the handicaps this group is facing (see also Chapter VII). \1 I.- Table U.1: Averae. Educational Achievement (in Years) of African Male Ul'ban by Occupation and Fconomic Sector, 19,h Sector I II III IV V VI VII VIII IX X Totals Manufacturing 10.1 1.3 1.0 9.0 1.8 ,.9 ,.0 ,.4 1180 Construction 10.0 4.0 7.2 5.3 4.2 3.6 194 Ca..erce and Banking 7.2 5.0 11.0 7.5 4.5 4.6 5.5 488 Tl'ansportandCoJIIIIUnications 11.0 4.0 5.6 8.0 10.3 8.6 6., 4., 6.4 220 Other Non-Government Services 7.0 7.7 6.7 7.3 5.2 4.0 363 I Q\ GoYernment Sector 17.0 16.2 1.8 11.0 13.3 10.3 1.3 6.1 7.5 397 CD 2842 I -Farmers VI • Administrative, Executive, and Managerial ~rkers II • Professional Scientific and Technical Workers VII • Clerical and Sales Workers III - Other Professional ~rkers VIII - Skilled Workers IV • Technicians IX • Semi-Skilled and Unskilled Workers v • Foremen and Supervisors x • No answer Source: Labor Force Sample Survey, Januar,y!February 1968. - 69 - 25. An examination of the relationship between occupation, age, and earnings confi::t"JIlS that sa.laries rise with age 'up to age 50 (see crable 4.9). The relative salary levels, as will be discussed in Chapter V, correspond roughly to the average education of the interviewees in the respective occu- pational groups. Table 4.8 shows other expected features: the Kikuyu, Wes'li- ern Bantu (Luhya-Kisii-Kuria), and Luo tribal. groups provide the lion IS share of professional workers (categories II aDd III); relatively more ot these three groups hold jobs in the two high categories -- 1.5 to 2 percent i" versus less than one percent for the other tribes; in any case, more than half of every tribal group works as semi ....skilled and unskilled workers. 26. Annex Table 4.14 comparea the average education of Asian and Ii European employees grouped by occupation with those of their fathers. It is \ rather surprising to see that in quite a tew cases the fathers had a better education than. the younger generation in the same occupation group; this is in contradiction to the generally accepted view that the long-term upward shift in general educational attainment will be reflected in an upgrading of occupations. Part of this paradoxical result ~ be accounted for by the large share of non-responses (about 35 percent). It could also be that many of the fathers immigrated after having received their education in countries with 8. higher gener-.l education leyel, and that the labor market in Kenya did not require the same amount of educa.tion for their sons and deughters, particularly in view ot the considerable private costs of education. f. The Social. Mobility of Urban Employees 27. Through the Survey data, two kinds of movements indicating social change can be traced: an individual. 's movement from job to job, and his choice ot occupation compared to his father's. Table 4.10 gives the present and previous occupations of African employees ~d the average education of per80ns in each occupation/occupe.tion "box". 1/ The amount of inter-occu- pational. movements appears to be considerable; only a minority within each category has changed jobs on~y within that category (top left to bottom right diagonal.) or not a,t all (bc\~tom l~ne). Furthermore, those who moved upward had on the whole recei veq, more edt.tcation than those who stqed in the 8ame group or moved downward. Y This suggests that their Buperior formal. 'rtion of tathers who were farmers, and upward: biues to the other groups, since mQst of these answera were given by older employee. whose fathers could' hardly have been in urban wage employment. o Table b.UI es n - Father's Occupation: I III - IV - V VI - VII - !!!! - IX X Totals I II 24 12 15 41 8 235 4 309 987 1 17 , 1648 :0;1 2 1 1 1 1 l& 8 12 44 IV 1 2 3 S 11 V 2 1 3 8 11 21 1 47 VI 1 1 3 1 9 10 13 38 VII 4 1 1 2 1 1-4 17 24 4 69 -..J l:-' VIlI 2 4 4 1 15 ?9 49 ? 106 IX 6 2 3 5 1 60 70 189 5 341 X Totals .2 47 11 28 .:2 16 2- 24 127 -- 230 811 38 1258 36 73 492 681 2112 61 3566 11 I = Farmers. II • Professional Scientific and Technical Workers. III = Other Professional Workers. IV = Technicians. V = Foremen and Supervisors.. VI = Administrative, Executiw, and Managerial Workers. VII z Clerical and Sales Workers. VIII= Skilled Workerse IX = Semi-Skilled and Unskilled Workers. 1:. =No answer. Source: Labor Force Sample Survey, January/February, 1968. - 75 - groups (white-collar and blue-collar occupations) and jra'f;hers' occupations in three (farmers in addition to the other two), the iuter-generation occu- pational distribution is given below and compared with similar data for the United States in 1952. 1-/ Fathers Farmers Blue Collar Whi te Collar Occupations Occupations Percenty!!. Kenya !!§! Kenya USA Kenya Y.[il Blue Collar occ. 79.5 66.8 76:,.4 68.4 60.1 34.4 Sons --=- White Collar occ. 20.5 33.2 23.6 31.6- 39.9 65.6 Ihtaber of ObsE!rvations 1631 151 441 275 208 208 The data show a greater upward mobility in the U.S.A. This would become even clearer if' the farming sons were included. A cOZlservative estimate / s,,"s that about three out of tour sons of farllers reDUlin in agriculture. 2 Although the comparability of the two sets of data ~y be limited, the ques- tion remains, which social, econo~ic, educational and political fac'tors con- tribute most to social mobili ty~" is education the pri:ncipal factor? This question is of more than theoretical importance for many of the emerging nations on the African continent who have made open educational opport,'ADity one of their principal goals. g. Conclud1ns Remarks 32. Marty important issues have been brietly spe~culated upon during our description ot Kenya's urban labor force in this chapter, all of them separate from the main is sue we deal with in the remfdnder of the study. We bring the most' important ones together here as a Jt'eminder ot the variety of phenomena connected with education, some whose et:fects are measurable in monetary terms, and aome whose are not. Source: G. E. Lenski, "Trends in Inter-GeneratiOrn&1. Occupat~ODaJ. Mobility in the United States'Y, American Sociological Re ..~, October 1958, re- printta,d in Wyes and Labor Mobility, Supplement. No.1 t OECD, Paris 1966. The original. sample alao included interriewees (BODS) who were farmers; they were eliminated from the above table in orcier to make the results comparable. 2/ It i8 assumed"",that the proportion ot sons ot (lnoll-tarmers who choose tva- i08 as their occupation is negligibly small (\th4e tigures in the U.S.A • • saple were 1.8 and 1.3 percent, respectively). - 76 - 33. The reduction. in fudly size, spontaneous and apparently associated with both greater education and urbanization, can be taken as a signal that an officiaJ. taaily' planning program might be well received in Kenya. In general, haring teverchildren will result in important quantifiable long- run saviDg8, which should be taken into account as indirect or secondary benefits. 34.~\ Education's role as a catalyst of migratory movements mq not be bene~~cial from 8ociety's point ot new: increasing geographic mobility mq tend in the long run to widen regional differences in wealth ratber than n&rrow'them. Therefore, if a aore equitable distribution of income is de- sirable, educational. opportunities JIl81' have to be accompanied by appropriate economic and fiscal lle&Bures. 35. As the educ·ationaliJystem has released more and more school- leavera into the labor market, there has been a gradual improvement of "standard"· educational qualifications acceptable in a given occupational category. This raises the question of the comparability over time of nom- inal educational qualifications. 36. The con8iderable movement fro. one occupational category to another in our 8uney 8ample is probably less an indicator of intrinsic social mobility and more an indicator of the transitory opening of employment opportunities following independence. Education should thus not be relied upon too heavily to bring about the best possible utilization of the country's ; stock of human talent. - 1'7 - v. EARNIllGS AND ,SCHOOLING, SOCIO-ECONOMIC AND OCCUPATIONAL CHARACTERISTICS 1. The Approach 1. A number ot studies in both developed and de"".rl~~_ ~.;,:g countries have established that the,~e is a posi tive relation between a person's earnings and the amount ot :torm.&1 schooling he haa received. 11 For the moat part, these studies estimate average earnings of individuals grouped by sex, age, urb8D and rural residency, aDd number ot years ot schooliq. They attribute the differences in lifetime earnings among individuals with Tarious schooling levels to the" schoolinS they have had. 2. other studies have shawn that differences in schooling are no·t the only reason for two ind! vi dual s ' earnings to differ. other factors such as home background and achievement motivation also affect earnings, and additional schooling contributes only a part of the ~tference in esrD- il1gs between those with different &BlOunts of schooliDg. Y But even those studies which make some adJustment for non-schooling tactors senerallJ do so unifol'llly at ever)" schooling level. ~ There are also other factors which .., distort earnings differentials between tho.e with var.ring amounts of schooling. They stem tram union action, gOTerDMDt employment, tarif:ts t subsidies, and national employment programs, for exaaple. When these effects are important, they will influence wases paid ad possibly ear"'ings d1fter- ences between 8chooling leTe1. as well. 3. We haTe attempted to improve on earlier treatment a ot educational. benef!ts by taking these 'lectors into account. Using detailed data col.- lected trom our 1968 Labor Force SurYeY in a multiple regreasion anal7aia, '!I Amonl others, see Gary S. Becker, "Investment in Human Capital: A Theo- retical AJlaly'si.; " Journal of Political EconOllY (supplement) LXX (Oct. 1962); Th.todore W. Schultz,· The Economic Value ot !ducatiOD, Bew York: Colcabia Uniftnity Press, 1963; Martin CarD07. "EarDias. _d Schoolins in Mexico.," EcoDollicDeftl~nt aDd Cultural Chap, Jul7, 1967; T. P. Schultz, lReturna to Education in _ t a t Col_bla, The Bud Corporation, MeJK)radui. RM-J645-RC/JJ.D, Sept_ber 19 8, a4Sa.lel Bowles, Pl8DDiDi () Education, :tor Econc:.ic Growth, Harn.rd Uni'Verait7 Pre.s, to be publi.hed in 1969. if See l4vud F. DeniQon, The Source. ot Growth in til~\ United;§tiatea ad the Alternatift. Before U., C~ttee tor ICODCllllic Deve1~Dtt New York, 1962. Aa,di tional e&I"Dinp due to factor. like background aDd aotiTation CaD be tbought ot as returns to education co.t.'i(i~;~d:~~uts:lde formal achoolins;. " JI Howeftr.Deni.on doe. differentiate between secODda17 aDd univenit7 leTela, •••-iga.iDg 60 percent of incc:.e differential. to schoolinl at the secoDd&I'J' ad 66percen~' at the universit7 left1. Ii - 78 - we can derive earnings figures which are corrected for indi vidu&1.s' socio- economic and occupational characteristics, including union membership and government sector employment. As is mentioned in Chapter VI, long-term distortion. such &8 tariffs, subsidies, and national employment programJ!, are not corrected for because they are a fairly permanent teature of the economic framework, and belong to a set of political decisions outside the scope ot this study. 4. The analysis of earn.ings is first done for urban areas. Each of the s1% sex-ethnic groups in Kenya is treated separately: Atrican males, African fell&l.es, European males and females, and Asian males and temales. Each ot these six groups is broken down into sub-groups with the sue years ot schooling, so that schooling, ethnic group, and sex are parameters, and the variation in earnings is estimated as a function ot the variation in the Don-schooling variables. The data basis, model, and earnings profiles ~e discussed in Sections 2, 3, and 4 below. 5. In same cases, lack of a sutticient number of observations limit. the anal7sis. For this reason, no multi var:l.ate regression estimates of earnings are possible for European males, regardless of schooling level, 01' for 8IlY single sex-ethnic group with more than 11 years ot schooling. For the same reason, we C&Dn!)t correct the proftles ot females, Asian males, or ot persons with more than 11 years of schooling for socio-economic or other variables. Thus, African males with less than II years of schQOling constitute the group for which information is most abundant and the most detailed analysis possible, and we start wi1:;h them. African females, non- Africans, and persons with more than 11 yeaJ:"s of schooling are deolt with afterwards • 6. In Section 5, we digress to take a brief look at the considerable amount ot detailed data available from our Labor Force Survey to see how the adJustments change earnings and what sane of the specific explanators ot earnings differences in Kenya are. Unlike the earnings figurel, most ot the results of this digression are tenta.·ti~, mere 'hints ot sociological md occupational phenomena in Kenya. Nevertheless. they give a glimpse ot the wq in which education, society, and the econo.my interact. 7. Section 6 consider. the effects of education on income tor Kenya's large rur..:L popultltion. Data 0:1 gross income, total and non-tarm, for familie. ot .mall 1~~dholder8 by level of education are available from the Economic Survey ot Ce'ntral Province, 1963/64. From these data we derive the incoaea ot the head of household himself. Two sets ot age-earnings protiles tor landowning heads of household, each based on difterent &8- auaptions, are pre.ented in Tables 5.7 and 5.8. No correction ot the urban rates tor aocio-economic background is possible. 8. Earnings data frail the present chapter and cost data from Chapter III are the input for the next chapter, Chapter VI, where the rates of return to education in 1968 are calculated. All earnings data used in Chapter VI are deriftd in this chapter, but not all of the age-earnings protiles ve calculate here are used in Chapter VI. The profiles for African males ~\th II years ot achooling or less ,unadjusted except tor age (Table - 79 - 5.1), aDd adjusted tor socio-econanic variables (Table 5.2); the unadJusted profiles for African females with 11 years ot schooling or leaa (TeibJ e 5.4, 4 columns 1-5), and the unadJusted profiles tor male Africans wi'lih 12-17 ,.ears ot schooling (Table 5.5, columns 3 and 5) are all USed in Chapter VI. Rates ot return in rural are.. are estimated froll both sets of rural pro- files (Tables 5.7 and 5.8). SaDe of the profiles are not developed into rates of return because ot statistical significance probleM and the rela- tive unimportance ot:these groups in the labor force, tor example, the pro- files to.r non-African females (Table 5.4, colUJllla 6 and 7) and tor Asian males (ALmex Table 5.6). The profile in Table 5.3, whi ch is adjusted 'tor both socio-economic and occupational variables, is not used. Because ot the close correlation between job characteristics and schooling level, correcting for occupational variables overadJusts for the effect of 8chool- ing on earninss. 2. The Data Bas is 9. The study makes use of cross-section urban earnioss data for Kenya, collected in the Labor Force Survey we made in Januar,y/Februar,y 1968. We took a sample of 4,742 wage-earners in that period in the cities of Nairobi, Manb as a , and Nakuru. The sample i8 .stratified b,. size of tirm, c1 ty, and private and government sector. An effort 'Was made to slIIiIlple proportionately to total employment in each city and size of firm group in the private sector. The . . :firms sampled were selected randomly trom the list of firma participatiDg in a survey with nearly complete coverage conducted in 1967 by the Ministry' of Econanic Pl8DDing. The wage-earners in the public sector were not ae- /) lected at random. They numbered less than 15 percent of the total sample. 10. Information was collected thr.ough a questionnaire, in direct inter.. views with workers. The questionnaire provided data on the worker's wage or salary, sex, the number of years ot schOOling completed, the type of .chool attended, age, father's occupation, puents' literac;y, ethn:i.c origin, the nwaber ot years spent in the city-where interviewed, age at finishing .chool and beginning vork, number ot yeFs with the present employer, whether or not currently neei ving formal education, Whether having taken or taking on-the-job training, ud, if African, his or her tribe. In addition, 8. work.er w&8clusitied by tbe type ot industry in which he lTorked, public or private sector, size ot firm and the citY' in which it was located, and Whether or not he belonged to a union. (For a more complete description ot the survey and a copy ot the questionnaire, see Appendix B, "Scope of Labor Porce Sample SurTe,.".) 3. The Model 1/ 11.1\ Since we want to analyze the data to see how earnings V93:'Y' u the other tacts about a wase-earner change, we construct a simple equation ex- pressing the dependencY' of income on all the other factors: (age, years ot schooling,) Incom.e • function of' (father's occupation... ) ( ••• union. memberahip. ) 1/ Sections 3 and 4 and Appendix D have greatly benefited trom the author.' discussions with Mr. R. C. Manning and tram his written comments. -80- A list ot the tactors, or variables , i . given in Annex Table 5.1. Most ot them are yes-no variables; in other words, tor any given individual, either they apply or they do not. They can be quantitied by conTerting them into "du.;;r variables", which take a value ot unity whenever the wage-earner bas that particular characteristic, and zero when be does not.!/ Thus, it an interriewee is, Sill', a clerk and his tat~er a farmer, the respective occu- pational categories (VII and I) reeei ve a: value ot un!ty, and the other nine categories in each group receive a value ot zero. Even in the cue c' ot -set a cODtinuoua variable, dU111117 variables are used, since ap is not' necessari17 linearly related to earnings, but is rather a proq tor job experience and maturity. 12. Using mathematical notation, the equation becaD.ea: j • 1, 2, ••. m k • 1, 2, ••• 1 where the Xj are age variables and the Zk are groups ot socio-economic and occupational. variables. Since there is no theoretic~ buis tor choosing a specitic tunctional torm tor I • t(Xj, ZJt}, the relationship between the depencJent and independent variables i~ assumed to. be linear. The error term Ui accounts tor randoll variations in income, which a,rc, Dot covered by the chages in the.:JX or Zt., and is usumed to have the usual properties J necessar,y tor least squares estimation ot the parameters. (5.2) J 1 .1 + \ Y1 • Bo + \ ' B X ' Ck Z1k + U1 J. 1, 2, ••• II L L i t . 1, 2, ••• 1 j k i - 1, 2, ••• n Xj and Zit are the independent variables and Ii i. the depend~llt one. 13. Atter the regressions are performed, one for each schooling group, the predictive torm ot the equation (5.2). ,obtained by least square., i. uaed to derive the edut!at10n~arn1ngs profile tor each achooling group. This i. given in (5.3). 'l:I. Since, tor each indiT1du&l, the Talues tor a group ot dlDlllD;Y variables covering ODe characteristic by detinition add up to unity -- the indi- vidual talls in 0Dl7 one age group, one occupation catelory, etc. -- the required linear independence exists neither among the d~ vari- able. vithin the 11"0up t Dor between those groups ad the implicit con- stat tera. However t it we vithdraw one dUllllr1' variable tna each group t we CaD proper17 .sti.ate a reduced set ot par~ter.. The intercept tera will then represent the income ot the subpopulation de.cribed by the withheld ~ variables, ad. the individual coefficients will be .st1Jlate. ot the income ditterentials attributable to their correspond- ing Tariable~. 'l'hat such ,ere-parametrization" is inevi tably arbitrary Call be seen by coaparins the reduced groups ot dUJllllr1' variables and the li.t ot vax-1abl•• withdrawn, both in Annex Table 5.1. - 81 - (5.3) iLj~l • bo + L b J X1J + L cit Zllt j, k, i above as j k 1\ where Ii • predicted value ot Ii 1\ b • B t the least aquares estimator of the intercept; o 0 1\ b J • Bj t the least squares eatimator of the age coefficienta; 1\ ~ • C t the least _quares esti1'd.tor of the aocio-economic k and occupational coefficients. 14. The data collected give valuea for Yi. XJ and Zk." one complete set ot valuea fo~, each of the 4,742 intervieweea. Theae set. are firat di'Yided into the<~1J;)lree major ethnic groups (Atrican, Asian, European) and then 8ubdivided into males IDd temales, resulting in 8ix data groups. Be- cause of their preponderance i~ the labor torce, we focus on African males t who make up 75 percent of thej~otal. Comparative data on females and nOD- A1"ricans are diacusaed later in the chapter. ~• The Relationships between Educati~". Age! PAd Earnings a. The Case ot Urban African Males with 11 or Fewer Yeare ot SChOOling ke-!t:d.1usted Education-,Earnings Profiles 15 (j To estimate for a given schooling group the _8Il earnings U80- ciated vith 81J7 particular ase requires grouping the data by years of schooling. C!/ The est'imate ot average monthly earnings for each age-edu- cation category would then take into account both the influence ot age and education separately t and their combined (interaction) effect. At two levels 'ot schooling (end of the seventh and eleventh years) the Dub-samples are turther divided according to the results obtained in the ex.. tha.t take place at these two points. 16. The age-ail.1uated education-earnings profile. t Q8 gi van in Table 5.1, are buedon the resrea.ion estimates given in Annex Table 5.2. In this resre•• ion, only ase variables vere held constant. It follows fram c the de.c~lption otthe model that the intercept terms tor each sChooling group in Almex Table 5.2 repre.ent the meUl earning. attributable to the vithclrawn dUllDl' variable, which in the case of the group reterring to ase i. the 35-~~ 7ean 88e category. Theretore t tor each IchooliD8 group t one mu.t add the coefficient. ot the other ase !roup. to the intercept ter.m in Annex ~able 5.2 to tind the mean earnings for thease group. in Table 5.1. Y We allO pertonaed the regre.aion analysi. OD the lample of African JD&lel using 7eare of schooling as a continuous variable. The re.ult. are described in Appendix C. ~~~ Table 5.1: Kenya: Unadjusted Age-Earningf!~prOfileSt African ~~ Hales ,. by Years of SchciColing, 19 B (KShlmonth ) Years of Schoolins ~ - 0-2 3 - 2 7Fl.1 7Pa 7of1. 7 All ...2.... nP& 11 All ~J.4 224* 224* 1.5 - 16 60* 60* 17 - 19 194 231 209 278 275 257 227 405 405 20-24 285 258 260 28;3 335 278 357 573 619 25 - 29 288 326 326 402 505 379 481 755 830 30- 34 325* 367* 410*- 581* 531* 495* 632 1,075* 1,182* 35- 44 335 390 427 604 786 505 898 1,404 1,372 115 -511 326* 3~7* 450*- 893 181** 616 739* 140H 140** co I'\) 55 + 314* 353* 673* 673* 3,000** Source: Labor Force Survey, January/February, 1968; see Annex Table 5.2. Note: 1. Primary school completed, failed in KPE. 3. Primary school completed, qualified on KPE for entry into 2• Primary school completed, passed KPE. Form I of secondar,y school but did not continue. 4. Form IV of secondary school completed, failed in CSC. * Coefficient of age dumrI\Y is not significantly different from zero at 5 percent level of significance. Therefore, earnings are not significantly different from those of the 35-44 years of age group. ** Significant, but represents only one case. - 83 - Ex!J!l')le: From column (1) ot Annex Table 5.2, we see that the intercept term tor the 0-2 years ot schooling 81"oup ia 335.0. Hence, this ti~ appears in Table 5.1 as the earnings ot the 35-44 years ot case group. Since for the same schooling level the coetficient ot the 11-19 Tears ot age group is siTen as -141.0, the earnings ot this age group at this schoolins level is giTen in Table 5.1 b7 335.0 - 141.0 • 194.0. The.· Influence of Socio-Economic Factors on EarniD8- 11 • Now that we have sh~ the :relation ot age and earniDS8 tor a gi'ftn IIchoo1ing leTe1, we want to correct the age-earniqa profile at each " Ichooling level for the etfect ot 8ocio-ecCl1C1Dic characteriatics. The cal- culation of tlle adjusted profiles (Table 5.2) ia baaed on the reault. of regressing both ase ad socio-economic variables on inco., ,ahovnin AnDex Table 5.3. While we uaed the intercept tel'll in ADDex Table '5.2 U a l-efer- ence point forderi~ng income protilea, ~@ intercepts in Annex Table 5.3 C&DDot be uaedin exactly the aue vlq becauae the latter intercepts in- clude the effect of the o1;b.er variables in the reINs.ion bes:ta.ea ase. We JIIWIt hold the aoCio-eCODaL'1c variables At lila. level to allow inCOMe of different schooling groups'\1;o be compared independently ot ,~oc1o-e~Qnomic charactel;"istics. To thia elta, each of the schooling group i'egreaa:1C1ls haa been evaluated at the _ans ot the 8ocio-econc:aic Taril&blea tor the whole aample ot African alea vith 11 or fever yean ot achOOliDg. 18. Our procedure is to 6~J'1 ve the earniq8 leftl: of the 35-44 years of sse group within each aCbooling cates0i7 b7 tint aalUll1ng that the value ot each non-ase Tariab1e which we had included in the regre.lion eatimate equals ita __ vi thin each schooling categorJ". W. then ad.1uat the 1evela ot the difterent age-earnings protile. relati"le to each other bY' correcting for difference. between the MUTalue ot each aocio-economic Tariable in each schooling group and it. _ a Talue in the a..-ple .. a whole. When ve tinish theae adJustments, vehave eatilllated the age-earniD88 pro- tilea 'at each education leYel .. it all A:tricaD male vase-eamera had the socio-economic ch-.racteriat1ca of the "aTerase ellpl07ee", described bY' the mean Taluell ot thoae characteriatica for the whole aubgroup ot Atrical1 malel with 11 or 1~.ayear8 ot achooling. 19. For each schooling group, thecalculat10D i. Mde by' takins the _an earniDga for the achooling group ad subtracting t:roa it the _va ot the product. of the age coetticients ad the __ ot the cor:reapond1ug as. peoup, i.e., the frequency ot the ase group within the IchooliDg group. The reaultins earnillS' tigure i8 added to the • • ot the products ot the coetticients ot the locto-economic variables and the difterencea between the corre.ponding _aDa ot the variablea tor the "hole a-.ple and tor the - 84 - ,+ Ichoolins group. !I We clarit.Y thil procedure by reconstructing the 0-2 ;year8 ot 8choo1iq colUIID ot Table 5.2. Ex..,le: ColUlUl ODe ot Almex, Table 5.3 gi ve. the _an coefficieitl tlt the ase cate'80ries at this acbooling leftl (1); each i8 .w.tiplied b7 the lilted in colu.. ODe ot ADDex Table 5.4 tor the corresponding .,;e group (2), U14 thele products are luaaed (3) and .ubtracted tna the M _ earnings tigure for the 0-2 7e&l'I ot achooliq group in the lut line ot Annex Tabl.5.~ (4): (1) X (2) • (3) (4) ,(4) - (3) - ace trca A.T. 5.3, tram A.T. 5.4, tna A.T. 5.4, groUP collmll (1) col\Dlll (1) "mean earniDl!" ~14 0.00 !J 15-16 0.00 17-19 -168.'8 0.01 -1.69 20-24 - 50.6 0.05 -3.67 25-29 - lt3.5 0.11 -4.78 30-3~ - 0«18 0.20 -0.16 45-54 55+ - 8.1 ... 18.9 0.22 0.07 .,11.78 -1.32 -12.37 320.6 333.0 Theretore, __ eaminls ot the 0-2 ye~1 ot Ichooling group corrected tor locio-eCODOJIL1.c ditte~lence8 vi thin the group is 333.0. ColUJlll ODeot Annex "Table 5. 3 Sivel the II ~oetticienta ' ot the categoriel tor mother'l &11dtather' I literacy, tribe, and tather' s occupation (1); each il JIUltip1ied bY' the ditterence (4) between the value ot the mean tor the correlpondiq catelOry .an tor the whole laple IiTeD in the lut colllB of Annex Table 5.4 (2) &114 the tor the correlpondiq catelOl7 tor the Ichooling group, IiYen in the tirst M_ C01UD ot ADDex T~le 5.4 (3). Theae products (5) are 1l8l84 U1d dad to the earning' tisure (6) to gi,.. a DeY _&11 e&l"l1in.. ctigure (7). 'AI '" '!'be a1c.b~aic Mri'YatiOll ot thil prQcedure and aD alt~rnat1ve toX"m ot .Itm.tion are Ii",n in Appendix D. /i - 85 - (1) (2) (3) (4) (5) non-schooling tram A.T.5.3, tram A.T.5.4, fro. A.T.5.4, Tariable colwm (1) last colUJllll colUlDl1 (1) • (2) '~ 863 1106 25 - 29 294* 229* 490-l~ 392 1090* 145i~~ 30 - 34 326* 262* 523* 909 1259-l{- 1471-l~ 35 - 44 325 306 383 1359 1649 45 - 54 345* 215* 234~~ 1797~' 55 + 415 1475* " (j Source: Labor Force Sample Survey, January/February 1968. Y . There are no significant coefficients in the 9-year schooling category e~t:iJnates. \\ ,. 'j ~" - 91 - years group. The age-earnings protile derived trom the.. regressions is shown in Annex Table 5.9. 29 • An attempt is made to construct a separate profile for the 66 African males with//post-secondary schooling, but -this produces few signifi!- cant results (see ,Table 5.5, ~olumns 1, 2, and ~) II Therefore figures tor African JD&l.es arel derived by il.ssuining that the d1.lllDD;Y' Tariable for African male equals unit)" and for other ethnic groups, zero. The coefficient thus obtained for Af'tican males is added to the intercep~" to get the earnings tor the 35-44-year-old AfricaJi males. (Table 5.5, c~o1umns 3 and 5). These figures are used in the' next chapter to estimate rates ot return. 30. While most ot the regression coefficients presented in Annex Tables 5.7 and 5.8 are insignificant ,and the 'lew significant ones so dis- persed that, individu&l.ly, they add 1ittle to our understanding, a few resul-t;s are worth mentioning. The age and sel,t-ethnic variables as a group account tor nearly two-thirds ot earnings variation in the 12-13 years of schooling group, and for almost one-half in the 14-17 years category. Adding all the other vari~bles brings the value of the coefficient of de- termination (R2) to 0.77 (0.89 for 17 years of schooling). These are much higher values than those obtained for primary and secondary' schooling. 31. Second, in comparing the percentage or-African males with '-l1igher education in each cluster ot variables with the comparable percentages for all Africans from Annex Table 5.4, we learn that more than h< of' the highly educated Africans' fathers are literate, compared to only 30 percent tor all Africans; a similar difference exi.sts as regards the literacy ot JDC)thers. Likewise, somewha't fewer highly educated Africans had fathers who were farmers than the average (33-40 percent va. 50 percent), and fewer had fathers who we,re skilled, semi-skilled or unskilled workers (33-31 va. 48 percent). A higher percentage of highlY educated workers are in fir.ms with 100 or more employees, and their occupations are predoDdnantly clerical in the 12-13 year category, and technical and protessional at the14-plus level. Vert high proportions -- 69 and 90 percent respectively -- of Africans 'With higher secondary or university education work in the public sector, compared to 10 percent for all Africans in our sample. 32. Earnings for Africans v;J.th 13 years of schooling are nearly three times as great as those for Africans with five years ot schooling, and 40 percent higher than the average earnings of Africans at the 11 yeara ot schooling level. An African with 15-17 years of schooling makes, in turn, twice as much on the average as an African With 13 years ot schooling. Average earnings of highly educated African males (13 or more years of schooling) remain below the aTerage earnings at all ages for the overall sample of all sex-ethnic groups with this amount ot education, and are considerably below the salaries ot similarly educated European males: - 92 - (in Ksh/mnth) (1) (2) (3) (4) (5) 11 years 13 years 12-1 1 zears !!! African Males African Males African Male African African Alone Alone as Variable Males Alone Male as Yare 17 - 19 405 675* \1 20 - 24 619 714* 796 1100* 2102* l(. 25 - 29 830 1326 1053 2205* 1932 30 - 34 1182* 1325 1379* 2661* 2614* 35 - 44 1372 1259 1655 2679* 3405 45 - 54 3020** 2320 3410 55+ 1017* 1909* * Not statistically significant at the 95 percent level. ** statistical.ly significant but r.epresents only one observation. - 93 - Averye Monthly Income (Kenyan shilliY8) Years 01' SChoolins 13 15-17 Ethnic Eou12 African male 1,122 2,429 European male 2,794 3,409 All sex-ethnic age groUp8 (Annex Table 5.9 ) 1,473-2,997 2,667-4,144 33. In conclusion, it can be said that data on urban African male em- ployees with 11 years ot schooling or le8s have permitted very- detailed ad- justments tor socio-economic and occupational characteristics. The protile8 obtained when occupational Characteristics are held cODutant are considered to be over-correct_~d for the eftect ot 8choolill8 on earnings. However, the protiles derived tram the regres8ions which include soc1o-econoadc T&riables are adjusted tor tactors largely exogenous to schooling aDd can be considered more appropriate measures otthe true benetits to education . than the initial unadjusted age-earnings profiles. 34. The sample size otAfrican males for upper seqondary and university schooling was too small to estimate in any meaningf.ul wat the eftects 01' the corrections tor other variables on the age-earnill8s profiles. At these highest levels of education, age-earnings protiles could be estimated in the broadest terms only, by aggregating all sex-ethnic groups. 5. The Significance ot Non-Schoolipg Variables -- A Digression 35. Betore considering data on rural incomes and schooling and the relationship between these and the urban data we haTe examined so far, we make a briet digression on the Gttect which each of the 57 non-age, non- schooiing variables has on earnings. Our observations will be limted to the group ot African males. 36. From thevut amount ot buic data in Annex Tables 5.2, 5.3. and 5.5, on which the protiles in Tables 5.1, 5.2, and 5.3 are baaed, we CaD identit.r the tew indi'Yidual variables which seem to be major explanators of income variance among African males. Annex Table 5.2 give. the coef- ficients tor the age groups without adJustments for other variables. Mean earnings and age are also tound tor each schooling group. The latter figures reveal three facts: (i) Earnings not only increase with years ot schooling. but the absolute and relative size ot the increments also increases; - 94 - (ii) Scores on the seven- and eleven-year exDS have a marked etfect on mean earningll: while average earnings tor the "qualifY" or "pass" group in the seven-year categor,r are 32 percent higher than earnings for the three to tive years of schooling group, a "fail" on the exam results in little earnings difference from the lower schooling group; (iii) Since older African men are less educated than younger ones, the average 86es as80ciated with the average earnings at each schooling leTel decrease as education !ncreues. Thu the average age ot an Atrica man vi th two years of achooling or les8 is 39, wh1.1e the average age of a man with 11 years is 25. The data imply that the Atrican male labor torce over 34 years of age has on the average about 1.5 years less education (3.9 years versus 5.4 years) than the African male labor torce under 34 years old. 37. Annex Tables 5. 3 and 5.5 shoy the results ot adjuating for Bocio- econa.ic variables, and socio-economic and occupational variables together, respective17. Amons the variables with inSignificant coefficients, i.e., variables vh1ch seem not to have aQ1 statistically relevant etfect on earnings in themaelves, when other variables are held constant, are: 'lather's occupation (with scattered exception8), f'ather' s literacy, city of' employment, and score on the esc (11th year) exam. Varinblea which do seem to be signifi- cant explanators, when other variables are constant, haTe to do with tribe, mother's literacy, tirm size, occupational category, union membership, and, particular17 for five 8Ild tewer years of schooling, sector of' employment. 38. The ti~s tor the firat tvo years of seconcla.r1 school show that the difference in earnings between those with nine years of' school and those who pus the KPE is increued by adjusting tor socio-econanic variable .. , and essentially eliminated by the correction for e~lo,y.ment characteristics. In fact, this latter correction causes earnings ot the 9-year sroup to tall sUbstantiaJ.ly below those for the 7-year group at all asea below the 35-44 bracket. Since we consider that emplo,y.ment characteristics vary system- aticaJ.ly with the amount of 8chooling received, the resulta of the adJuat- ment for occupational characteristics imply that the return to the first two 7ears ot leconcla.r1 schooling is due largely to the advantqe that these years prOTide in getting difterent sorts of Jobs than those available to priaary school leavera. 39 • Almex Table 5.4, wbich we haft alread1' used in the profile calcu- lationa, i. a further Usplq of the detail of' the data ana.l.yzed. It give8, by schooling'sroup, the percentase of individuals characterised by each of the variable'., the a'"rage ase aDd earnings, and the number ot observations for each schooling group. It ShOW8, for example, that 75 percent of Atrican males with two years or less of schooling haTe illiterate tathers and 92 percent illiterate mothers, compared to 34 percent and 54 p~rcent respectivelY tor those with U year. of' Ichooling. More interestingly .. while 70 percent I, ' - 95 of those iil the l.ut educated group are selli-skilled workers. ad half ot those with the most education work as clerks, adJdDistratorll. or foremen. rousbl1' the s . . proportion ot these two education groups have fathers who were farMrs (~6 V8. 53 percent). ~o. ~ ot the lION notable trends which show up as we 110ft :t'roIl lover to higher le...ls ot schooling tare: (1) the increue in parents ' literacy; (ii) the steadily increasing share ot members of the Kiku7U tl1be and the decreas1ng,share of Kamba tribe members; (iii) theincreue of tathel'S in semi-skilled occupations of interviewees with up to nine yeare of scbooliD8; (iv) the rapid increase in the share of vhite-collar workers ( Occ. Clerk), :trom 2 to 51 percent ot the sUIPle, as we move from the 0-2 to the 11 years ot schooling cateS017, the parabolic increase and decrease ot skilled workers t and the rapid decrease ot semi-skilled workers; f (v) the decreasing number employed in manufacturing and services, end the correspondins increue in the public sector; (vi) the increase in the percentage ot those who went to public rather than private primary schools; (vii) the increase in the percentage ot ·~ho8e taking extra-mural schooling and on-the-Job training, and (Tiii) the decreasing share ot union members. The wealth ~t data in Annex Table 5.4 could by itselt be the baais ot severaJ.. essq. on the nature ot education, society and the labor market in Keny-a. Sane of the tacts it contains prove ot Wle when we attempt to in- terpret the ditterences between adjusted and unadJu.•ted qe-earningl pro- files. 6. Rural InCCMs aDd Schooliy 41. Because ot the gap between urban ad rural income. and the limi- tations ot urbUieaployment opportunities compared to the aDDual add!tions to the labor torce, _ _ &1.781s ot the rates ot return to education in KeIV'& should not be baa.d on urban wage emplOJ1Dent alone. However. our Labor Force SurT.". includes 01117 urban workers. Consequently t the concluaiona aDd data ot the toregoing part ot this chapter, which are baed OD the survey data •. are r,leTant only to the quarter ot wqe-employed Africans who York . in KeDY&·. urban areas. (See Annex Table 5.10). The other three- quarters ot the yorkers in the monetU7 sector, not to speak ot those in the nOD-.,net&ry' sector t an not covered. We there tore need to know whether - 96 - the phenOMnon ot a return to inyest_nt in education oblerftd among urban eJlPlo7ees exists at &l.l tor the rural aaJority ot the population. 42~. Education brings little or no monetary return. to tvo larse aepents o{t the country' a Il&le labor torce: the unempl07ed, and tarm laborers. Fara laborers' earninss are a tunction ot ase rather than schooling. Their wacea raD8e trom about !Cah 2 to 3' per dq and are generally supple_lted" by rations valued at Ksh 25 or ao per JIOnth. by housing, and by a ..all plot ot land to be operated by the tar.. worker a and their t.-ilies tor their own nee48. There is no indication that peraons who went to school have a better chance ot tinding thia kind ot eaplo:plent than haye illiterate people. On the contr&r7. African aDd non-African emplO1'ers alike seem at present to be 80118What skeptical about the luitability ot educated persons tor farm labor, quoting their uarealiatic Job expectations and critical attitude .. maJor 4H.vbacka. ! 7 43. On the other hand", the l~pat aepent ot the laboi"' torce, aelt- maplO1ed all&ll. laDdcnmerl, detinitely seem to earn aore when they haft had JIOre education. These men operate small farms, CUltivate cash crops to seae extent. ad more otten than not derive incc:ae trom occ.. ional or regular Don-tara emplO7MDt .. well. The total nUllber ot howseholda whose heads are amall laDdovnera DIal' be anywhere between 900,000 ad one mUion. In acae cues, the head ot houaehold or other taaily _bera aq be purauing urban ell,P107Jll8Dt while the taily at~. in the village ad works on the small t&l"Jl; tood aDd otber cOllmOdities are sent to tovn frca. tiM to time. ad cuh remittancea flov back to the village. Y Por the y ..t majority of Africa tudliea, haweTer, their tara and the village are their only sources ot income. ". Table 5.6, bued on data collected bY' the KeDYaD gonrDJDent !rom a~ 800 rural. houaehold. in the Central Province in 1963/61J t shows that total e~nsa ot auch • ..:u. landowning tailies cl.~J.y increue with the education ot the head ot houaehold. 'J/ In this table t ve der1..., figurea tor head ot household inco_ trca gros. tamily inco.-e tigures. We firat aubtract trc:a groaa tui17 t&l'Jll incOlDe the average inplici t value of taaily tam labor other than the taiJ.y head and ot h~red labor (Kah 165 per acre !I Ccaaunicated by R. Poaner, Un!w:rsity gt Michigan, Ann Harbor, Mieh.. who ia at pr.aent c~letiDg a study on the rural labor torce in KeQ1a. gj AI haa been pointed out in Chapter II t this ayste•.•~_ to be gradua.l.l7 41lappeariQ. the .At:ricUl "tm:-Set laborer" ot toner tiM. beins repl ..~ed by a per.u.ent urban v01;'ker. . *JJ The.. data are b . .ed OIl the Econaaic l;Surfty ~t the Central Province, 1963/640 UD4ertak. b7 the Statistic. Division of the Miniatry ot Eco- nQlic Pl-.uiDi aDd D~Teloplllent, _d coaaunicated b;y Dr. B. F. Ma•• ell, Staotord rood B •• earch ID.t1tute. Although it 11 not spec1fiedin th~1 • ..,le whether .:the rural. hOWl_bold head 11 !\ ladOvDer, we ~ave u'UJ¥ld here that he is. ,It is be,lieved that all but three or four in the sa\Jllple did ow land. 7,:",,:..:< '* I , (::""-"'''''':'''';IIi'lI'.ilq,:::~''')iI-..p'''~;J't"·· ~ ~:r,- ,'iii ..... ~. ,~ Table 5.6: ~: '---- ~ ~- - "':'J~~ -==" -=--.=-, - of Household Head, 1963-64, T Illiterate Literate, --1-3years 4-8 years 9 or more years nO.schooling no schooling schooling schooling schooling Gross Family Income (Ksh/yr. ) 1709 (1566) 2553 (4539) 3341- (4234) 4116 (4377) 6284 (5916) Non-Farm Family Incom (Ksh/yr.) 804 (812) 1528 (3700) 2050 (3791) 2635 (3508) 4352 (4702) Age of Household Head (Years) .51 ( 15) 48 ( 14) 45 ( 14) 35 ( 10) 39 ( 11) Number of Observations 408 130 80 149 21 Average Fazm -'Si..ze (acres) 6·,1 (13.1) 6.7 (5.9) 9.2 (9.5) 7.2 (8.8) 9.5 (9.7) I Farm In~ome of Ho~01d \0 Head' (~b/yr.; ) -196 (1779) -147 (1407) -312 (1746) 228 (1664) 263 (2216) -.l Non-Farm In~ome of~use- hold Head (Ksh/yr •. 502 955 1281 1647 2720 Gross Income of Household Head (Ksh/yr.) 306 - 818 969 1812. 2983 Standard deviations are in parentheses. l! Estimated as gross income minus llon-farm. :income minus [[165 Ksh/acre imputed labor cost + 10 Kab/acre _ imputed land rent) times average nUDlber of acrey /2 Assumes that 0.6 .family meriftJers (in m&le equivalent terms) besides the head of household, hav.ing the - same schooling as household head, eam non-fam income. Source: Economic SUrvey of Centra:;t PrcYvince, Statistics Ilivision of Ministry of Economic Planning and Develop- ment. Data conmnmicated "Py- B. F. Hassell, Stanford Food Research Institute. - 98 - UIl~ Y) t ad al.o aubtract a ainiaull rentu value ot lud (ICah 10 per acre armually). Thil t&:rJI incale to the taai17 head, that 11, t&1"ll inco_ Det ot tud17 labor, hired labor, and lad inputs, i. negati ve or ve;ry low. ~~' ~5u;1 We theD ••tiaa'te non--fUll head ot houaehold incCllle in Table 5.6 by ...ua1as that _ a"rase ot 0.4 _n other thaD the t.a117 head aDd 1.5 VCMD (0.2 :I:n MIl-equiTalent I Y), or 0.6 men other thaD the taa1~ head, are available to York· OQtlide the tara. It 1s turther ...u.ed that taaily _libers other thaD the taaily head haft about the same Ichooliag he doe •• Hence, 0.6/1.6, or a Ihare ot 3/8, il 8ubtracted fro. a household'a non- tar.a iDcome to 7ie14 the part that can be attributed to its head. ~6. The figure. in Table 5.6 Ihow that while the incre..e 1n tarm inc~ which goel with tldditional education of the household head. is signiti- CUlt, .uch ot the pq-ott to more education is in hisher non-tarm incOM. The ,protitability ot education in rural Kenya, therefore, dependa to a large extent on the .uppl1' ot ott-t&1"ll jobs. ~ T• FrCI'A the data in Table 5.6, we malte tvo estimates ot the gr088 inca.! ot head ot household br education level, one which is corrected to'r acre.._ aDd tudly lize differentials between age groups and between edu- cation levels tor a 8iTen age group (Table 5.8), and one Wiirch is not (Table 5.7) • 48. To arriTe at the Table 5.7 protiles, the implicit mean earnil18s ot the head ot h0U8ehold shown in Table 5.6 must first be distributed by age. Mean inc~ ot the head, ot household i8' distributed by sse and edu- cation leTel in the I . . Y8¥~hat total tamily earnings are distributed by ace 8D4 etiucation leftl in the rural lample. Annex Table 5.11 .hon mean tud17 incc:.e by' leTel ot education and 88e, trom which ye derive age.... earninSI protile. tor houa.hold heads 8howritn Table 5.7. These protiles are used to e.t1Jlate one set ot rates ot return to investment in 8chooling tor rural laDdoVner. in Chapter VI. Y Data troa the Economc Burny ot Central PrOYince, 1963-64, indicate that the ayerage nu.her ot tami~y man-d.,s put in on the average 3.9 acre tan is 154 dqa tor _n, 188 dqs tor wc:lMn, ad 26 dql tor chil4ren.-- There are 1.2 men, 1.5 yeaen, and 2.8 children in the awrase tara taail¥. Aftrap vas.s are 2.19 Ihillinp/dq tor _n ad 1.60 tor w~n,. It 1. . . . . .d that children earn 1.00 .hillins. d&1l7. The aYer_ lap11c1t taaily labor cost, other tha tor the tudly head (..- .,..d to be ODe ot the _n in the tu1~) 1. there tore 152 .hillingl/acre. 13 .hillinp/acre i. the awrase co.t ot hired labor. Y WClMn h .... a probability ot 0.2 relative to _n ot tindiD8 vork ott the tara ad l'ecei'ft about 0.73 the vqes ot men (2.19 ICahtor _n VI. 1.60 ICah tor vc:.en), .0 a woman is veishted as 0.146 ot a JUIl-equiTalent. - 99 - 49. Since the estimates ot Table 5.7 are UDcoloerected tor acreage ditterences tOl" difterent age . CQnaider each set of adJuatments on the urban rates aDd 1ta ettect on. tl;1e; rat••. separately ,. ~~i,. ~)le rural. ratea, calculate the cc.bined rates,? an4. pi'8aent a .UIIII&l7 ot a.ll rates. TIle comparisOD ot thea. rate. with alteraat1Te iD'Yestment rat•• is :reserved for Chapter X. 2. The CalCNlatioD ot the Rat,••, • the Source. ot Data j 12. .et. The' general proced~.. rQZ, calculating a r.te ot ret~~ to an in.....t- ment is to'. find that rate vhicb~ the discounted value ot the t1me stre_ of its costa equal to that of the', time atre .. ot the benetits accruing to -it. In equation torm: t Yi (6.1) ----~-.~ 0 ( 1 + r )i, i =1 where Ii • the 41fterence :til 7ear (i) between the net costs aad the Bet benetita ot add!tional schooling r -= the internal rate ot return n • the leDgth ot the pro.pective york lite in 7ears. This _thod ot relating costs aDd, benet1ts was chosen because, unlike cost- benetit ratio't the internal rate ot return to educaticm lnftat.ent ts comparable to rates ot returnoQt al.ternatl'Ye kinds ot i~'Y••t_nt•• "!I For the purpose8otthis 8;t.~'. gO.,.JnPD8nt empl07••s in rural. are.. , such .. priaarJ' 8Qhool teu••rs., ~ be considered p..p of urban e1lP1OY~ ment t. aince their reDlunerai;::l\QIl, to~ the S8Jlle ~.. .. urbaD govern- MDt apl07.e.-. - 107 - a. Benefit. 13. To translate the Chapter V earnings profiles, considered to be the total benetits ot having a given 8IlOUIlt of schooling, into net benefits of taking an additional amount ot schooling, we calculate the difference between the earnings strelUU ot the two schooling levels to be compared. Linear interpolations are made between midpoints ot 8I;e sroups in the profile tables to obtain continuous earnings profiles. Where an earnings figure in the pro- tile table is statistically inl:!.gniticent, a linear interpolation is made between the·· tigures tor the next higher and next lover 88e groups; where there is no significant tigure in the highelt age group, it is aaaigned the income ~f the next lower group. 14. The first step is the calculation of unadJusted rates: unadjusted net benefit data are derived from Table 5.1 (unadjusted ege-earnings profile tor Atrican males, 0-11 :rears of schooling), 8I1d Table 5.5, C01UlDll8 3 and 5 (unadjusted proti1e for African malel, 13-17 years ot IChooling). The other tables of Chapter V are cited &I they become releTant to the discU8sion. b. Cost. 15. The costs ot schooling borne by the private individual are d1tter- ent from those borne by the larger social. unit. Both include earnings fore- gone while attending school, but private COlts add only school tees pqed whereas public costs add the entire operating and capital COlts incurred by the schooling systEm. 16. Earnings foregone by age and scl1oo1iDS groups can be estimated trom the earnings protiles in Tables 5.1 - 5.5. The timing ot earnings foregone can be determined by adding to average school-leaving age in Annex Table 3.1 the averase del", in tinding work tor school 1eavera , given in , J Annex Table 6.1. The starting point tor work lite is tound to be 16-18. ~ 17. Because very tev children work, gj priTate coata up to the end ot primary school consiat ot Ichool tees only, Ihown in Table 3.15. Though the prim&r1 school courle run. leven 7earl, Itudents need on the average eight years to complete it; halt of the cost of thil extra year i8 .. signed to the cost ot the tirat tour years, aDd half to that ot the 1ut three. Starting at ~e 17, earniDg_ tore gone beco_ aD important part ot private cOlta. The ri~~cond&r7 Ichool tee .tructurs ia Ihcnm in Table 3.16. /} ------------~i ~f 1:1 We ..SUlletr ~t no one under 17 il ellPloye4 in the urban sector; o~ one --------------- Atrican JRa.i~J 11114 26 Atrican t_ale. in our lurvey were younger than 17 (le.s than 1 percent ot the (~ricaI'J oblenationl). In some developing countriel which have had long perioct. ot rapid growth, c • t., in Latin ~ric.. , children can find work in II.Ul7 urban lenices such as 8ellinl neWSpaperl, Ihining .hoes, working .. h0U8e-b07l. However, 1n KeD7&, such politions are tilled by adult •• Y Children conltituted onlY 0.2 percent ot urb_ employment in 1966. - 108 - 18. Pri...ate co.t. tor African. by qe, aex, aDd educatiODal leyel, doeri'ftd trca all thia intonaation, are ahown in T&ble 6.1. Since higher eecondar7 aDd univerai ty .tuden:t. lenerally do not p~ tees. theiJ:" private coat. are It.ited to earnin,. tore~net .hown in the last two column•• 19. Intomation on operating &,\d capital coata nece •• ary to calculate the aocial rate. ot return i8DOt quite .. complete as data on .fees and in- come toresone. PriJIarT teacher salaries are available by school in a depend- able torm tor two counties. Ion-teacher expenditures are only .a 811&11 frac- tion ot,total operating cost. at the priJDa17 level (about 10-15 perc!!,nt) and tb~ are d8rived on a per pupil buis trcathe county cOUDc~r1:)UCllets of tour q,ntral aDd Eastern ProYiDce countie.. Aa &ascribed in Chapter III, a r~dol1 selection ot schools in those two counti.s 'i. used as the baais tar the oper- ating cost estimatea. 20. These figure. do not pendt a differentiation by' 7eU"~1.'e,~ the s _ total. cost ••t1aate per year haa beenUlled throu8houtthe pri.Ju.ry course. This i. in obvioua contraditiOll to the general. rule that the mo~e qualitied aDd hence better paid teacher. tend to instruct the hipergr&de., aDd that theretore more ot the cost should be distributed tohiper grade•• Distributing th. cost eq~~ will lead to a distortion of the •.ocial r~te­ ot-"turn struct,ure taTOr1\1a the 6-7 yean ot schoolinl group at the e~;'t;ense of the 3-5 years cateSOJ7 ,ud ot the latter in ccmpari.on to the 0-2 tl1~ua group. 21. It in order to pt aD idea ot'the;,JDaPitude ot tl~e di.tortion we 2J.as,.., tor exaaple. thatpubl1c .ubsid1zat~\on ot prilllUTj/achool1ng cOTered alyqs the . . . .hare ot total .ocial ~Q8t. ~nu. income tOr8I01le. then the ditterence between tee. paid in StUl~\d81 and VII (!Cah 53 VB. IC.h 70) 'would imp17 that tQtal .oc1&1 coat,>:) pe1\, pupil misht T&rI' by u :lIUch .. Kah 30 (trca Kah 132 in Studard J: to Kiah IT4 in Studard VII)."::~er thu not at a;J.l .. we ........ Even thi8 estimate could turn out to ~,1Dadequate. If local authorit~es deliberately' or UDcon.ciou.l1' tollow .. path iQt progr.s- siTe ')1b.. 1di~~tt=, / If"" '<, the e1l'Or "', could haft .erious proportions. 1:/ Hovever f a coapar~a{l~c:~t the d1ttere~~es bet"een the .e-ad3usted rates ~or the 5-7 ~d 2-4 y• .,.... intenal. .hon le.. d1 versence ot the .ocial rate. from the ~!''t..,?e n structure. .. repre ••nted by' the pri T''!~ rate-ot-returntisurea, th.,n one would haTe expected in such a cu ••r' // ,-, _/ II 22 • ,v COTerase of .ecOIlda'Q' operating .co.ts i8 110" complete than ot prtmar.r co.ts: est1~at•• ot teachinl aDd non-teaching expenditures are ~'Y,.tlable by sc$ool'j\ tQr all ... i.ted ad Mintained secondar7.choo1. in Ke~,ga. The school '~o.t .data are de8cribed in detail in Chapter VII. Uni- verai tJ' operating cost. are taken trom the Uni ver.ity College Plan. The Y Datat:rom a 'rural :&:rea in neipbor:l.DI Usuda, where per-pupil costs tor the .eTenth yeremn thaD. twic. theee ,tor the tint srade, sugest that tb1. is .ore thlUl a theoreticalccmaideration. (Source: P. Fo.t.r ;aDd L. Ya.t, Population Growth UldRuralDe'Y81opment in BUl!Dda. A ~jmulatiOil ot a'Micro-Socio-Econcmc ,Syat_, College Park, Md., 1968). - 109 -- Table 6.1: Kenya: Annual Earnings Foregone and Other Private Costs, by Sex" AgelancL Years of Schooling, Africans, JJ68 (Ksh) Education Level (Years) .----------.-------.----------------~--~----.----------.------------ N&F M&F Males Males Males Females Males Age 1 - 4 4- 7 7Q - 9 7 All - 9 9 - 11 7 All - 11 13 - 17 9 .53 - 10 56 11 .59 12 29 35 13 69 14 15 70 16 494 494 494 17 1701 + 494 1602 + 494 765 + 494 18 2043 + 307 882 + 307 19 2331 + 307 1008 4 307 20 4,617 21 5,094 22 7,164 23 7,623 24 8,082 , 25 8,541 Source:- Primary ~6hoOl fees: Ministry of Education dat,a on fees by district, 1966. - 13 - 11'11Iears of education: Table- 5.5 monthly incomes are multip~f.ed by nine to obtain annual income foreBone. - Secondary school fees: In. 1966, 33,000 students in maintained and assisted schools paid an averagect 307 Ksh/yr.; about 30,000 Harambee school $Judents paid 700 Ksh/yr., or an average of J,.94 Ksh for these 6),000 s'tudents. No attempt is made to allow for bursaries or fee ,-4 ;remi.ssions. Earnings foregone are estimated by taking monthly earnings from Table 5~1 aM Table and multiplying by m.ne. For lack of data, it is assumed tha1j a neglieib1e number work while attending school, so these ea:mines foregone are not adjusted for th9se who do not forego them when going to schOOl. - 110 - , Plantigure used 1s ::tor 1969, when the university viII be J,operat1,tlg much closer to capacity., 23. Capital cost data are approximated tor primary and lIecondary schools from discussions with Ministry ot Education officials and trom con'rtl·uci~:t~,ln and equipment COlt figures tor several recently built aecon~ar.r schools. University capital. costs are de:c! vedtrom the Uni verst ty College Plan. At every level, the= value ot buildings ad equipment is estimated 8lld the capi- tal cost ill taken equal to depreciatj,on plus an altern.:~iv:e cost ot ~apitaJ. in the economy, usumed equal to 10 percent. We .. sume a ;litetime o't 40-50 years tor most school structure•• Y Primary' school capital cost proves negligible compared to current operating cost, and secondar,y school capital cost is not a very important item in total. secondary expendi'tures. At the university level, however, capital ,coat amounts to about 50 percent ottotal cost. The oper~ting and capital C08~ figures used are shoWn in Ilrable 6.2. 3. The Unadjusted Urban Rat ••' 24. The unadjuated priTate and soci&1. rate. of retu.rD to Tarious levels ot achooling fer A:rricaD m.al~ •• baaed on the coat8 and benetita just described, are ,given in colUJIID ODe ot Table 6.3. The private rates v&ry' amoD8 educational. levels but are all, YeZ'J high aDd taTOr cOlapleting 88 much education u possible rather than ~p~nl out. Private rat.s are highest tor lowel' secondary schooling as "whole (36 percent); PJ"iJaarJ cc-.e. next with 33 percent, uni versi ty vith 27 percent ad higher aecondari vi th 24 . percent • Th~ soci&lrates are lowe~ ,at each level because socially' incurred costs err running the schooling syatq, are JaUch higher than priTate tees. The SOCii~ rates are roU8hJ.y one-tb1:rd lower than the pr~v.te rates throup higher s~~onda.:ry and tvo-thirds low.~ .tor university; the higheat aocial. rate i. 2";'\ percent to secondary 8eh~,11nsj theD pri1ll&!7 .ld. th 22 percent, hiSher secondary with 15 percent f&l\d univerai t7 vi th 9 percent. A Note on African Feaalea, 14ffi'! 25 ~ Our survey data on Afric. 'female yap earnera p':nD1tted estimation of unad,1ustQd rates only' tor three ,ducational lev.ls! the whole primar;y course, the lut three years ot prim-.ry education, and the first four years of secondary school. The primary rat.s are much lQWer than the corresponding rates. for African males, but are very similar tor secondary schooling (see Table 6.:3, col'Wlln ~;)ne). 26. It is possible that tbe lIQadJuated rates vould 'be higher if' ve vere able to include .. benetita the nOD-~ketable but econo-1call1 rele- vant contributions made at hcae b7 educate'd temales t such,.. a JIIOfe etticient management of the household, auperior tutorinS ot children, aDd improved t Y The exceptiOD8 were semi-perm...JR· ani temporary DlUd-_d~attle structures used for; primary schools; see ObltP1;e:r III, pSge 48. - III - Table 6.21' Kenya: Annual Coats Per Student Used in CaloUlating Publio Rates of Return (in Kenyan pounds) Level of Operating Capitat/ Sohooling Costs Costs- Primary 1.1 0.1 - 0.5E! Lower Seoondary 87.OJ/ 3.0 -lO.oW r.:j Higher Seoondary 1~75.OJ/ ••• ~ University 690.021 625 Y Undiscounted annual aJOOrtization charges per pupil place. y >The primary range encompasses bot,h the usual mud-and-wa:" ~;~Ei structure and the mre costly send-permanent and perma.ne~it structures. Lifetimes of these structures are assumed ;iO be 10, 30, and 40 years" respectively'. J( Current cost plus Ksh 250 per student annually for the 8th and 9th years, and 500 Ish in 10th and 11th years for rent and dep- reciation on school buildings. W The lower figure refers to a day school located in Nairobi, the upper to a predonIinantly boarding school upcountry; both schools have !three streams. The assumed lifetime of the buildings is 40 years. 21 No data wailab1e; figure is probably in the neighborhood of 10-15 Kl,. §! The costs per stUdent a:tth~>U:niversity College of Nairobi have been falling steadily ov~ time as t~p.e College begins to reach capacity. The total cost figure of KIt 1,.3~\? may therefore be an overestimate in two or 'hhree years. See Uni ve~si ty of East Africa, Univer.&i:~y " 1/ \I \\ Development Plan, 1967. ~, Table 6.3: Kenya: Average Social and Private Rates of Return to Schooling for Africans b, Years of Schooling, Adjusted for !se Oiiiy, for Ase, Taxes and l-brtality , Only, and for Age' 'and Socio-Economlc Variables Only. '~-~ ----=::'~"'.':z" ':::;', (1) (2) (3) Xee.rs~ '\\ Qchoo _ Adju8t~d £or Age Only Adjusted for Age, Taxes, and Adjusted .for Age and Socio- ",'\~, -::"'-- It>rtali ty Only Economic Variables Only .I Males '~!m a1 es =I.. '<-='0"" ' , , 0. . Mal e s c'cX~- . Hortali ty- Hales Adju.$ed Adjustt3d Primary Private Social Private Social Private Social Private Social 2- 4 25.6 16.4 n.a. n.a. 25.6 15.1 30.8 16.7 5 - 7 55.1 38.4 7.1 6.6 55.1 37·7 21.7 18.0 2- 7 32.1 2l.7 M 7.1 ~ 20.9 26.0 17.9 Seconc:1aq: l-' l-' N &- 9 23.6 16.3 n·~· n!f#.. 23.6 14.8 20.6 13.7 10 -ll 52.2 33.5 n.a. n.a. 40.2 33.5 )6.1 25.8 2/ 8 - 11 ~ 23.6 33.6 19.5 31.6 24.0 32.0 ~ Higher Secondary 12 - 13 23.8 14.7 n.a. n.a. 22.9 14.7 23.8 14.7 Uhiversity . 14 - 17 27.4 8.8 Jl:a! n~~. 19!9 8:8 27!u 8.8 1/ Investment in the schooling periods in this co1wnn yields the rates of return listed in the other columns. For exanp1e, the private return to males f"or investing in the second through fourth years of schooling __ i.e., of taking f"our years instead of only one year -- is 25.6 percent, adjusted f"or age only. Y The adjustment for public sector employment brings this dow to 17.2. Sources: Co1wnn (1): Tables 5.1 (males), 5.4 (females), and 6.1 and 6.2 (both). Co1unn (2): Tables 501, 6.1 and 6.2 and Annex Tables 6.4, 6.5, and 6.7. Column (3): Tables 5.2, 6.1 and 6.2. - 113 - " · .~ tamily health. 1/ The significance of mother's literacy as shown in Annex Table 5.4 suggests that education Of females ma;y have a positive etfect on children's earnings. However t proper evaluation of the returns to temales not retlected in the wages they earn is outside the scope of this 8t~~; 27 • Insufficient survey data on females means that we have no satis- tactory returns estimate tor a group which absorbs a large part ot public and private investment in education -- 40 percent of primary pupils are girls -- without contributing a comparable part to o~ral1 benetits as . . .ured by earnings -- onl7 II percent ot total urban employees are fe- males. Same allowance is made in the Chapter IX employment projectiona tor low temale participation in the labor torce. But leaving out~.;t:-!:~ temales and considering African JD.&lea aJ.one &s we do throughou't::this chapter probably reaults in rates which would overestimate the social&lld priTate returns it applied to &11 Africans, males and temales together. 4. AdJusti. tor Tax aDd Mortality 28. Because indi vi dual a exclude direct taxes when evaluating their earnings benefits, the benefits used in tinding private rates ot return muat also be corrected for taxes. From information on tax rates in Kenya by range ot income, tulily status, aDd number ot children (Annex Table 6.2), on the &'Yerage number ot children tor African male. in our sample (Annex Table 6.3), aDd earnings data trom the protiles of Chapter V, we calculate average taxes tor Atrican males by age end level ot education tor the first 11 7eara ot schooling (Annex Table 6.4) and tor 12-17 yeara ot schooling (Annex Table 611,> • 29 • Difterences in taxes paid between schooling levels are 80 uall th&t they' do not aftect the rates ot return up to the 10-11 Y'f!arI ot .chool- ins level (Table 6.3, column (2». For the latter two ;year., the tax adJuat- Mnt cuts the rate ot return by one-tifth (trom 52.2 to ·40.2 percent). For the higher secondary cycle, (12-13 years ot schooling), the tax ettect is nesligible, since the additiona! average taxes paid by per. on. vith 11 year. ot schooling ditter little trom those paid b7 person. with 13 :veare. The !I It should be noted that even traditional marriage practices, tor these or other reasons, such &8 prestige, place a definite value on education. A recent newspaper item (New York Times. NtV' 21, 1969, p. 30) reported. on localeNort. in one AfricaD country to e.tablish a staDdard bride price cWpend1ng on tb~ bride's education. The repOrted price ratio ot 1 I 1.6 I 3 tor uneducated, pr1ma.ry educated ad .econdary educated brides , respectively. vithout • doubt implies aD ..ae._Mnt ot the 'co.ts 'W'lUrlODe 'b1 the t_~ which s.nd. a 4&upter to .chool. - 114 - tax eftect i. mo.t marked oYer the tour :rear. ot un1verl1t7, where it re- duces the 'UIladjuated rat. ot return by a quarter (27.4 to 19.9 percent; see also Annex Table 6.6). 1/ '. : , 30. In contrut to the con.1derable reduction ot priyate rates broU8ht about by the correction tor tax•• , the aocial rate. are artected hardly at all by mort&lity. 2/ Mortal1 tT" rat.. aes\Ulled to be represelltatl ve for Atricus in Kenya in the recent -put are 81 Yen in Almex Table 6.7. They .how tha.t a l5-year-old h.. a better than even chlUlce (62 percent) ot Ii viDg to the age ot 50 but a 1es8th.,;) even chance ( 45 percent) ot11ving to be 60. Linear interpolations are ~d. ~ between the t1 ve-year iDtenals in the table to permit earnings ditterence.~ to be adjusted tor mortality on an annual buis II so that the earning" profile. are continuous. The resulting adjusted social rates, cshown in TabU~' 6.3, column (2), &re'~ unchanged tor those with 10 or more yearl ot schooling. ad. are 5-10" pe,rcent less than the 'unadjusted ratel tor the lower aehooling group•• 5. Ad.1ustiy :t'or Socio-EconOldC Ditterences 31. The beneti ta (ss.-earniDU' protilel) 1.18.d in calpulating the unad- Justed rate a of return in the t1ra~ col~ of Table 6.3 1u-.p together tbe return to schooling and a beet ~'f';' other iuputa vh1cl1 JII8l'. h~ft little con- nection with addi't;ionaJ. schooling:,. To isolate the effect ot scbooliq, we moat use net benefit tigures adjuat_d tor tactors which ~ exos.Doua to schooling and yet influence earn1bp in w.,.. which do not depend on the - - 1/ The figure ot 19.9 percent ia.~8bDgt·aix percent lover~thaD the private tax-adJusted rate ot return e8'tUlated bY' Daniel C. ROgers tori univerai ty educated teachers IISee "The~l Returna to Risher Leven ot EducatiOD in Kenya", Discua8ion Pape,r 110. 59,~r, Inatitute tor De'ftlopaent Studiea, University College, Nairobi', JilDuU7 1968, Table 3, p.l6. The lover rate eltillated by Rogers 1s, PJ!O'bDq the r.ault of bot~.~lover iDComeS tor USC holders uaed by h1.~ c~dto thOlie WI_ by this 8t~ aDd somewhat higher coat t1~F.I.J Be,; theretore would have~ e.timated lover incOM': toregone by un! verli tr, a,tudenta th-. we do. ~ . Rogers &lso estimatel that the:. tiu-adJuated p;r1vate ~.te ot return to six years ot seconciary schoolills, (13 years in total) oYer tour (11 years in tot&1.) is negative, while tl1•. preaent .tudy shove .~ 22".9 percent ~..te ot ret"arn. The difference 11.. probabl1' &110 due to thcf ditterence in the elltllllate of incOMa ad coat., at the l3-,.ear level. Rogera t' 12.2 perceDt e.tiJlate tor the IQCial rate to univera1t,. training is 3 percent more than oUr est!mate, but thia :La!" '5ec... he appliea. a weh lover unua1. a... operating COlt tor uniTerlit,:' educat1oa.: 0872 ver.eua'the KJa,3l5' we Wle. He does not include capital coat. aDd & per student operating coat tor 1967 rather thll1 the lover figure ot. 1969 we' adopt,. Y High overall mortality rates in. lela cle...lopedcountriea' are usually the result ot high infant lIlOl't~ty ad; hiper mortality at.· -sea over 40-45. Inf&llt mortal! ty doe ..~ net ent.~' into the ret~~ t.o schooling, and hiper mort..:J.ity at late!" ap., ia Hanly diacOUD::ted • f - 115 - " amount of schooling a person has t such as better home training, tribal at- tiliation and other socio-economic differences, and union membership. ~be age-earnings profiles ~1U8ted tor socio-economic va~iaoles (Table 5.2) are therefore the appropriate figures to use in c.aJ.culating rates of return adjusted for socio-economic differences (see Table 6 3 s col'umn (3»; the fi effect of union membership, along with that of anoi:;her occupa.tion&l variable, government employment, are discussed in the next section. 32. Column (3) of Tab.la 6.3 shows tha.t adjusting incOme for t'l'-::]~\e, parents ' literacy and father's occupation cute the rates of ret'C.lX'tI. to pri ...· mary education by about 25 percent, reduces those to the las"t ':.. -:-rQ yes:c's of secondary school by 25-30 percent, but leaves those to the first two je~rg of secondary schooling essent1aJ.ly unchanged; the rates for h:1.gher st';:conn- ar.r and university education alao remain the same. 6. Adjusting for Occupational Variables 33. As discussed in Chapter V, adjusting the age-earnings profiles for b'oth soeio-economic and occupational. variables tends to overadjust the ea.rn- ings, because most of the occupational variables are closely related to school.. lng. Likewise, using such profiles for rate. of return is lj.kely to over- correct the rates and underestimate the returns to add! tional schooling fI Non~'theless t two of the occupational. variables, like the soeio-economic variables, are not .,the result of additional schooling and should be consi~ered separately for the:f~r partial effects on the social rate: union membership arid ·publie sector ~loyment. These factors do not change the flow of' benefits aa perceived by the private :J.ndirldual, but they do change those relevant to the society as a whole; thus their effects bear only on the social rates. a. Union Membership 34. Union membership wa~. included in our questionnaire and used as & variable in the regression 81uUysis in Chapter V. We determined that the percentage of' wage-earners who\. are union members decreases somewhat with decreased school1ne;, but among.' schooling levels the variation is not substan- tial (see Annex Table 5.4), ana, therefore union membership as a variable has no noticeable effect on age-e&rl~,~ngs profiles. The one exception i. the lower proportion of union members in tl\e 11 years of schooling group, but the rele- vant regression coefficient i8 not significant (Annex Table 5.5), so that even in this case the effect on e ..~.rnlngs is not important. Therefore, we .., safely conclude that in our pal~icu1ar case union memberahip doe. not chase the rates of return to .choolil'lg. !I y There i8 a separate argument that exogenous constraints such as union actirl\~.,~ tariffs, corporate taxes.i;~.~ubsidiea. and minimum wage laws mq caUse wages to be di~terent from those in equilibrium. It iI, how- ever. more accurate to use rates biued in this wa:y for planning pur- poses it the ai.tortion. are expected to persist. , '. - 116 - b. Public Sector EmplOl!ll!nt 35. Government wagel tor KeIlY_- with esc or highereducat10nal achieve- ment are hisher than wag.a tor similarly trained Kenyans tn the private sector. If we believe that the higher wages retlect high,r product!vi ty of gOTernment-hired secondary-school and univerlity graduate", then no adjust- Dent i. neceas&r.y. This belief would be consi.tent with the view that be- cause the government otters higher wages, it CaD hire the moat productive graduates, leaving their le.a capable claesmate. to find j'obs in the private sector. 36. ,However, in order to account for the pOllibility that government wages are arbitrarily high and do not reflect marginal prOductivity of secODdary school and 'Wliversity-educated Kenyan., we havet'*'eJC&Ddned the earnings data ot persons with this l~vel of education to:iJdentity and cor- rect for any earnings differentials which can be aasigne~;)~o government sector employment. ' 37 • .Annex Table 5.4 .howe that aD unuaually larse proportion ot KPE qualifiers (25 percent), perlona ri th tvo yeare ot aecondary' schooling (18 percent) , with full lecond&ry' achoolill8 (51 percent) 8Ildupiveraity training (69 percent l~ work tor the government'. Thele poups, then " are the ones whose protiles would be mOlt likelY to contain acme benetits from working , at higher government w88e.. However, onlY the 11 year. Qt .chooling group I shows a significant regression coetticient tor public aector emplo.yment (see Annex Table 5.5), and a DWlMtrical.ly large one at thai;. We therefore recalculate the He-earningl protile tor the ll-year srolq), correcting tor both socio-economic variable. aDd government sector employment. When co~ bined with the earniDgs protile'tor the 7-year sroup aqd the .oci~ cost of taking the tour extra years of schOOliq. this profile yields a aocial rate of return ot 17.2 percent <.ee Table 6.13). The comparable rate ot \ return for1;he tour year. ot secondary Ichool not corrected tor gO'Vermaent employment is 21. 0 p~rcent (see Table 6,.3, 001'-1 (3». 38. Which ot the two rates ve chool. .. the marginal rate ot return to 11 years ot schooling 4ependa op. whether ve belie" that the government il pqing its employee. more becaule they are luperior to their counterparts in the private sector, or whether lI'e beli..,.e that esc grliduate8 in the tvo sectors are about equally productive but that the goverDJlient over-pqs. It Pllblic sector salaries accurate~ retlect marginal produ~ti vi ties, the higher rate, which represents the averase rate to both private and public sect()r8, should be used. However, it public sect'or salaries are higher than both public aDd private 8ectorproductlv1ty left18, then they clearly exaggerate the aocial benetit. tram lnveating in the fir8t fouryeara of secondary 8chOQling, and the co~ctere gifted pupill bY' overrewarding in! t!aJ. indications of ability and underrewarding less promising students. This kind 01' school- ing system reintorces the more able students' advUltage. In auch an elit1st system the spread of exam performances will exaggerate the spread of the students' abilities, and the ability correction made above leads us to attribute too large a part of the rate ot return tor &11 the students to ability and not enough to schooling. Likewise, it the syste. is egalitarian in the sense of. making an ef'tort to intervene in faTOr 01' disadvafitaged pupils, exam scores will not reveal Itudents' ability ditte~n~s, resulting in an overvaluation of the portion ot the rate of return due. to schooling. But if we find that the schooling ayatem ia neutral to ability in the sense of discrim.i.nating not at' aJ.l 'on the grounds of ability, then individuals passing and failing an exam can be laid to have both receiTed the same amount end qu&1.ity of schooling. 47& In fact, there 1. lane' indiaatioD that Keroran schools ... a group mq be ability neutral, though individual achoola mq not b~. This impres- sion is derived trom an ~alysia ot' tbe KPE pertormance data tram MUranga and Meru Counties in which we soUlht to determine the ability neutrality or ability biu ot inputs into the .4Ucation process. We find in Chapter VII from this data that average teacb~r salaries are poei tively and signiti- C8.lltly related to the average exam· :p.~formance 01' a school.: We want to know whether' the improve~ averase' performance is brought &bout mainly by below-average candidates, mainly 'b7 above-average candif3,ates, or to a more or less equal extent by the two ~oups,. 48. In the first cue, one would expect the dilpersioJl ot individual results, measured by the standard deViation ot the candidat'es' exam Bcores from the school's mean, to 1n'creue 1.tls thlll theDlean~ fnthe second caee, more than the mean, and in the third: c'ue, by the same amdunt as the mean. This wouldcorreapond to an egal.ite:1an, elitist and abilitY neutral sc.hool system, respectively. CaJ.culating' 't~e regression ot stan. deviations on average eXaDlli.cores, we would thus enrision the slope of the regression line to be less than unity in tbe- e~al1ta:r:t&n model (and possibly ne~8,- ti ve if the cOD'Y8rgence ot indi vid.\taai rea\lilt. 1s due' to' a deiibelltate ~,ffort - 121 - rather than a concurring te.~ure ot quality improvements)) greater than un1t7 in the elitiat model, ad equ&l. t~ unity in the eaa e ot abil! ty neu- tral1t7. 49. In the case where individual schools i l l the system show any ot these three resUlts when expenditure on teachers is increased, dependJ.ng on such tactor. as, s8.Y, the quality and dedication ot the staff or the headmaster, one would expect to find the correlat1.,?n coefficient to be close to zero and the t-values of the regression coetticient not to be Ittatiati- cally s1gnit1cat. This is what actually happened in our f:i.rs'li :t"egl"ession including all the 89 schools, which yielded an R2 of 0.004 8ll(\, f\ t-value ot -0.59. 50. However, when considering the ,4.ata of the two COV.lltit'lB sepa..'rately, it turned out that the8l;I,'up of schools :~n Muranga County shcn.,ed an in8ig~ niticant positive corr~l&tion between avll,rage exam .corea and their standard deviatioDs (R2 • o.oq;~ t -= 0.14)" while ~Jhe other sample from Mereu County ahowed a significant:; negative correlation (R2 • 0.13, t -= -2.75), suggesting the presence of ratt~er strong ega.litarian features. 1/ Unfortunately, the available data did ~~ot permit 8DY conclusion about the underlying reasons. These results warrant',t'urther stud,ies on a broader geographical buis and including addi tional vai~iables. 51. The return to education for KPE tailers JD.83" be biued downward because of the certificate effect: the lower incomes of KPE failera &8 compared to KPE passers may be largely' due to lack ot a certificate, not to lower ability. Posseasion of a; certificate ie in itaelf an advantage in obtaining a better job. / 52. To useas thia certificate effect, we want to know what part of the difference in mean earnings of the 7F and 7P groups (Ksh 96) is attri- butable to the advantage the 7P group haa in choosing an occupation. To ~e an estimate t we weight the differences in occupational structure i/;;b~~ween the 1F and 7P categories by the difference in wases between the occupational group. which cha.-racterile the 1F ad 7P categories. Annex Table 5.4 sbowa two maJor ditferences in occupational structure between the 1F and 7P groups. Nine percent fewer wage earners with a KPE tail record are clerk. than those with a KPE PaBS (14 vs. 23 percent), while 13 percent more KPE tailer. are semi-skilled workers. When we weight these ditterences by Kah 201, the amount by which earnings of semi-skilled work- ers with 7 7ears of schooling differ tram thoae ot clerks with 7 years ot 11 This interpretation disregards the possibility of rigged examination re- sults in individual. schools. Since a teacheros 8ucces~', is quite often evaluated only in terma ot the number ot his pupils who\\ quality tor secondary' education, there i. .. certain temptation to mI.4nipulate the eXUl reaults. In such a cue, one would naturally expect both a high averase leore and a reduced r&D8e ot indi ridual reaults. - 122 - , , schooling (Annex Table 5.5, 7-All column), ve tind that abOut Ksh 26 (0.13 times 201) of the Kah 96 ditterenee between the 7F and 7P _811 earnings 18 attributable to the certificate effect. Tho the traction of the differellca in adjusted rates of' return to KPE tailers (social rate • ).,0.8 percent) and KPE passers (aoQial. rate • 22.7) which can be elq)lained by" the certificate ~f'fect is approximately 27 percent. Y This eftect wouldraiae the abil! ty adjusted social rate of return to completing priJa.ar1 schooliUS to about 14.0 percent. (See Table 6.13, column (5». 53. The~orrection tor exam acore. ~ also neglect the differential effecti veness of teachers. In other vords, .tudent. who pus the exam mq have better teachers than those who do not, and Y"would be underestimating the contribution of schooling it we correct t~~ exam score' ditferences. Even ii' we had differenti&l. schooling cost by student, we 1night not pick up the contribution of better teachers if they are not paid more. 'l'hie is an unknown factor in the Kenya case, but eTen .. 50 percent upward adJustment to the railers' rates as a compensation tor poorer teachers wOuld leave both social and private rates to the last three years ot primary' school tor tail- ers well beloy the rates tor all studenta. 8. Rates ot Return to Education in the Rural Sector 54. The pr:!.vate and social rate. we have been working with apply only to the 'W:"ban labor torce. In this sectioo we eltimate ~..~e.. ot return to primar;;r education in the rural. sector. We use· the two'.eta ot 1963 earnings profiles estimated in Chapter V tor the small landholding segment ot the rural !<;lector (Tables 5.7 and 5 .8) u the 'benetits in calculating the rural rates of return. T~ble 5, 8 ~f~s corrected for acrease and tara air.e difter- entials between age groups and between education levels tor a giTen age group, and Table 5.1 is not. The tinal rate. tor the rural sector. ahown ip Tables 6.6 and 6.7, have been adJusted tor the probability in each ase g.,l.oup of becoming an owner ot farJll land -- a JIL8D who can experience a concrete positi ve return to investment in schooling. The; rural r'ates are therefore corrected for the probability ot being in a poaitive yield activity, as opposed to being a hired rural laborer, tor whom additional. achooling seems to have litt1e if any pq-oft. They are not corrected tor unemployment in the rural population as a whole. 55. . Table 6.6 uses as co.t figure. the aYer~e ann~ achqoling costa per pupil in 1963: Ksh 54 in Standards 1-4, Ksh 7es in Stapdarde 5-8, and 1/ There lU\Y' also be outright discrimination asainst thoae Who failed in the KPE; this cc!,';1l4 mean that the 27 percent tigure 11 dovDward biased. Occ"lpational distribution is also _probablJ influenced by "ability", defined in this c~~text as any in~~a Dot accounted tor by schooling and socio-economic,\factors, includi~)g "innate ability" it it is believed to have economic ret\~s. It we assume that this kind ot ability plqs a role in occupationll distribution, then the 27 perC~nt f'igure is probably upward biased. - 123 - Ksh 1,2_3 in secondary schools. Fees were Kah 45, 60, and 250, respectively.!! Income tore gone tor those taking 7-10 years ot schooling is estimated as 0.75 (9/12 months) timea the income of those with _-8 yeara ot. schooling in the 18 and 19 year 88e group. Table 6.7 ia baaed on the aUle 1963 earninge protile. but on 1966 schooling costs aDd tee. as given in Tables 3.15 and 3.16. These are the aame achoo1iq coats and tees used tor the urban rates. 56. The correction tor the percentage ot thoae who own tar.ma at a given age is made by weighting the income ditterentials tram Tables 5.7 aDd 5.8 by the propability that a male ot a siyen ace vill own a farm. This requires that we eatimate the total number ot heads ot households in each age group who are a.J.l landholders. Aal\1llling that the Central Prov- ince Survey tigures are representati ve tor the whole ot rural Kenya, that there are a total of 900 ,000 8II&l1 tarma in the country, that the age ccapo- sition of the African male urban population is the same; &8 .that of the 1962 census, and that the urbu Atrican 1I&l.e population haa increased 8 percent 8DDually since 1962, it is poslible to der1.'ve fro. the, ase-brealtdovn ot the S~y population the absolute numbers ot owner. ot small tar.ma in each age- group. When this di8tribution ot landovners by age is 41.v1ded by the cor- re.ponding di.tribution ot AfricU1 lII&l.es by ~e, the age-specitic percent- ases ot landowners Ulong all Atrican JD&l.ea Y in KeDYa are obtained. They are shown in Table 6.5. The table ahows that a younguter who ieaTes priDiarT school at the age ot 15 h.. somewhat le.s than a 10 percent probability ot becoming a landowner and thus realizing the ditterential income ot Table. 5.7 and 5.8 by. the time he ia 22 years old; a more than 30 percent proba- bility at the age ot 27, etc. 57. We multiply the benetits tram Tablea 5.7 and 5.8 by the proba- bility ot receiving these benetits in Table 6.5, and discount the streUl8 ~J of costs and benetits to get the rural rates ot return shown in Table 6.6. av !I All tigur.a reter to the tormer African schools and are calculated tram budget figure. published in Ministry ot Education Triennial Survey, 1961-1963, Ministry ot GoverDllent ot Kenya, Nairobi 1964. The .econdU7 tee figures are tor d., pupils in general seeondar,y schools. Fee. in technical secondary schools were considerablY higher. gj Whether the rural rates should be adJu8ted tor the probability ot &D7 African aale being a small landholder or tor that ot a rural African male holding land depends on one's hypothesi. about posaible remigration trom the urban are.. to the countryside. In 8IlY' cue, there i8 little ditterence between the two set. ot probabilitie., since almost 90 per- cent ot African. li'Ye in rural. areas. Thus the rural. rates in Table 6.6 are very cloae to being the OTerall rural.. sector rates. 'J/ It we did not adJust tor landowning probabilities by ase, the rates tor the first /~hree years ot pri11&r7 achooling would be much lower thaD thoae presentea/illere, aince income tor both 0-1lliterates and 0-literate sroups ar~ IIl\\Cj/higber up to 25-30 years ot as. thUl incomes ot thoa. with 1-3 year8Jit schooling. However, the r.tw1~ to inve.tment in the _-8 and 9-p1ua year. ot schoolins would be Dlch higher without the adJuatment tor the likelihood ot owning a tarm. . - 124 - Table 6.5: Kenya: Dl.stribution of African Landolmers by Age, 1963-64 1/ Estimated Shares Age Group Averase !gr of LandownersY 15 - 19 17.0 0.4 20 - 24 2'2'.0 6.5 25 - 29 27.0 25.4 30 - 34 32.0 3h.o 35 - 44 39.5 44.3 45 - 54 49.5 72., 55 and over 63.3 95.4 Averas:e for all ase sroups 46.6 $1.6 Y Mldpoints' of age-groups, except last figure, 'Which' is mean. y Assull1.ng that the sample ,population was representative of Kenya's small farm sector,. and that there e:xi steel approximately 0.9 ntl.llion small farms in the country. ~ Source: Economic SurVey of Central Province, 1963-64.• - 125 - The private rural rates to inve.tment in primary .chooling c.re considerably lover than the unadjusted urban rate, shown in the lut colUllll tor purpoa.a ot comparison. At the lover secondary level, the rural social ratea are much lover than unadjusted urban .ocial rates. Rates corrected tor acreage and tamily .ize ditterentials between age group. and between education levels tor a given age group are h1sber than the rates unadjuated tor these biues, except tor the 6-10 years ot .chooling category. Table 6. 7 shovs rural rates ot\,~~turn baaed on 1963 incame .tre.... and 1966 co.ts. These rates are relevaDl~. it it i. as.umed that absolute income ditterences betveen education categories did not change trom 1963 to 1966 but that the costs ot schooling did. They are very similar to the Table 6.6 rate. but slisbtly lover. All the rural rate. tend to be biued upward becauae inco. toregone is included as a coat beginning only at ase 17 though it ~ be 1Ilportant in rural areu at lover qe.. Y 58. Three immediate conclusions can be drawn trom these figures. i. These rates for th~ rural sector are low in comparison vi th those tor the '!segment ot the labor torce employed in the urban sector, eapecially when considering the tact that the rural rates have the upward biues de- scribed above. ii. From a comparison of the first two education groups, it appears that a sizable traction ot the rate of return which per.ons with one to three years of prim&r.1 education receiTe when their income. are confronted vi th tho.e ot illiterate. C&lUlot be attributed to tormal .chooling. Y iii. The rate. to inyestment in the first year. ot aecondary .chool tor rural labor are JaUch ,lower tban tor urban labor. 9. Rural aDd Urban UneDlRloyment 59. The social rates which we have pre.ented so tar tor the urban an~ rural sectors fail to retlect the fact that the earninls protiles on !/ See Phillips Foster and Larry Yost, "Population Growth and Rural Develop- ment in B ana: A Simulation ot a Micro-Socio-Economic S .tem", Miscel- laneous Publication No. 21, Agricultural. Experiment Station, Unlnr.ity of M&ry'land, April 1968 t Collese Park, Md.. 2/ Since a perBon's becoming literate by his or her own ettort. is obviou.lJ much more a question ot individual ability than ot the aocio-economic tactors dealt wi til in the earlier .ection. ot this chapter. the ratea for the first two education groups .., haye yer,y difterent ability aDd background components and thua ~ not be atrictly ca.p&rable. Table 6.6: Kenya: R u R A L U R BAN Dlfference in Education Income Unadjusted for Income Adjusted for Income Adjusted for Acreage and Fam:tly Size Acreage and Family Sise Age Only Private Social Private Social Private Social Illiterate to 2 years 11£.5 13.8 15.6 15.0 n.a. n.a. Li terate to 2 years 10.3 9.9 12.5 12.1 n.a. n.a. No school to 2 year/1. 13.5 12.9 14.8 14.3 n.a n.a. I ..., 2 -6 years' 20.0 18.7 23.8 22.0 n.a. n.a. ~ 2 - 4 years n.a. n.a. n.a. n.a. 25.6 16.4 6 - 10 years 8.8 6.2 7.9 4.4 n.a. n.a. 8 - 9 years n.a. n.a. n.a. n.a. 23.6 16.3 fl Weighted average estimated using number samp1edin literate and illiterate categories. SOurCffl -Rur-al: Tables 5•7arrd 5.8 and 1963eo'st figures in text. Urban: Table 6.3, column one. l, Table 6.7: Kenya:- Rates of Return to Scoooling, Landholding Household Heads in Rural Central Province, 1966 Costs and 19630:64 Benefits, Corrected £orProbability of Being Landholder R u R A L U R BAN Difference in Education Income unadjusted for Income Adjusted for Income AdJusted for Acreage and Family Size Acreage and Family Size Age Only Private Social Private Social Private ~al Illiterate to 2 years 13.7 10.4 14.9 11.5 n.a. n.a. Idterate to 2 years 9.8 7.5 12.1 9.6 n.a. n.a. No school to 2 yearrll 12.4 9.7 14.2 11.0 n.a. n.a. ~ I I'\) -.1 2 - 6 years 18.9 13.6 22.3 15.4 n.a. n. a. 2 - 4 years n.a. n.a. n.a. n.a. .25.•6 16.4 6 - 10 years 7.6 4.6 6.3 2·3 n.a. n.a. 8 - 9 years n.a. n.a. n.a. n.a. 23.6 16.3 /l Same as in Table 6.6. Source: Rural: Tables 5! 7 and 5.8; costs from Tables 6.2, 3.15 and 3.16. Urban: Table 6.3, o:>.lumn one. () o - 128 - ,; wich theY' are baaed are rele'YUt ~ tor thoa. who work. Wh.n un.mploY'- Mnt· il"videspread, the rate. ot return . .t be adJu.ted downward to account tor the tact that total marginal productivitY' ot 8QJ cohort ot .chool l.av- ers i8 not retlected b.J' the v .... ot tho.e who work. ~J In this aection we dev.lop a serie. ot combined urban-rural .ocial rateD veilht.d tor emploY'- ment probabiliti.s in 1960-1966. 60. The rural rate. have alreadT been a,dJuatedtor the probabilitY' ot AtricU1 11&1. beiq .-p101'.d in the kiDd ot rural work in which edu- aD cation 1ields a po.itive, return. To coabine the.e rate. with the, urban rate., we muat weisht the' latter by the probabilitY' ot employment in that .ector. 61. A slobal. W8¥ to deal with this problem i. to correct the rate ot return (p) to a given a.QWlt ot .ducation ( i) b7 IIUltiplyins it~ bJi the. re- 8pective probability (e:t) that a siven cohort ot .chool leaven, (marginal ney), or all personll ldth thi.. educatiODal qqJ.itication (a""nee view). will tind. employment vh1ch'J1.lde them ditterential earningll cozre.ponding to their education'.. The: resulting rat. Pi yould thus, be: the .ubscript. (r) and (u) reterring to th. rur&laa.d urban aepentll ot , employment , respectively Y. 'l'bia adjU8tme~t, while an acceptable approxi- ...tion tor social aDd private rate. ot return in III&D7 cuell, 111111: tail to sive adequate results tor ~ aocial rate it only a 8mall traction; ot a g1ven cohort ot 8chool-leaver.' .ucce.dB in obtainiDg protitable emplOlMnt: The di8cOWlted co.t. ot .clueatillS the cohort ~ yell outweigh the: total dia- cO\lllt.d diff.rential HI'IliD8. ot tho.. who tind protitable uapla7Mnt t 80 t~at the 80cial rate otreturn becaae. n.lative. A .1m11ar ar....nt ap- plies. to the private rat•• , except that the rate. ot return are· I,caewhat lells aen8itive to chaase. in the. pro.pect. tor earning. and e~lo,rment becaWie 8chooling co.teare, partl1' .ub.ic1.i ••d bY' public tunela.. The marsinal rate ot return to inve•.tMnt 1n pr1.m.arJ .chooling -..v thue be _sative in Kenya, .ince, sa ve .how below, a hish proportion ot peraons DOW'! leaTing pr1J-.ry 8chool are UD~1IP10J'ed or under_pl07ed. 1:1 In tiM. ot he.....,. UDeJlp1071UtDt, ••tillat.4 v.... mq al80dftter trom the co.petitive equilibr1ua vage eYeD it theY' have b~en correct.d tor Ti.ible UDtmpl~nt'. Hovever, tOJ! pluming purpo.e. the rate. should not b. corre~ted tor th1.' bias it the c11.tortioD is expecte.d. to per.1at. Y '!'hiB reter. to profitable emplor-nt only, i.e. t employment in' yhich a .ore educated per.on, will receive a poaitive income differential in c_pari.on with 1... educated per.on.:.. Unemployment aDd non-protitable aplO1MDt are equival.nt in the ••nae. that they b.oth re8ult. in nesatiY. rates otretum. - 129 - 62. The probability ot obtaining urban employment is estimated below on an annual buis tor 1960-1966. Annex Table 6.1 indicates the average number ot years which ~..aed between a peraon's leaving the school and his taking up his tirstjob in our Labor Force Sur"y. This intormation is given tor the urbau Atrican aale employees in the survey, crosa-classified bY' level ot education and year ot leaving school. The 'ligures ahow that the waitiilg time is shorter the higher the educational. level, and that it has tended to decrease in recent years. '!I By i ts.lt, however, this obser- vation is meaningless since it gives no indicatioD ot the probability that a person leaving the educational s,.stem in, a"". 1965 at Form II level will tind an urb8l1 Job. To .awer this question, we malte two usumptions: tint, the Labor Force Survey result. .uat be representative tor all Africans in non-88ricultur&l. occupations; second, the enrollment ditterencea between consecutive years and grades ~t correspond to the numbera of achool-leavert trOll the various levels ot the educational ayatem (aee Table 3.2)Y. The results are given in Table 6.8 (tor incomplete and complete prtmar,y educa- tion only). 'J/ By dividing year-bY-Y'ear recruit_nt tigures tor persons with gi ven amounts ot education ( derived trom the Labor Force Survey data and Annex Table 6.1) by the corresponding numbera ot apparent entries from the school systell into the labor ~ket, we arri ve at estimatea ot employment probabilities for the two lowest education categories tor the period 1960-66. The private rates ot return to the 1960-66 cohorts ot African malea with incomplete and complete prtmar,y education weighted by these employment prob~­ bilities but not otherwise adJuated are given in Table 6.9 and the comparable social rates in Table 6.10. 63. But theae adjusted rates reflect returns to much more than achool- ing alone, &8 the socio-economic and other adJustments ot the urban rates have shown. If we estimate a correction tor socio-economic background on the cOlllposite unadjusted rates ot Tables 6.9 and 6.10, we can expect them to fall substantially. The estimation requires us to assume that the pro- portional ettect ot socio-econc.ic backgrOWld on the increase in income !I However, Boae ot the 1966 school-leavers who did find jobs were not yet recrui ted by the time of the Labor Force Surve;y_ Another observation to be taken into account il the decrease we observed in the Itandard deviations in the last years. Thia sUSg.sts that nowacla¥s yOUDg Keayul either 'lind an urban job soon atter learing Ichool or not at all. 2/ For the actual c&l.cu,ationa, Ministry ot Education statistics reterring to bo,ys' enrollments welre used. A tixed sum tor non-Atrican pr~ enroll- men,tl was deducted. ~Secondary leYel enrollments reterred to general sec- ondary schools only. \~ H01IeTer t allorinl tor &11 torma ot secondary edu- cation was considered\_ spurious reti __ nt in Yie", ot the very crude nature otthe estimate~. 'JJ The data tor perlonl with 10 and aore yeara ot education suggested a CQll- plete absorption ot the 1960-66 cohorts into urban employment. For per- Ions wi th 8-9 ;years ot education, the employment, probability dropped to approximately 90 percent in 1965 _d 15\perceDt in 1966. \ \\ Table 6.81 Kenya:-:- Estimated Numbers· of African. Male.. Pr.l.man; School' Iieavera . and Estimated Urban :&nplo:pent Probabilities 1960-1966 .' " 1960 1961 1962, 1963 1964 1965 1966 School: Leavers (thousands) stan~ds·J: -v· 52.4 26.0 49.5 Y 43.2 20.5 7.5 Standarda:.; VI - VII (VIII) li.7 23.5 28'.4 33.0 68.6 66.7 68.8 ~ ¥flop_to Probabi.lit~.a (percent) standards I - V Leavers 6 9 5 Y 4 2 Y Standards. VI - VII (VIII) 42 20 18 16 10 9 5 >:" t-' \.&,) 0 .. 11 Contained in 1962 estimate .. 2/ Contained in 1965 estimte. Sourcet,'; Calculated from enrollment statistics and Annex Table 6.1CB indicated in the text. -1 ! ,j Table 6.9: Kenya: Urban and Rural 'W3i hted and Combined PrivAte Rates of Return to Prima~ Education, 19 0 - 19 1960 1961 1962 1963 1964 1965 1966 Inco~Iete Primarl Education Urban Component!! 1.5 2.3 1.3 I.,;! 1.0 0.5 0.521 Rural ~mponenty 19.3 19.3 19.3 19.3 19.0 18.6 18.3 Total CO!!l!lete Primary Education 20.8 21.6 20.6 20.6 ~ 19.1 - 1B.8 Urban Conpmentll 23.2 11.0 9.9 8.8 5.5 5.0 2.8 Rural Component~l 15.8 15.8 15.8 15.8 15.3 14.8 14.3 Total 39.0 26.8 ~ 24.6 20.8 19.8 17.1 f-J w f-J Note: No adjustments made for socio-economic and ability factors. 11 Differential rate 2 - 4 from Annex Table 6.8 multiplied by probabilities in Table 6.8. Y Average of rates for 0 - 2 and 2 - 6 from Tables 6.6 and 6.7 corrected for acreage and family size. Rates between 1963 and 1966 interpolated linearly. J/ Ilif"ferantial rate 5 - 7 from .~ex Table 6.8 mult~plied by probabilities in Table 6.8. 4/ Average of rates for 2 - 6 and 6 - 10 from Tables 6~6· and 6.7 COITected for acreage and family - size. Ra lies between 1963 and 1966 interpolated linearly. 5/ 1963/62 and 1966/65 urban employment probabilities for persons with incomplete primary education - assumed to be equal. -Source: Annex Table 6.8 and Tables 6.6, 6.7 and 6.8 •. Table 6.10: Kenyar 1960 1961 1962 1963 1964 1965 1966 Incomp1~te Primar;y Educauon Urban Component!! 1.0 1.5 O.B O.~ 0.3 0.3 o.-P Rural Coi1q:>onentY 1B.2 1B.2 18.2 18.2 16.6 14.9 13.2 Total Coeelete Prima~ Education - 19.2 ~9.7 !.2.& 18._8 16.9 15.2 13.,. Uri>u.lComponent21 16.1 7.7 6.9 6.1 3.8 3.4 1.9 Rural Component!:JI 13.2 13.2 13.2 13.2 ll.7 10.2 8.8 II Total 29.3 20.9 ~ 19.3 15,.5 13.6 10.7 ..., w I\) ~Qte: No adjufJtments made .for socio-aconomc or ability factors. 11 Differential rate 2 - '" from Annex Table 6.9 multiplied by probabilities in Table 6.8. 2/ Average of rates for 0 - 2 and 2 - 6 from Table 6.6 and 6.7 corrected for acreage and family - si~e. Rates between 1963 and 1966 interpreted linearly. 1 llif£erential rate 5 - 7 from Azlnex Table 6.9 multiplied by probabilities in Table 6.8. hi Av~~age of r~te~ for 2 - 6 and 6 - 10 from Table a 6.6 ~d 6~7 cQrreGted for acreage and family " size. Rates between 1963 and 1966 interpreted linearly. 5/ 1963/62 and 1966/65 urban employment probabilities for persons with incomplete primary - education assumed to be equal. Sources: Annex Table 6.9 and Tables 6.6, 6.7 and 6.8. - 133 - usociated with more schooling is the same in rural areas as in urban, and that it has not changed betye~n 1960 and 1968. !I We find that the 1968 private rates adJusted tor aocio-ecoDomic background trom Table 6. 3. colUIDD (3) are 120 percent ot the private unadjusted rates in Table 6.3, colUIDD one for incomplete primary education (2-4 years of schooling), and 39 per- .. cent tor complete primary education (5-7 years ot schooling). The social rates in 1968 adJusted for socio-economic background are 100 and 47 percent ot unadJusted rates tor these two levels, respecti vely • ApplYing these percentages to the combined rates in Tables 6.9 and 6.10, We get an approxi- mation of the rates adjusted for both employment probabilities and socio-' economic background (Table 6.11). These adJusted rates more correctly re- flect the contribution ot investment in schooling alone as distinct from the combined contributions of iuvestment in schooling and extra-schooling qualities. 10. Final. Rates! Adjust_nts Accumulated 64. Tables 6.12 and 6.13 present the final social and private rates of return, showing each step of adjustment which we have made in sections 4 - 9 of this chapter. These rates are recalculated using corrected benefit data. The crude internal rates to urban Africans, derived by relating the net additional earninp:s associated with addition~l schooling to the cost of that schooling, are seen to vary considerably amop,g educational levels, but are generally high, ranging from 9 percent for university to 24 percent for secondary schooling. When the rates are correc'ted for taxes (private rates) and mortality (social rates), at least some are still acceptable investments. But the adjustment for the influence of socio-economic variables, including the correction for ability based on examination results, leads to a marked reduction in both the private and social rates of return. 65. At the primary school level, we then combine the unadjusted urban rates, weighted for theprobabili ty of urban employment, and the rural rates, adjusted tor the '-proqabili ty of being 8 small landholder, to yield over'all combined rates to investment. As we noted at the beginning of the chapter, the combined urban-rural rate is a suitable yardstick for social investment decisions only if the dlvidln~ line between; the two sectors is thought to be easily permeable. If this is not the case, the rural and urban rates should be considered separately from one another. Rural rates would appear to be more relevant for policy decisions, given the population weight of the countryside and its lag in educational development. ~ Since African primary Bchool students probably were much more hetero- geneous socially and economically in 1968 than in 1960, the eftect ot a correction tor socio-economic background probably would have been smaller in 1960. The rates in the earlier years ot the ••ries would thus be biued downward relative to the lat.er rates. - 134 - Table 6.11: Ken"ya& Combined Urban and Rural Rate's ot Return to PrimaEl Education, AdJusted for'Socio-Economic i"'1 Backp-ound l 12~ - i9~ II I' , \\ Inc9mplet'ePri~~ Edu,cation ~\,\ \\. 1960 1961 1962 1963 1964 1965 1966 'I '\ \' \ ,I Private 25.0 25.9 '24,.7 24.7 24,,0 22.9 22.6 Social 19.2 19.7 19.0 18.8 16.9 15.2 13.5 ( Complete Primary " Jiiucation Private 15.2 10.4 10.0 9.6 8.1 7.7 6.7 Social 14.1 10.2 9.1 6.4 7.2 5.0 Source: Tables 6.9 and 6.10 total t!gures,multiplied bY' percentages derived in te~. \J -( // , ' Tab1 e 6.12: Kenya.: Average PI" vate Ra:.es of ~turn to SchQol5.. ng, All Adjustments 07 Years of Schooling, 1968 Years -:.f _ • U!"hi.ll rates corrected for: -.. -LI Seho o~n§- Rural rates based on head of hshld. Combined urban-rural. inc., corrected for: corrected for: age, taxes, proo. of holding land, ~loyment employment age age and soc. -ec.var. acreage, and family probabilities probabilities o:-.ly taxes and exam score size only and soc.-ec. var. Prir:t 2- 5 - 7 26 55 26 31 1Jt£ 1~ l2~ 2 - 7 33 55 33 13 l8 22!l 17- 7- Seconda17 B- 9 24 2h 9 cLl± I 10 - II 5Z 40 30 t-' 8 - 1"1 Jt..- 32 19 ~ Ri~her 5e~ond~ l2 - 2.3 2L 23 23 liliversity ll- 17 27 20 20 ~k)'te: For eaC!1 set 0-£ adjustments, each rate is recalculated using adjusted costs and benefits. /l See footnote L!, Table 6.3. Sources: T2 1 - 2 years of schooling, 1966. ColUMn. 1: Table 6.3, column 1. /3 3 - 6 years of schooling, 1966. CoIUl11l 2: Table 6.3, column 2. 7IJ ? - 10 years of schooling, 1966. Column 3: Based on Table 6.3, column 3, and 75 Incomplete primary schooling, 1966. Table 6.4. 76 Complete primary schooling, 1966. Column 4: Table 6.7, column 3. ColUMn. 5: Table 6.9. Column 6: Table 6.11. Table 6.13: Kenya; Average Social rta,7.es of Return to Scmolins, All Adjustments. by Years of Schooling. 1968 Years of Schoglin 4, Rural rates based an head of hshld. inc., CoJlbined urba-rural Urban rates corrected for: corrected for: corrected fer I age, morta- age, mortality, lity, soc.-e!:. age, blrta- soc.-ec. var •., var., ability, age, mortality ..,loJllSlt ..,lo~t lity, soc.-ec. ahill ty, and KPE cert. and prob.of holding age age and and soc.-ac. probabi- probahili- var. and !FE cert. govrt a..,l. ].and, acreage, ~ Jl)rtality var. lities ti_ md exam score effect effects' and f!!l.l;r Sise onlY 8OC.-ec. ftr. Pri~ 2 - 16 15 11 ~7J ~ 15 ~ 5-7 15 15 38 38 17 10 14 14 L:..1 22 21 14 11 II. 11& 15ll SecandarZ 8 - 9 ~6 ~5 19 6 lO -11 3h 34 28 26 6 26 6 2l1! I t-' 8 - 11 ~ 21 24 21 18 18 ~ 12 Hi~her Se'Cal(]ar.y 1.2 - 13 15 15 15 15 15 15 Universi:5l iIi - 17 9 9 9 9 9 9 Note: For each set or adjus'"wnents, each rate is recalcula.ted using adjusted costs and benefits. See roo~nn~~ /1, Table 6.3. or J 1 - 2 year 5 schOoling, 1966. Sources: ColUl'ln 1: Table 6.3, colu.n 1. 3 - 6 years of schooling, 1966. ColUJWl 2: Table 6.3, column 2. 7 - 10 years of schooling, 1956. ColU!ll1 3: Based Oll. Tshle 6.3, colUlma 2 and 3. Applies to incollJ>late primary schooling, 1~. Colurm 4: Based on Table 6.3, oolu.ns2 and 3, and t'able 6.4. Applies to complete priat;-y schooling, 1966. Col'Wll'l 5: SamE>, ~s Co1uJIIl 4 J plus text OIl IPE certiticate effect. Co1wm 6: ~as CoIUll1 5; plus text on goye~t ~l.oJamt ettect. Column 7:' Table' 6.7, CoIUJa'l It. Co1UJ1'!1'l H: Taole 6.10. t - 137 - 66. We have shown that the unadjusted rates to investment in primary schooling, which haTe led the authors of aame earlier cost-benefit studies to &ssign high priority tb public investmsnt in primary education, are probably overestimated. The social rates to primary schooling in Kenya and seven other countries in Table 6.14, most of them unadjusted, Y are .,. relatively high, and most ot the rates to univeraity training are relati vely low ,compared to inveltlllent '~oth in other education levels and in other torms ot capital. The exceptions are Mexico ad Venezuela. which have high rates to investment in university training; these countries had sustained rapid growth over a long period proceding the date ot the estimates. 67 • In the light ot our experience, much ot what· haa been attributed to primar,· schooling il more likely the result ot non-8chooling inputs. and therefore the rate ot return to investment in primary Ichooling i8 probably much lower than the estimates indicate, when these other factors are constant. Furthermore" since in sane countries significant numbers ot primary school graduates remain in the rural sector or are unemployed, the rates based OD urban earnings data uncorrected tor either of these factors clearly overestimate the return I • Once the moat presling demands for medium.- lev'e1 personnel have been met, a aimilar situation might well develop at the secondary education level. Last but not least, the omission of temales trom the e.,timates of overall returns gives them a considerable upward bias. Y 68. Helpful as it may be to see the average rates of re;turn at a given point in time as we have presented them, the education planner is more, interested in what the rates are at the margin and how they will chSDge in the future. The steady decline in the combined urban-rural rates over the period 1960-1966 in Table 6~11 vividly showa that rapid changes in the returns to education have already occurred in KeDya. The essentially static exercise of Chapters V and VI is an indispensable point ot departure; Part Four of the study (Chapters VIII and IX) will uae 41tterent tools to extend cth~ &Dalysis into the tuture. I 1!jl.1!'.;'::)~? _ _ _ _ _ _ _ _ _ _ _ _ _ _. . . . . . . . . Jj Beme stUdies do corNct tor abil! ty or unemployment in estimating ratea ot return to schooling. M. Blaug's work contains a correction tor ability, but the rates are acljusted unitorml.y bY' the same percentase at' all schoolins levels, and the rate to inveatment in primar1 8chooliq remains relatively high compared to those to other levels. M. Selowstr's eattmates ot rates ot return tor Colombia do not adjust tor ability but do correct tor une.plor-ent &D~ labor torce participation ratee. Pos- liblY' becaua. the rates are estimated on17 tor Bolota, the uneaplO)'Jll8nt correction does not result in great chaD,.s ia the estimates. if The unadjuated rate or return to pr1uz7 schooling tor urban temales is 7.1 percent. compared to 21.7 tor VbUl African males ; it is 19.5 percent tor secon4ary" school, compared to 23.6 percent tor males. The emplo,y.ment probability tor urban African temales is ot the order of 0.2 (Table 4.1) aDd tor rur&l African teules. 0.05 (ott-tan a.plo;y- MDt; see Appendix E, Step 6). . Table 6.14: Social Internal Rates of iiBtum to Schoolini;for Males in Eight Developing CoWltries, Based on Urban Sal!>les: Kga, rthem Hileria, Uganda, India., Hexice" ChUe, Colombia and VeneBUe1a. (1) HoA~rn (3) J (4) I (5) Mexico Years of (6) ChUa!! Years of (7) Colollbia Years of (8) Venesue1a Yarse()f KenTa Migeria Uganda Years of India I Years of Scboolingl968 1964 1965 Schooling 1966 Schooling lS6) School.iDg 1959 Schooling 1965 Schooling 1957 --I- - - 2-h 16 1 - 5 20 2 - 4 17 -I-til 2Ia 1 - 3 35 1-6 82 5-1 38 t 17 66 6 - 8 17 5- 6 38 ~ 7 - ~ 29 h - 5 30 7 - 11 17 :8-9 16 9 -ll 16 9 - 11 14 \7 - 1iY 17 6 - 8 18. ~2 - 15 23 1:0-1.1. .34 t 5 22 12 - 191 13 I2 -13 12 . ~3 - 17 - 12 9 -11 29 1.2 - 13 3 18 12 - 1 # 11 14 - 16 30 12 - 1h 7 15 1$ -~6· ... I ~ 'l.h - 17 9 5 12 negatLTe Source•• ~ - Table 6.9, cOlU1ln (1). !I •.tes are ror .ales and f..:1.... Camo),,-cornctecl Northern Nigeria - SaJ!lUel Bones, Planning Edu&ation for rates for ad.1e (see -Bates ot Btrt.1lftl to ScbDol.1Dc EconOJlic Growth, Cambridge, Harvard thivers1ty in Latin Allerica,· p.~) are ~2 perc_t to Pria17 Press, 1969. schoo~ inst.d ot 2Ia perctllt and 12 p ..ClDt to .-nl ~ - John SIQ'th and Hi.coo1as Bamett, "Rates of Return secondary scb00111urtead ot ~1 peclllt. !be other on Invest.nt in Education: A Tool for Short-Term tllO rates were not corrected. Educational Planning, Illustrated with Uganda Data," in George Bareday and Joseph z.uver.J's, ibr1d Year Sf Average schoolin« • 5.S years. Book of lducation, 1961, London, 1968, pp. 299-323. India - M.maug, F.R.G. Layard, and H. 'Woodha1,l, The Causes J/ n~ecial" s8COIldar;r scbooling (average • 8.5 Tean). - - of Educated thetijloYJllent in Indla, manuscript to be published in 19 9, Table 9-1. W Gtmeral secoodary scbooling (average • n.s ;rears). Mexico. Chile, ~uela - for cOllplete reference, see Martin Carnoy, "Rates of Return to Schooling in latin AMerica," 21 First degree oVer _tricul.ation. Journal of Human Re~, Vol. II, No.3 (Summer, 1967), pp. 359-374. §! Jhgi.neering degree over matriculation. Colombia - Marcelo Selowsh.yJ The Effect of Unenployment and Growth on the Rate of Return. to Education: The Case of Coio,bla Harvard University, center for International _~fairs, 196A. - 139 - VII. THE DETERMINANTS OF EXAM PERFORMANCE 1. The Quality of Schoo). Output ... 1. .The previous two chapters focus on the intluence ot non-schooling factors on earnings and the calculation of average present rates of return to different levels of schooling, adjuated for these influences. Such rates give an idea at which levels within the schooling system investment gives the highest returns. In this chapter, we examine the influence of 1n- schooling factors on the quality ot school output, and calculate rates of return to schooling inputs. 2. The quali ty of output is estimated by students' exam performance: ~f average exam performance in one school is higher than in another, the quality of output of the first is considered to be higher than that of the second. It would be more accurate to :measure the quality of schooling by estimating the change in students' performance over the schooling period rather than the absolute level of exam perfor.mance at one point in time. Also, exam scores do not necessarily D1euure the training which students receive in discipline, consciousness of ttme, and hierarchical interactions. Because such training improves productivity and contributes to an increase in national income, it is very important in an industrializing society .• 1/ Our data does not permit U8 to take either ot these tactors into account in our estimates of schooling quality. 3. The concern with quality of schooling arises for two reasons. First, we would like to be able to project the quality ot the supply ot edu- cated labor through time in absolute terms on the basis ot assumptions about schooling inputs, and to use thi8 component in the supply projections of the Part Four, "The Future". Second, we want to explore the relationship between expenditure on schooling inputs and the quality ot schooling output in Ke~a. While it is commonly held that increasing expenditure per student on IchQ01- ing inputs, especially teacl1er expenditure per pupil, raises the quality of school output9 the veraci~i~ ot this relationship has never been assessed for developi~ countrie6,' .nd haa rarely been tested even tor industrialized countries. g/ F 4. The question of ~~e effect ot expenditure per pupil is important, not oply for the efficient al:location ot school budgets this year and next, but in the long-range plannin~\ ot .. school syatem. Recent studies have \\ \\ \\ ----------------------------~\r__ !.I See H. Gintis, "Production Fl.)pctions in the Economics of Education and the Characteristics ot Worker';~~~rroducti vi ty", Harvard Un1 verai ty, 1969 (mimeo. ) \ \ 2/ J. Burkhead t T. G• Fox, and J .'~ • Holland, !nput and Output in Larse- City High Schools, Education in \'\Large Citi~s Series, Syracust' Uni verai ty Press, Syracuse,· N. Y., 1967; TlJ)~' Riblf)h, Education and Poverty, Brookings Institution, 1968. - j~' 1 1 (/ (I - 140 - shown that a student'. pertor.aDce OD exam. i. highly correlated with his socio-ecoDcaic backgrOUDd; thus. hi. background JIUQ' be largely respoDsible for his .uc~ee. or failure. Y Furthermore • it i. probable that new students attracted duriD8 a period of rapid educational expansion in a devel.oping cOUl,lt17 cc:ae fro. Ie.. taTored socio-economic backgrounda than the pre- exp*n8ion .tudent.. The earlie.t .tudent. are likely to be the moat gifted children ad the children trOll familie. who are keen OD education. As schoola 'If increase their cat=-ent areu. the Ichool populatioD is likely to include growing nuabere ot pupil. who are either 1 ••• girted or come tra. families le•• intere.ted in educatiOD. 5. ThlUl, in the .hort run, a rapid expa8iOll of enrollments is likely to lead to a decreue in'the a"rase ab.olute quality of output .. .eaaured by ex. . pertoraaDce. ODly' in the lODg run vil1 the gradual increue in the averase educatiOD of the: population and the concc.itent .ocio-ec~OJDic change reTer•• ''i'4is development. The tiM-path of quality of .choo1 Qutput can be illustrated b,y the follawins figure. I I Q' ----- - - - - - -l.---- - - - - - - - - t I' I t t n Fisure 1.1: TiM-Path ot School Output Qual! ty as a FUnction of Exa.d nat ion Pertor.mance 1/ J. Coleaan et. al., !4ucatiCll&1. Op:port1Uli~, u. S. Ottice ot EclucatiOll. WuhiDgtOD ,D. C. 1967; Thor. ten Huen, Talent, Opportun1t;y aDd Career: a 26-Year Follow-up", School Ren_, 76, (2). June 1968, pp. 190-209; D. Wolte. in S. E. Barris ed., Bisher B4ucation in the Un! ted Stat•• , The EcODc.ic Problema. /\C"ridge,. Mus.. 3 . 960, pp. 11a:119. - 141 - 6. The quality ,of output (Qt ), vhi ch is believed to be adequately typified by examination performance (Pt), 1s uSUlled to follow aU-shaped course over time. At point tn' the temporary negative eftect of the rapid expansion ot enrollments ia ottset by the positive etfect of increued parental ed~cation.!I Iil the figure, the value added ~y schooling (Q It) is assumed constant. This need not be the case; if the quality of teaching . graduates, the most import8llt determinant of' s'chool output, follows & simi- lar path as the quality of &ll graduates (Qt ), then the latter will in turn be influenced by the change in the tormer t and Qt will tollow the same para- bolic path as Qt (Pt ) • 7. An adult. literacy capaign, the widespread introduction of tele- vision, or rapid migration into the cities might shift the curye upward. It is more expeditious and relevant to influence Qt (Pt ), i. e. t to improve exam results, by working with schooling inputs. Including these inputs &8 variables, we express the curve ot Figure 7.1 in equation (7.1). S (-;- L k P t (~-L 0 (7.1) Pt = a + b2:Xi + C + d --------- + e 1-=1 S 0 0 P t - j 0 where P • exam performance; t t • time in years; j • interval at which the expansion ot enrollment. is . . .ured; if Xi • the schooling inputs; !I More elaborately, Pt could be expressed as a tunction ot t sq, the rates of' increase in the percentqe of the school age population enrolled, and in the average education level of the parental populs.tion. We might expect the age-specific enrollmen~. ratio to be positively correlated with the average parental education, but the change 1n the ratio would tend to be negatively correlated. Y The choice of this interYal is largely arbitrary. For data convenience t it could b~ geared to the frequency with which other locio-economic indicators (such as census results become available. It co~d also be linked to the length ot educational cycles. In the present study!) the interval 1962-1967 w .. chosen. - 142 - s • enrollment .leve1 in the area -~- under observation; H -~- • averase education, or stock of schoo1iDg, ot the population: per capita in the same area (meas.ured in thia study by average male education) ; Uld a.,b,c,d,e • parameters. 8,. Our intent in this chapter i8 to e.tima.te exam per1'ormaDce as a :fUnction ot the above variables, vi thcro•• -aection data. Because. information on enrollment rates over ttme aDd on socia-economic background of' pupils i8 un- aTai1able, we have to rtt2z:esent sIP aDd H/P by proxies which ~ ayai1ab1e only at· the county level. The pro~ tor change in secondary eDEollment rates i8' the share ot secondary achool students in the total cO,UDty popu- lation in 1967. 11 The, pro~ tor the stock ot .choolill8 18 the: ayerage years 0,'1; education ot males IIIld temales oyer 30 (uaed .. separate var,i,ab1es) in 1962 in the county where, the 8econdary school ia located. Y Since the primary" 8choo1 examination re8ults .and schooling input data could: be, col- lected, only for two counti,es, the use ot the proxy variables :1:. re.•tricted to the analysis.' or 8,econdary examination re8ults (CS,C and HSC). However g sinqe no systematic or' aigniticUlt results emerged, for all intent. and p~oses our analysis is, limited to the tir.t two terms ot equation (7.1), tliat i" the ef1'ect on exam re8ult. ot various schooling input8i" and rates ot return, to increasing them, largely ~corrected tor enrollment', ratios and tile' sC)cio-economic composition ot the .tudent population. 9. Although the actual analY8is is thus inevitably a secpnd\-best com- p~i..e, J/c: it neverthele8s yields a good . .y worthwhile resuJ:ta:" some ot whieh ar_ in contradiction' to widely accepted premiaes ot educau±onal. policy. For example, the analysi8~ make. clear that simply, increasing expenditure per p~pil cannot be relied on' to raise ,exam pertormance, and that the impact ot the different expenditure cClllponents, (teachers' salaries, boarding tacili ties, etc.) must be examined.. While we are far from suggesting that concrete action 1/ This implies that the' nUllber, ot seco~.dary school students in',1962 (the year t o-5) was negligibly .ma¥, not ';m unreasonable assWflP:tion for JD08,t counties, Siven· tb,e ~cr.t eight~Jold increase in the total number ot Atrica second&17 pupi181n that pl!riod. It also as8ume.; similar age structure. in diUerent counties. 2/ We ........ that a .chool' s catchment area· is by IIIld large restricted to it. county-, a p11.U1ible aimp1iticati.OD, it one ignores Kenya~! 'It dozen or so extra-provincial .,choo1s "bich recr.ut:t country-wide. J! The most serious effect ot. omitting socia-economic, background: as a vari- able is the biu: it: caWie. in the e.t1JDat.es ot the coefficient b. - 143 - be taken on such a narrow factual basis, we feel strongly about the necessity of a thorough educational procels analysis preceding any major reallocation of resourc.~est whether from other public aCtivities to education, trom one edu- cational s'ubsystem to another, or wi thin a subsystem between groups of inputs. Also, while~ we are unable to estimate the coetficients c and d in the above equation, &'ld, therefore cannot project a qual!ty component 'into the supply projections of Chapter VIII, we believe the model to be correct and verifiable given more detailed data. 2. Primary Educ:ation a. The K~n,ya Preliminary Exam (KPE) 10. The KP'.E was introduced in 1961, superseding the eX81D8 that had formerly concludt~d the Primary and Intermediate courses in African, Asian and European schools. A centralized exam designed, supervised and evaluated by the Ministry 01" Education , it consists ot a battery of 240 quest:t.ons in three subject groups 1/, 90 of which form an English language 'test, whereas 50 refer to mathematics and 100 to general subJects. The maximum score in each subject group is 100 points, giving a theoretical total range of 0-300 points. 2/ This score has replaced the tail-pus-qualify classification used until 1966 (only a qualif,y could makE a student acceptable to aecondar,r school) • 11. Not only is the KPE the leaving exam atter seven years of primary schooling, but performance in it constitutes the main, if not the only, cri- terion for admission to secondary education in maintained and assisted schools. Only in a few marginal cases do secondary schools resort to addi- tional. inquiries (personal interviews, information trom the former headmaster, etc.) to assess an applicant's aptitude. In the last few years the KPE t S role as a selection screen has gained an ever-increasing importance, as sec- ondary education has become a virtual pre-condition tor urban wage employ- ment. Due to the educational system's lack ot flexibility and its~ vertical orientation, the KPE is a once-and-for-all deciaion that entails a good deal of indi vidual hardship, no provision being ~e tor late academic de.velop- ment, stronglY biased ability profiles and the like. 12 • Since perm~$sion to repeat Standard VII and KPE depends on the availability ot school places and ia certa1n~ not encouraged, unsuccessful JJ The following description r~terl to the 1967 exam (see: Ministry of Education, Kenya Pr.lim1n~r Examination 1967 II English, Mathematics and General Papers). g/ These extreme scores do not t)ccur. Since all but five composition tests are multiple choice questions, involving tour alternatives, a completely random anlwering and a failure in the composition text would already yield on average a total of 71 points. On the other hand , it is, obvious that the theoretical maximum is not attained either, given the relative dit- ficultY"ot the exam. (In 1967 the best pertormances were in the neigh- borhood ot 270 points). - 144 candidates often display considerable determination and ingenuity to circum- vent the official policy, a popular device being to change one 'a, name and. repeat Standard VII in a more remote school. A recent study has shown that out ot a sample ot 834 pupils who sat KPE in 1964 in Central Nyanza, Plnbu, Kericho and Kitui districts, 68, or more than 8 percent, were still repeating in 1967, the corresponding 1966 ad 1965 'ligures being 37 and 16· percent, respectively. More than h'~" of the 1965 repeaters had obtained, a. KPE pass in the previous year and obvlc~1y stayed. in school in the hope ot improving their Bcore to the point Wh~~l)~:y 'they would be acceptable to a secondary school. Similar results were obtained ln a sample, ot 203 KPE candidates (1964) in four schools in the Tetu location o~ Nyeri district. There is evidence tha~ repeatera had a better chance of being &Qcepted by secondary school& than first-timt! candidates. 1/ b. The Data and their Limitationa 13. The 1967 KPE register provides a unique source ot da'.,; giving in- formation tor ~very individual candidate OD the following characteris,tics: sex, age, numb'er ot previous exam attempts, total KPE score, SCOEe. in each of the three subjects groups" previous ranking in the class, 8Q.d. the candi- date's tirst, second, ad third choice of a secondary school in. case he or she would quality. However, the register does not give inf'orm-.tion on the socio-economic background ot candidates. 14. Once the KPE SC9re ia accepted .. a reliable indicator of' scholastic achievement, the question arises ot how the score is intluenced, by certa~n educational policy measures, if all extra-educational factors ..,e. kept equal. Ideal.ly, this question should be approached on the basia of completely diaag- gregated data, i.e. at the. level of the individual pupil. The 1967 KPE re- sult. does contain incH.viduaJ.s' re.ults by .ubject group, but time and persOD- nei constraints did not~ permit compilation and analysis ot dat., in this d~gree ot detail. Furthemora, information retle.cting environmental in:eluences iron the individual' pupil 8I,ld other extra-educational data are availllL'ble at the county level. at be,·t (and evep ~there mo.t~ in the form ot rough estimates), so that matching these two kinds ot data would have been unfeuible, parti- cularly since data are t,.en from school. in onlY two counties ,. 15. For theae re..ons the intor.mation is aggrega~ed and, aDalyzed at the level of the individual school. Only, the tir.t four data items are used: percent ot girls among a school's cmJidates, average age, perc.ent ot re- peaters, and average total KPE score. The data buis is a 20 percent random sample ot all primary .chool. in Muranga aDd Meru Counties that presented KPE candidates in 1967, totalling 89 school. with 3,405 candid..tes, or roughly !I L. Brownstein, Prelimin&!'l Results, ot a Surny of 1964 KPE'· Candidates in &abu, Kitui. Kericho and Nyanz., Institute tor De,velopment: Studies, University College" Nairobi, 1967, mimeo. J. Anderson, "The Adolescent in the Rural,. C01IIJIll.1D1ty", in: Education,. EDploYJD!nt 8Zld Rural. Development, ReHn ,ot the. Kericho (KeDYal Conference" 25th September to 1st .October ~, Nairobi 1967, pp. 420- 21. . - 145 - 2Js percent ot the national total. Although these two counties t situateCi in the Central aDd Eastern Provinces north of Nairobi, have a a_ber ot Bocio- economic teatures in common. there are sufticient diiterencea between the. to . .e them adequately repre s ent at i ve of IUUl1' or KellYQ f i rUral are... ' 16. This information is .upplemented by school data contfdned in the budget. of the tvo county councils, including enrollment. in each standard. nl8lber of c1...es, number ot teacher. in each category, and total teach.rli' salaries, &ll at the level ot the individual. school. 17 • On the whole the data appear to be fairly reliable, 'with the exception ot the share ot repeaters and the ave rase 88e of candiuates. Y In the cue ot Meru COWlty, in particular, the understatements on repeater• .ust be ot maJor proportions (only 11 out of 41 .choo1s reported ~ re- peaters at all, against 46 out ot 4:8 in MuraDga.) Y No allmtance can be aade tor this biu, which might explain why the percentqe ot repeaters ia not a signiticant explanator ot a school'. average ex... perf'orma.uc6. On the other had, average age ot candidates is a aipiticant var1ab1e. c. Resreesion J1..naJ..ysis Beau!ts 18. Initially, we designate teachers' salaries per pupil (Et/S) .. the sole determinant ot averqe KPE Icores. (T-nluea ot the coeftic1at. are in parenthese•• ) (7.2) KPE' • 150.312 + 0.152 EtIS (0.187) Both the t-val.ue and, F-value are in.ignificant, aDd expenditure per pupil appeara to haTe a negligible fmpact on averase KPE acor••• 19. In an ettort to obtain better resulta, we diaacgregate the teacher.' .alarie. variable into its two m&jor components, a"fer... teacher .alar1e. (!!tIT) and teacher-pupil ratio (TlS). The latter does Dot . e . to haft aD. eftect 'J/, but the tormer doea. .. Y The data on the averase 88e ot candidate. appear to be le.a subject to miareportill8,' aD impre.sion which i8 borne out by the Tirtll&l. ab.eDce ot correlation between the share ot repeater. ad the _...r _ ..e ot candidates (R2 • 0.0002). , Y In a,.10 percent random s_ple ot all prilu.r;y achoo1. conducted b7 the Min1atJ)' ot Education in 1967, this tendency .howed even more atroDSl:r. The per\~ent~e ot ,repeater. in Stadud VII ".. 14.1 percent tor the whole c'~t7, 1.7 percent tor Meru and 14.5 percent tor MUraDl_ (3.7 per- cent and 16.6 percent, respectively, in our .-.ple). "JI InclWliOD ot TIs .. 1m independent variable ca.ed the '-val. ot the equatioa to becOM ina1gniticUlt. Regre •• ing -partial teacher-pupil ratios tor the ditterent grade. ot teacher. (P1 t P2, P3 t pa. Uld UDqual.l- tied teachers) on KPE score. resulted in negative aDd i ..ipiticaDt n- gr••• ioa coetfici.nt. tor the P2, P3 aD4 \lDqual~ti.d teachen. t' - 146 - 2 ', 0.06 KPE • 136.146 + 0.079 (Et/T) R ":' (2.35J In leeking to under.tand the reuon wb1' teacher .&lari.. ...11 to a:tt.ect eXQltl score, we ex8m1ne the three tactors which intluence the av.erage teacher salary: perf9rDlSJlce-related bonuses. tbe aftrage formal qualification ot the teaching staft 5 and it. ..-.erase aeniority. In Kenya. there ue no bonus incentives, except tor headma.ters' adm1D1.trative allowances which are not related to teachingpertormance. Formal qualitication unc10ubtedly bas a strong influence on the Bchool aal.ary' bill. The starting 8&1&1'7 for the moet highly quaJ.1t1ed teachers (Pl) i. mc)re the tour times that or an un- qualified teacher without KPE. However. the compoaition or the' teaching statt does not turn out to be a good explanator of KPE score. In DO case does the sh~ of a specific cateS017 of teachera in a schcol '. teaching force have a 'signiticant t-value. nor significant; .' . w..the F-v&1.ue tor the whole· equation 20. The aeniority ot the te.biD8 .taft eM Dot be . ._ _ 4 tirectly from the avd.~able dat'a. Howe'Yer, we are able to aat1etactoriq correct the .choel aaJ.ary bill tor seniority. W. calculate two ~ical sal&l'1 bills, Et* and !:tit... To tind Et*·, we" aultip17 the nUllber ot teachers in each cat'egot-y by' the correspondins initial salaries aDd a.the products tor each school.·~ y Et* 1a derived in the .~ wq, except that it i8 based OD the average ~alary tor each categGr,J Observed in the two counties ~or which we have data.. We then formulate two eq~1;ion8 anal8fl;oua ~o" (7 .3) • 2 : (7.4) KPE m 14_.152 + Q.039 It·/T R • 0.006 • /J (0, •. 56 ) (7'.5 ) KPE • 144.432 + 0.045 I*··/T (0, •.48) ~ • 1 21. A !compariaoo ot (T. 3) aDd (7.4) ahon the tull tilpact of aenior:1 ty on exam reeU:J.ts, whereas a ccmpar1lOD' ot (7.3) aDd (T.5) brinp out the effect of' ttle deviations troll aYeZ'". ..Diority • In both c ...s. the in- clWlion of seniority makes tbe ditt.nace between obtainias inaigniticant and significant t- and F-valuea at the 0.05 left1, thOUSlJ, the absolute ai se ot the f'raet1on of variance removed ia rather ....u.. Thu, aw.or1ty aeems to be a better predictor of exam pertormance than is theto~ qualification of the teach1DS statt. . 22~ However, one should not 3uap to the conclusion that the effect ot seniority is 801ely due to the accu.ulat.4 teacbins experience ot senior teaChers. The pos! ti ve association ot ICPE reaults aDd ••Q~orl ty 118¥ retlect 1/ Since .·eniority increuea are abCNt the a _ in relative teru tor &11 categoriea (exQept tor unqualified. teachers, whoae aalaries rema:l.n con.taut). this procedure imp11e. that increases in ~.achiDg experience are pa.it1vely' related to the la_lot protea,ional qualif1cation, which doe. not aeem unrealistic. - 147 - the beneficial effects of having a stable staff, since the older, more senior teachers are less likely than younger ones to be trsnsfered to other. achoolii; ~ regardless of qualifica.tions. Furthermore, the deep-rooted respect for older persons in African societies mq add to the classroom effectiveness of the more senior teachers, particularly in rural areas where traditional values have been conserved. Finally, the senior teacher has probably come to terms wi th the prospect of sp_ending his life in a rural environment, whereas the novice, particularly if he ie more qualified, ~ experience a .tee.ting of futility which is bound to affect his work. 23. The next equation includes Et/T and three additional Y8.rlables related to the si,Lze end composition of the school's student body~ 1'.J.um.'bei." of' pupils (S), share of KPE candida.tes (K/S), and average age of IG?E cSJH11dB.'tes (A). . (7.6 ) KPE = 212.325 + 0.090 Et/T - 0.048 s - 59n250 K/S - 3.943 A 2 (2.64) (2.82) (1.67) (1~a7) R = 0.173 These results ere difficult to interpret. The t-values for the first two regression coefficients are significant at the 0.01 level, and those for the last two at the 0.10 level. The combined explanatory power ot the four vari- ables is only moderate -- roughly 17 percent of the variation in exam, score -- which is hardly surprising given that we have no data on many educational variables (teacher turnover, commuting distances, equipment standards, etc.) nor on the still larger number of socio-economic factors. 24. The negative correlation of size of school and KPE results seems to fit into our description of the educational growth process in the beginning of this chapter. 'The first pupils in a new school are likely to be the most gifted children and the children from families who are keen on education. Y As the school increases its catchment area both in width (by attracting pupils from more remote homes) and in depth (by enrolling more pupils wi thin the school district), not only does the size of the school increase. The school population is also likely to include growing numbers of pupils who are either less gifted or who have longer commuting distances, families less interested in education, or more limited financial resources. Conaequently, the ave~~e exam scores are likely to decrease as school size increases. "I However~\ i;ncreasing total inputs into the teaching prq~ess and/or impl'ove- ment in the input. mix JIl8¥ bring abOllt an increase in e~~i,am .cores. 25. There are two possible explanations for the negative correlation ot the share ot candidateei; in the school's .tudent body (K/S) aDd the KPE ) 1/ One might object that the most important factor in a family's decision to send th~ children to school is ability to pSJ the school tees. How- ever, since family income levels are closely related to the education of the 'main income earner, the families whose head earns more and is I better able to P8\Y school fees are also the families whose head is better educated and more likely to be keen on educating his children. r ~Ic a ) ~. - 148 - results. First, in a school which screens its pupils severely during early standards, the surviving KPE c~~-&i.dates are likely to perform better' than those in a school with casufl~t:,screening. Second, a low K/S could indicate that the candidates are those; more favored, t,~wer childrl.en who started school prior to a recent rapid enrollment expansion. The average age of candidates i. also negatively correlated with average KPE score, since older students are'more likely to have repeated a standard, to be, KPE. repeaters, \1 or to have temporarily dropped out of school. 26. As' we mentioned when discussing the data, the percentage of re- peaters is not a significant explanator of ex~ score • Neither is "the per- centage of girls among candidates , ,possibly as' a result of the conflict during a girl.'s education between'negative attitudes on the' part of society and positive ability and motivationoD the. part of the indivi,dual. 27 • In 's~, we hav'~ regressed eight variables on ,Jove a pass, for example, both above- and below-average c,mdidates would be credited with a salary gain of Ksh 0.67 even though the bellJw-average candidates 'would actually re.alize a gain of Kah 2.16. Without knowing the actual shape of the distribution, it is im- possible to assess 'the dimension of this error. 34. Fourth, o;ur treatment of the differential earnings 88, perpetuities implies some overstatement of the rates of 'return. Fifth t anaddi tional up- ward bias results from taking the average salaries for the thr.ee categories regardless of ,age , given the fairly low average age observeq. in the sample - 27 .3, 28.2 and 28.6 for the qualify, pus and fail categori,s • respectively - and the fact that salary curves tend to flatten out with incre ...ing age. Sixth, the semple is assumed to be repres,entative of rural Kenya though it . is relatively small. As has been pointed out in paragraph 15, Meru and . Muranga are distinct socio-economic entities. Since a dUJIIIIIY var.iable iden- ti~ing the county proved to be insignificant in one of the' subsidiary i variants of the analysis, the validi tyot the results may not be as restricted geographically as it would appeer at first sight. " 35. The most serious objection is that urban wqe differentials have been used to represent benefits, though we showed in Chapter VI that a weighted aver~e of wage differentials in available forms ot profitable em- ployment i~amore appropriate measure of benefits. If we were to make this correction, which illcludes 81.1 allowance for unemployment, t)le rates of return 'Would be reduced to a negligible figure. Thus, while the markedly low rat~ of return to increasing teachers' salaries once a school has reached an average score of pass or better suggests that salaries should b.e. raised in i3chools with below pass scores, 8uch "a, meaaure could tail to produce economic effects beC8UQe these schools ~ be in areas where wage employment outside agriculture i~ rare. 36. The .tact that most of our simplifying 88sumptiona tend to give ail" upward bias to these rate$ ot return cuts· an unfavorable l:t,ght on the rate~:, which are already very low. Though a more thorough and extensive inquiry , - 151 - would be necessar,y to consolidate these results, it would seem that in- creasing average teacher salaries would be a rather uneconomic means for improving students' performance on the KPE. 31. Recalling our analysis in the previous section of the components of teacher salaries per pupil, it seems doubtful that it would be economi- cally profitable to upgrade teachers on a large scale even if some cos·t in- creases could be absorbed by an increase in the pupil-teacher ratio, which appears to have little effect on KPE Beore. In that analysis, th.e compo- sition of a school's teaching statf by formal qualification explained little ot variation in KPE scores, while seniority was a more potent explanator. In the absence of data on performange-related salary components &,d on geo- graphic differentials in salaries, it is altogether uncertain whether ef- ficiency-tied or autonomous salary increases would lead to better KPE scores and higher quality of schooling output. 3. Secondary Education a. The CSC and HaC Exams 38. As we explained in Chapter III, the Cambridge School Certificate exam (CSC) is administered at the end of Form IV, after four years of sec- ondary schooling, under the auspices ot the Uni versi ty of Cambridge Local Examinations Syndicate. Only a small proportion of those entering the CSC exam continue their education in Form V or in teacher training colleges (Sl courses). The CSC is thus the main entrance requirement for most of the ~dium-level clerical and administrative occupations. The number of schools which present candidates for this exam has increased sharply, in line with the general expansion of the secondary school system, from 42 in 1956, onlY 13 of which were African schools, to 144 in 1966. Those schools which &iso have an upper secondary cycle (Forms V and VI) leading to the Higher School Certificate exam (HSC) constitute a nucleus of first-rate schools whose pupils perform consistently better than the average, perhaps because many of them are "extra-provincial" schools, which recruit their pupils from all over the country. b. Data and Assumptions 39. The data on the first four years (Forms) of secondary education leading to the Cambridge School Certificate examination (CSC) cover 133 schools, of which 115 reported the composition of the teaching statf and expenditure on non-teacher inputs in 1966. 11 These 115 schools form the basis of the regression analysis. The available statistics include indi vi- dual school data, reported by the schools, on: (i ) the number of students taking the CSC exam and their breakdown by results (Division I, Division II, Division III, GCE (0), Fail); '!/ Source: Ministry of Education. - 152 - (ii) the number ot teachers by grade of qualification; ( iii) non-teacher current expencli ture ; and (i v) the number of students in the school by sex and form of attendance (boarders/non-boarders). There are also data on: (v) the average education of adult male. and females in the county where the school is located (associated with each school for the purposes of the analysis) 1/. In the case ot entirely African schools, the average education is estimated tor the African population in the county. In the case ot mixed schools or pre- dominantly Asian or ~Jropean schools, the average education ot Asians and Europeans in the whole country trom the 1962 census is combined with the average edu- cation ot Africans in the county on the basis ot the percentage of Asians t Europeans and Africans in 'the school. 2/ (vi') a d\UDII'JY variable indicating the presence or absence ot a Higher School Certificate cycle (Forms V and VI) at the school; (vii) the number of CSC students per one th()usand population in the coun'ty whe,re the school is located. This figure is estimated by summing all pupils in Forms I - IV in the 133 esc cycle schools by county and then dividing the sums by the total population i~ each county. In both (v) and (vii), the administrative division existing at the time of tQe 1962 census has been folloved. 40. In order to be able to use the data to estimate relationships, the various esc exam scores have to be quantified. It is assumed that failure Qf the esc equals 1.0, Di vi~iOD I pass equals 3.5 , Division II pass !.I SO\1rce: Ministry of Economic Planning and Development, Statistics Division: Ke!8 Population Census, 1962; Vol. Ill; ot Atric&;U Papulation t Nairobi, 19 • . ' , 2/ Source: Ministry ot Education; this intormation was available tor Forms V and VI only ; where necessary:, adjustments tor Form IV enrollments have been made on the basis of the 1967 indi y,idual CSC results as pub~ished in the East Afri~an Standard, February l3"!"14, 1968. - 153 - equals 1.75 , Divis ion III equals 1.25, and GeE equals 1.5. 1/ The number of candidates by school achieving each level is multiplied by these indices~ The sum of these products is divided by the total number of candidates to give an average score for each school. The mean score for all school is 1.82, or an average Division II result. 41. Data on teacher expenditure do not cover all the schools, hence estimates are made on the basis of the number of teachers at each level of qualification and the average salary for tha.t level. gj Overseas tSl'1.tl teachers 3/ are assumed to cost Kt1740 in addition to the averagl.~ I)d.Y of a. Kenya citizen or local term teacher. '!if Annex l!'able 7.'1 shows the salary ranges for~he teaching force in Kenya,. 42. Since many schools ha.ve both esc and HSC cycles ...,but their tea.chers are not identified by cycle taugh'c, some sort of a:.),location has to be made. On the basis of information given by the Ministry Glf Education, it is assumed that only graduate teachers tea.ch in the HSe cycle i~(Forms 5 and 6) and about; 1.5 times as many gradua.te teachers per student ar~ emPI~yed in teaching HSC as esc students. rrhe number of graduate teachers in the' esc cycle can there ... fore be estimated as follows: Total Graduate Teachers Graduate Teachers in Forms I through IV = ~~e number of graduate teachers in the esc c,ycle estimated in this w~ is added to the total number of teachers at lower levels "'of ........ qualification re- ported by the schools. 1/ These are the reciprocal values of the midpoints of the theoretical ranges ot the various esc scores (Division I, Division II, etc.), multi- plied by 100 and divided by the value obtained for the Fail category; the GeE coefficient ot 1.5 has been fixed arbitrarily. Later adjustments allowing for an uneven distribution of exam performance within each of the ranges on the basis of ~pe 1966 esc resUlts in Uganda did not produce with the exception of Divisi~n I Significantly di1ferent weights (1.0, 1.24, 1.65, and 2.56 for failure, Division III, Division II, and Division I passes , respectively). 2/ Since most local teachers are of below average seniority, this estimate probably has an upward bias. 1/ Defined as teachers who have overseas leave and passage privileges, re- gardless of the scheme under which they are provided. '!l/ Es'timated Oil the basis of data published in United Kingdom, Ministry of Overseas Development, Statistics DiVision, British Government Economic AidA Statistics of Aid to Developing Countries 1962 to June 1961, London, 196 • - 154 - l 43. In 8.ddition to overall teacher expenditure per pup,il, a measure of graduate teacher input is used aa a variable in the regression equations. The graduate equivalent number ot teachers per pupil is estimated by veight- i08 each teacber reported by the ratio of hi •• alary (88 approximated by the average salary paid. to his level ot qualific ..tion) to the aver8@e salary of a graduate teacher. The weighting scheme assumes that relative salaries re- flect the relative efficiency of teachers 8Ild that the graduate,equivalent teacher/pupil ratio is a much more meaningful figure than t~e simple teacherl pupil ratio. c. Regression AnalYsis Results esc All Schools 44. The ~~othesis tested tor the esc cycle is that exam pertormance is. positively·related to total expenditure per student, non-teacher expendi- ture per student, and teacher expenditure per student, holdingoth~r.·, vari- ables, such as school size, constant. In addition, we assume that exam per- formance is positively related to averac. educatioD ot' adult males and females in the county (these are proxies for the propenaity of families to have their children .educated) and tb.at performance is negati 'Yely related to esc student concentration in the county (i.e. the percentage ot the 15-19 year age group enrolled in secondary schools having Form IV), bolding adult male and female education constant. This meaDS that we expect average exam per- formance on a standard exam to rise as, parents' education rises because of non-schooling inputs in the heme., and we expect it to tall ~ the nUDiber of students attending secondary school increase. .. a percentage of the total population (this 88sumes a constant age distribution fram cqunty to county and a strictly performance-oriented selection ot secondary school pupils). 45. The croBs-section regre.sion results estimating these relation- ships are the following: Y 2 . esc =1.6788 + 0.0016 (E/S) R • 0.039 (2.134) (7.9) log esc • 0.0669 + 0.1012 log S + 0.0947 B (3.410) (2.713) + 0.0114 log (E/s) (0.335) (7.10) log CSC ~ 0.0183 + 0.0811 log S + 0.1117 B + 0.0081 log (E/S) (2.154) (2.787) (0.236) + 0.0088 HIP R2 • 0.151 (0.866) [,I 1/ Figures in parenthes 1 s refer to t-values. - 155 - (7.11) log CSC = 0.2747 + 000584 log S + 0.0902 B - 0.0114 log (E/S) (1.445) (2.162) (0.318) 2 + 0.0093 HIP + 000792 C - 0.0008 (sIp) R = 0.178 (O.8l6) (1.824) (0.556) (7.12) log CSC = 0.1501 + 0.0686 log S + 0.0785 B + 0.0776 c (2.028) (2.393) (1.920) R2 =: 0.172 where esc = average ex~performance in points for each school E/S = total expenditure per esc student in ~, net of' boarding expenditure; S = number of students in Forma I - IV; B = percent boarders in Forms I - IV; HIP = average male education in years in the·district where the school is located; C = dummy variable for whether or not the school has an HSC cycle; . sIP = number of esc pupils per thousand inhabitants in the district where the school is located; log = natural logarithm. Other variables, such as the percentage of female students .in each school, the percentage of overseas term teachers, non-teacher expend! ture per pupil, teacher expenditure per pupil, and the percent of total expe!ldi ture on teachers were also run but were either highly correlated with the variables shown or contributed very- little to the regression estimate. 46. The results show tha.t if expenditure per student is the only vari- able used to explain exam performance, it is significantly related to per- formance, but covers only a small part (4 percent) ot the variation in exam score. The linear relationship (equation (7.8» also indicates that expendi- ture per student would have to be increased by more than KE6 per pupil to affect exam score by one percentage point. Since the average score among all schools is 1.82, or a slightly above-average Diviaion II pass, it would be possible to drop expenditure to about ~36 per pupil,. or approximately 40 per- • cent of the av~rage in all esc school (KZ87/pupil) and still achieve an aver- age Division II pass (1.75). J.f.7.. As we move down to equation (7.9), however, it is clear that expen- ditUre per pupil is not a significant variable when school size and the percentage of boarders ·1n the school are held constant. If expenditure per pupil (net of boarding expenditure) and the percentage of boarders do not vary, the size of the school (ae measured by the number of pupils), is positively related to exam score with 'an elasticity'Of 0.1, i.e. if the number ot students in the school is increased by tenjpercent, exam perform- ance is increased by one percent. ~be average number of pupils in the esc cycle is 258, or 64.5 per Form. Similarly, holding school size and expendi- ture constant, increasing the percentage of boarder.s by 10 percentage points increa.se$ average exam score by one percent, or cloa,e to two percentage - 156 - points. However, all three variab1esexp1ain only about 14 percent of the variation in exam score. 48. Equation (7.12) indicates that e.neJ.yzing schools on the basis of whether they do or do not have an HSe cycle reduces the coefficients of school sl.ze and percentage of boarders considerably. This reflects the larger size and higher share of boarders in school with an HSC cycle. 'l1Jle average size of lISe schools is 387 students, 70 percent of whom are board.era, whereas non-HSC schools average 196 pupils and 50 percent boarders,. HSe schools also spend more per pupil in the esc cycle (KEl13 compared to ~70 in non-HSe schools, per year). The result is a. difference of 8 percent or about 14 points in average exam score between the two types of schools, holding size of school and percentage of boarders constant.. The coefficient of (C) changes li ttle when expenditure per pupil, male education, and esc pupils per thousand population ar~ also held constant (see equation (7.1l». This indicates that esc schools with HSC cycles attract a type of pupil that does better even when a numbel' of other variables that affect exam score are held constant. esc Schools Grouped bY' Education Level in School' ,s, Counil. 49. In an effort to explain & larger fraction of the variance in exam scores, the group of' 115 schools was di"'dded into three sub-groups, based on average male education in the school's county namely: (i) schools in counties with average male education of less than or equal to 1.9 years, which typify most of 'the poorer areas of the country; (ii) schools in counties with average meJ.e education 'between 2.0 and 3.5 years, which includes all other areas; and, (iii) schools with both African and non-African students, the latter group being recruite~ from a population with an average eQ.ucation~ level more than twice as high as the highest district average for the African population. A~ could be expected from the statistically in8igni!~cant coefficient for ,male education in equation (7.8) ,differences in average exam. score between the three gro~ps of schools are not large: the first group has an average • score' of 1.74, and the other two, 1.85 and 1.80, respectively. ~be average values of the other main variables ~e shown in Table 7.1. 50. The results of the regression estimates within groups are as follows" Group 1: 35 scpoola in districts with an average education African m~es in 1262 'of 1.2'lyears or less: Q..f adult (7.13) log esc = 0.3554 + 0 .. 0404 log S - 0.1l.40 G + 0.1960 B (0.571) (1.702) (2.847) - 0.063 log (E/a) + 0.0540 (S/p) (1.000) (2.437) - 157 - (7.14) log esc = 0.6278 + 0.0119 log S - 0.0997 G + 0.2080 B (0.172) (1~554) (3.366) - 0.1281 log (E Is) + o~o433 (SIP) R2 Ie 0.389 (2.006) n (2.091) Group 2: 0 schools in districts with an aver e educa~ of adult African males in 19 2 of 2.0 - 3.5 l,e!:!!.: (7.15) log esc Ie 0.9511 + 0.1940 log S + 0.0317 B + 9.4890 ('l"/S) (2.452) (0.589) (2.336) - o. !~389 log (E/S) - 0.0007 (SIP) (2.085) (0.242). esc = 1.760 + 0.3933 C R2 :I: 0.201 (3.472) Group 3: (7.17) esc =1.606 + 0.0023 (E/S) R2. -.0.170 (2.478) (7.18) log esc = 0.5117 + 0.1304 B + 0.1484 (OT) (2.404) (2.108) (7.19) log esc = -0.0673 + 0.0862 log S + 0.0076 G + 0.1844 B (1.524) (0.157) (2.727) + 0.0263 log (E/S) - 0.0002 (SIP) (0.584) (0.107) R2 = .320 where esc, E/S, s, B, ,sIP, e, HlP, and log are 88 defined for equations 7.8-7.12 'and G • percentage girl students TIS = school's teacher-atudent ratio OT • overseas term teacher. as percent ot total teachers. En lS• non-teacher expenditure per esc candidate in ~ (net of boarding expend! ture ) • 51. This division into three groups does yield higher c~tficients of determination (R2). The coefficients ot the percentage of boarders are sig- nificant in the lowest and highest groups and almost equal with the same in- dependent variables held constant (equations (7.13) and (7.19». An increase in the percentage of boarders in both cases raises exam score more than in the aggregate esc regression estimate. A ten percentage point rise in board- e;rs raises the exam score by about tour percentage points. Table 7.1 shows the average number ot bo~ders in the two groups ot schools to be at opposite ends ot the spectrum. Since the Group 1 schools are'DiOst1y in poorer and , . '. ~ • i~ • • .•. . <. , Table 7.1: Average Value (Mean) of Main Variables Used in Exam Performance ~essions J esc and HSC Cycles) 19' > Percent Number Mean Non-Teacher Teacher Total Overseas Percent of Number of Score Expends/PuEil ~./Pu2il ~./PuEil Teachers. Boarders Students Candidates n) I CSC: All Schools 1.82 (KL) 20.5 ICb) 66.6 87.1 0.42 0.54 258 --- Group 1~ adult male. population 1d.th average of 1.9 year s of 1.74 21.8 67.3 89.1 0.42 0.76 178 32 ~ducation or less (less pros- perous and less populous rural counties) Group 2 : adult .male population with averate of Z-3 years of 1.8.5 19.8 64.7 84.6 0.44 0 •.58 219 39 education more pro~erous an"d populous rural," and Afri can " ,I: urban cOlUlttes) f-I' \.rt Group ~: adult male population (X) wi th 'average of 3.6 years or 1. 80 18.7 66.6 85.2 0.38 0.21 431 80 education or roore (mixed " urban schools) HSC: All Schools 2.66 30.9 143.8 174.6 0.67 0.69 65 24 %of Schools Avg. Male Education esc Pupi1s/10 3 Graduate Teacher Equivalent with HSC Cycle in District (years) Population in Dist. Per Pupil (number) esc: All Schools 0.23 3.27 11.0 0.054 Q:0up 1 0.08 1.52 2.4 0.054 Group 2 .)'. .. 0·24 2.~0 a.'3 0.,052 Group 3 0.38 6.56 26.4 0.054 HSC: J~l Schools 2.85 0,060 --- Source: Calculated from Ministry of Education data. - 159 - more thinly populated rural areas, a much higher percentage of students must board because of the large extension of -the school's catchment area. A nOl~­ mal level of boarding may be about 50 percent of the students attending, at\d the 76 percent average of boarders in rural, less-populated county ~ thus be equivalent in boarding intensity to the 21 percent among the ethnically mixed schools, which are all in urban areas. The 58 percent level in the middle group of schools, loca.ted mostly in urban. or densely populated rural areas, represents a comparatively higher level of boarding than the other two, and the substantially smaller regression coefficient for 'bof~ders ~ay reflect diminishing marginal returns to additional boarders. 520 When other variables are held constant, the coefficients of. expendi",.· ture per pupil and school size are significant only in the case of G:t·oup 2 where the coefficient of expenditure per pupil is negative, but the coeffi- cient of teachers per pupil is positive. This implies that although increas·... ing overall expenditure has & negative effect on exam score in that group of schools, increasing expenditure on teachers has a positive effect. Teacher expenditure per pupil as a percent of total expenditure is so highlY cor- related with overall expenditure per pupil that running the two variables together result in multicollinearity. The teacher/pupil ratio is therefore merely indicative of the teacher expenditure ~ffect. 1.1 When expenditure per pupil is the only variable used, its coefficient is significant and it explains 17 percent of the variation in exam scores in multi-racial or mi- nority schools. Even so, for every ElO increase in total expenditure per pupil (representing about 12 percent of average per pupil expenditure in that group), average exam scores would only go up by two percentage points 6 53. In Group 1 (1.9 years male education or less), non-teacher expendi- ture is negatively and significantly related to exam score, and the number of esc pupils per thousand population positively. '!'he latter effect is inter- esting, since it means that for this group of schools which are in countj.es wi th very low secondary Bchool enrollment ratios, increasing the number of esc pupils would result in some sort of external econoiny in the exam perform- ance of other pupils. 'l'he enrollment ratio e/P i8 not a significant variable in the other groups. !/ The positive correlation of the teacher-pupil rat.io and total. expenditure per pupil might be expected to eliminate the significance of the coef- ficients in equation (7.15). However, if this equation is estimated without including T/S, we find the'following results. (7.20) log esc = -0.3205 + 0.1484 iog S + 0.0301 B + 0.0278 log E/S (1.846) (Ou404) (0.402) -0.0010 log (S/P) 2 (O.324) R • 0.136 Thus the correlation between T/S and EIS leads to a more significant coefficient for E/S and a significant coefficient for T/S. - 160 - 54. The schools were &1.80 divided into those with HSe cy,c1es ,and those wi thout. Wi-chin the latter, there was little change in the a.ggr,egate re- sults; in the group of schools with the HSC cycle, the results of the regres- sion were very similar to those for the HSC exam itself, except that the coefficient of determination (R2) was regularly lower. HSC All Schools 55. Exam scores on the Higher School Certificate are differentiated by number of passes in principal and subsidiary subjects. rJ.1hewe:Lghti~g used here assigned a 5.0 to five principal subjects passed, 4.0 to four'"principe,l subjects passed, down to zero for no :principal subjects passed. Wi:th these weights, the l.'egression estimate yields the following relation . between inputs and exam score: (7.21) log Hse = -1.2470 + O~2614 log S + 0.6455 B - 0.2399 (E/S) (20456) (2.006) (2.001) + 1.4998 Et/E + 0.5198 log Hlp R2 = 0.548 (1.827) , (3.631) 'Where S, B, E/S, HlP, and log are defined as fOl· equations '1.8 - 7..12, and Hse = school 's average exam' 'performance E.... /E :; school fS expenditure "on teacher saJ.aries .:88 per.centage v of school's total expenditure. 56,. The HSe cycle averages a .DlUchsmal.ler number of pupils per Form (32.5) and a m'll.ch higher annual expenditure per pupil (~175) than the esc cycle ~ mostly due to much higher expenditure per pupil on teachers. As in Group 2 of the esc schools, the coefficient of' exPend! ture p~rpup.il is negative and the coefficient of the p.rcent ·of expenditure on teachers' salaries is pos:i.tj.ve ~ but not significant at the .tiv~ percent leve.l. As in the overall esc estimate, the coeffi'cient of the percentage of ,boa:r.d~rs in the school is pO$itive. In this case, tbe coefficient of adult male edu- cation in the cov,nty where the school is located is significant; it is posi tive and sho·ws an elasti ci ty ot 0.5: for every one percent increase in adult male education in the school's .county, there would be a.O. 5 percent increase in ~xam score.. These variables together explain 55 percent of the variation in exam score in the 33 schools observed. d.. ~e Rates ot H~turn to Addi.tional Inputs \\ 51'. From tbe Labor Force Sample Survey and the age-education..:.,earnings analysis .derived from it, we have found relationships between exam perform- ance at the, esc level and earnings (.see Chapters V and VI.) Gi ven the esti- mates of thecontril:>ution of' various inputs to exam score obtained ,from the previous sections, it is possibie to .assess, in analogy to section ,2d of this chapter, the returns of additional. inputs into the education process. The rates are s·u:mmarized at the end ,of the ,chapter in Table 7.2. - 161 - 58. The additional income associated with an increase in exam score is derived from the mean incomes of secondary school leavers with different results on their terminal exam. We estimate income differences between those with different scores on the esc exam (Division I, Division II, Di- vision III, GeE, fail) from two regression equations. The first estimates income differences by relating earnings of urban African males to exam score, no other variables held constant. The second relates e81:'Ilings to exam score wj.th age variables held constant, thus correcting for differences in age distribution. The results are as; follows: no other vari- age variables ables constant constant (Ksh/month) •- $ _ . . . . . .- Income difference between scoring Division II and Division I pass 20 34 (1.75 - 3.50 points on esc exam) Income difference between scoring Division III and Division II pass 133 120 (1.25 - 1.75 points on esc exam) If there were more observations in this subgroup of the Labor Force Survey, significant coefficients for exam score could be obtained when correcting for socio-economic background as well. With an even larger number of obser- vations in the sample group, we could make exam score a parameter, as in the case of the primary schooling analysis. Increasing Percentage of Boarders 59. The average cost per boarder derived from the school data is Kt25.8 annually. In order to determine the additional. private andaocial costs of boarding, however, the average expenditure of a secondary school student while Ii ving at home must be subtracted from the costs at school. From a survey taken in Nairobi in July 1963, the average expenditure of an adult equivalent member of an average size middle income family on food plus recreation plus household operation is Ksh 66.3/month. Housing expenditure was not in- cluded because it is probably not reduced when a young adult leaves home and goes to a boarding school. For a nine-month school year, in 1966 prices, this comes to K£3l. 7/school year. A similar survey of the former Central Province in 1963/64 yields average adult expenditure of KE5.5 (in 1966 prices) on. food, beverages, and services in rural areas per school year and of KE24 (in 1966 prices) in eight Central Province towns. J:/ Taking the simple average of these expenditures as an approximation of the deductible component in family expenditure yields KE20/school year. Net secondary school boarding costs are thus about Kt6 /year. 60. Holding the HSe dummy, size of school (8) end total non-boarding exp~,nditure per pupil (E/S) constant, equation (7.11) yields a coefficient ~!~--------------------------~ 1/ Ministry of Economic Planning and Development, Economic Survey of Central Province,. 1963164, Nairobi, 1968. - 162 - of B equal to 0.0902. To increase an average esc score of 1.82 by 5 per- centage points to 1.81. for example, wouldreq'uire an increase in ,-the per- centage of boarders of 24 percentagepoints~ SinRe the average percentage of boarders in esc schools is 54, this repr~sents -an,\ incre~e by nearly one- half' (44 percent). The number of boarders averages 1~2 per' school, so a 44 percent increase means 54 more boax:ders per school ,~'b;:t a" net i.ncrease in total social cost of KE324 per year and per school, or KEl.,296 'for Forms I through IV. Since the average number of stUdents per esc s:chool is 258, the average increase in the social cost per student as ares'ult of the additional boarders is KE5 for the four-year cycle. 61. On the benetit side, the increase' of five percentage points in exam score on the range 'between an average Division II (1.75 points). and. an aver- age Division' I scoI~e (3 .. 50 points)res,ults in.an addition.of (5/175) tim~a 20-34 shillings/month = 0.5 ...1.0 Ksh ·a:;',month to the averagesal~ry of urb811 African JIiales. If this salary increase of Kro. 3-0.6 per year is taken as a perpetuity, the rate of return whicheq...,tes the additional earni~s to the addi tiona! cost of boarcUng rti.ng~~ from r :II °5· 3 , or 6 percent perannwn, to 0.6 .~ r = ~ = 12 perce~t per annum. $~e increase of five percent~e points in exam score betlTeel? an average D1'v:4Iilion 1:11 (1.25 po~nts )'i-and8il average Di- vision I I score (1'.75 points )resu1ts in an addition of 5/50 ·times 120-133 Ksh a month , or Kt7 -8 a year. If this increase is taken as a perpetuity the rate of return ranges :f.rom 140-160 percent per annum. Increasiing Total Expenditure 62. The relationship between.~am score and total expenditure per pupil estimated for the entire set of obs'ened CSC schools yields a . slope' of 0.0016 .0 (equation (7.8». For every u.6. 25inerease in expenditure per pupil ~er year ~he average exam score is, expected to rise by one perc,ntage point. The estimated relationship does not include ~ other variables.. A five percent- age point increa.se;:~Jn exam score, ·from the observed mean ,of 1 •. 82 to 1.87, woUld require KE31 ~~5 additional. total expenditure per pupil per year, or ~125 f9rthe four-year esc cycle,. & substantial increase over the observed expenditure of RE84. 7. Nevertheless, a five point increase in exam score 1 between Pi visiqJl, II (1.75) and Di via i-on I (3. 59) resultsin a ;maximum 1n- (fi crease in earnings of (5/175) times Ksll 20-34 monthly, or KfX) .3-1(£0.6 an- nually • 63 • As~uming again that theaddi tional income is a perpetuity, the r$te of return can be estimated as: r ~:I: -= 1 25 0· 0025 , . 0.6 it r Ir: 125 -= 0.00;. ;), T'he 0.25-0.5 percent rate of returnca.~culated in this way is "an . upward biased estimate, since income differences from higher exam scores are lower in the t] '\\ - 163 - early working years and higher later. The rate of return is higher to a.ddi.- tional. expenditures made at l01lrer a.verage exam score levels because the income difference for those sco:t.·ing a Division II l)a.ss and those scol"'ing a. Division III or GeE pass is greater than that between Division I and Division II. Since both Divisions I and II are clearly above··average passes on. the ese~ it is a plausible result that the income dif.ference between those who obtain a poor pass or who fail (Division III, GCE, and fail) and those in the Division I/Division II group would be greater than the diffeltences w:(thil1 the latter category. 64. If the est:l.mated income differences are taken from the regl"ession results when age va.l\:t'!lhles e...t'e included, a five point increase in exam s~ore from, s'ay, 1.70 t9 1 .. 75, adds Kli7 annually to the average incoJlle of Africt Armitage and c. Smith, !!!.!. DevelOpment ot Computable Models ot the. BI-itiah EducationalSyatem and their Pos8ible Ulea, OECD paper DiSJEI:07ii~ .15, Paris, 1996 (miIIleo). For Ul application of thea. teclll'liquea, aee E~ducf!!iion, .Human Resources and Development in Argentina, Annex: Methodological Problems 14'"ld Statis- tical. Data, OECD, Paria, 19~8. 2/ Fittins a logiatic curve to the 1955-66 data did not yield a satisfactory resUlt:; the value tor C (1i,h. utpper limit towards which the enrol.lf1ents Y ~ cony,el"ging) being 1.16 mi?;lion, which is smaller thauthe 19~,8 obser- vation).;1 . ! 1 IL \\ • Table 8.1:: Four Estimates for Total Prima12'. Education Enro~lmen~, 1969-75 (in 1,000) r~~~;- A 1/ B ~/ c 2/ nW 1969 1,285.9 1,293,,4 1,198.2 1,169.8 1970 1,366.9 1,32-7.4 1,227.4 1,189-9 1971 1,453.0 1,359.9 1,253.7 1,207.2 1972 1,544.5 1,)88.7 1,276.7 1,221.9 1973 1,641.8 1,415.8 1,297.0 1,234.5 1974 1,745.2 -1,441.5 1,)14.6 1,245.1 1975 1,855.1 1,465.6 1,330.1 1,254.2 ~ ()\ \0 '. t A_ !I Assuming an annual i.ncrease of 6.3 percent between 1968 aDd 1975. (}.. 567 - 0.958 • O.918t • 2/ Derived from Y • e ' tl - 1959. ./ ;. 1 426~r. 11 Derived from y. 1 + ·1 1 7· • 2 -0 162 t J t l · 1959. • e ..1" t l!I Deri.ved from Y .. eO. 265 - 1.427 • O.S42 ; t1 .. 1955. Table 8.2: Fducation 1969 1970 1971 1972 1973 1974 1975 T9tal Enrollm,!!!! 1293.9 1327.4 1359.9 1388.7 1415.8 1441~5 1465.6 Outflow of Pupils !I Standards I - V: 11.& 83.1 ~ 75.5 74.2 76.8 Standards VI - VII: 11£1.2 131.5 13B.9 140.5 160.5 ~ 180.0 Studen~ who continue th~ir education1 37.4 3.8.3 39.4 40.7 42.2 44.0 Bet outfiQV Ire:. Standards VI VII: 103.8 93.2 ~ 22.:! U8.3 136.0 '"'! H oN 9 .. ".!I Calc;~lated on basis ot pupil nov ticures tor the period 1960-68. -; '- Source: Based on columnB in Table 8 .1 and on progression assumptions as outlined in the text. 1: ..... 4 - 171 - model as referred to on the first page of this chapter. However, there has been an attempt on a geographically reduced scale (rural Central Prpvince) to analyze the demand .for family expend! ture on education (among many other expendi ture items) from household survey figures. !I The results show elas- ticities of expenditure on school fees of roughly 0.8, with regard to total household expenditure; 1.4, in relation to household size and -0.9, in re- lation to age of the ho'Usehold head Y. The total expenditure elasticity has two components, namely that relating to cuh expend! ture (0.5) tUld to subsistence expenditure (0.3). Assuming (i) that cash expenditure If' 5J.wnys a fixed share of cash in,~ome, (ii) that subsistence expenditure eqtl£i.J.s sub- sistence income and (iii) that the composition or households by size remains unchanged, family expendi'ture on education will grUIl at a rate eql1A.l ,to thnt of population increase, augmented by half' the rate of ~ncrease in cash in- cOmes and 0.3 times the rate of increase in subsistence incomes. ~ 7. While the author we cite is careful to avoid generalizations, he concludes from the insignificant inter-district differences in' the sample of Central Province households that the results can probably claim validity for other rural areas of Kenya. 3/ We made a crude attempt to check these re- sults against available global data - per family cash and Don-cash income in rural areas, 1962-65, as estimated by the Minietry ot Economic Planning and Development, and the fee and enrollment figures tor 1963-66 (corrected for estimated urban enrollments). If one &Bsumes an annual population growth of 2.75 percent in Kenya's rural areas (which ia close to what most observers consider to be the present rate of increase), B. F. Maasell's findings for the Central Province appear to be consistent with the overall development of enrollments during that period. 8. In another attempt to assess the differences in the demand for pri- mary education separately for boys and girls uaing county data, we relate the percentage of boys and girls ot age 6 - 12 enrolled in primar,y education (Yl and Y2 respectively) to the ratio of the cumulative fees per pupil trom Standards I through VII and the per capita GPT (Graduated Personal Tax), a 1/ The following paragraph draw!J to a large extent on> B. F. Musell, Deter- minants_ of Household Expenditure in Rural Kep..ya, mimeographed paper, 1968. His observations referred to expenditure on both primary and secondary' education; the results are thus not strictly comparable. How- ever, the secondary school enrollment. in 1963 were so small relative to primary enrollments that even the considerable inverse fee difference, tailed to give private secondary education expenditure a sizeable share in total education expenditure. 2/ We disregard the statistically insignificant coefficients for the other variables. 3/ A similar analysis for Nai,robi by B. F. Musell and J. Heyer yielded an elasticity of education eXpenditure with regard to total expenditure of -0.232, but the coefficient was statistically insignificant. - 112 - proxy for income (X). 1/ The results obtained from cross-section, data,of 14 counties trom all provinces are: (8.1) log 11 • 5.93 - 3.74 loge X; (8.2) log Y2 • 5.67 - 3.39 loge X; 9• A one percent increase in the tee/GPT per capita ratio, i . thus' associated with a 3.7 percent decrease in the enrollment ratios' for boys (3.4 percent tor girls). It GPT revenue 1s a good income proxy, then the enrollment ratio for boys would increase at approximateiy 3.;' (3 .. 4 for.' girls) times the difference between the rate o~ 1ncreaee in fees and th&~ of' increase in income (or decrease, if the ditference' vere negat1 ve) Y.' 10. A s1m!iar relationship is fOund 1letween the family expeuct1tur.e on school fees (W) and GPT revenue (Z). The respective equatiOlls' are:, 2 (8.3) log Wi a 1.13 + 1.49 lOSe Z tor' bo.y. (R • 0.94) . 2 and (8.4) log W • Q~76 + 1.84 108 Z tor sirls (R • 0.96) 2 e It is not necesearily a contradition that th~ elasticity ot demand· for' girls' education is higher than that for boy. iD the second pair of eqU&~ion8. and lower in the former. This m8J indicate a greater income elaaticity aDd smaller price elasticity of the demand for educa~~on for girls relative to that for boys, which would be a plausible resUlt ~'. ' 11. These equations are obviously Dot auitable for projecting the" i~di- viduale' demand tor education. This i. Dot .0 much because ot' the modifi- cations the original data had to undergo ~ but rather becauae of the double 1/ Since there, are no geosraphical~ di.qp-esated income data, the GPT had to be talten .. a not quite satistactoJ7 8ub.titut_ • The la.tter would be a pertect proxy ot inca. onl1' (i) it the lnccae di.tributions of the counties vere larsely similar Uld (Ii) it the efficienC7 ot the tax collecting author!tie. did not ,how inter-count,. difterence. • Neither ass~tion is particularly cODTinciDi. 2/ It should be emphaaized that this approxtmation holda for small to mod- erate change. ot tees aDd inca.. 0Dly. 3/ Another .et of equatioll8 (vi th very similar re.ulte) related tees paid and averas_ GPr per t_ily. TIlie a4.1utMnt for size o( t.a1ly "ea made on the baai. of the 1962 CenaUl tisur•• OD th~ size ot houaeholds. 4/ These consisted in an upda~"ns ot the 1962 population data, applyins the same population growth rate to all the 14 cOUDties. ...UIlins the .&me 88e/aex breakdown &8 in tbeCellllU81ear. Another difficulty, naaely that of ditter~nt reterence year. - 1967 tor the e~llment .tatiatlcs, 1966 for the tee data, and 1965-66 for the GPr figures - was 'not tackled at all. - 173 - role of GPT as indicator of average individual wealth in a county, a.Yld as the county councils' main source of finance. If the potential demanci for edu- cation persistently and universally exceeds the supply of school places, an increase in GPT receipts would enable the county councils to estl!1blish and maintain more primary school classes, with a corresponding expanllJion of en- rollment figures following automatically. This may explain the asurprisingly '. 'r high coefficients of determination (R2 II: 0.94 and 0.96, respecti,l'ely) in the second set of equations. 2. The Demand for Secondary and Pjst-Sec;ond14n Education 12. The past growth of secondary school enrollments (we :i.Delude general and technical secondary), has been such that it is pointless to use time series analysis for projections, particularly in the yeus after iridependence • .An attempt was made to i it logistic and Gompertz curves to the 1959-68 en- rollments in each form, but only in the caSe of Forms V arid VI do the results turn out to be- suitable for projections. 1/ . 13. The ~oots of these difficulties lie in the 8p~ntaneous growth ot Harambee schools beyond governmental control. A:rter an\ eari¥ and unsuccessfu1 attempt in 1964-65 by the Kenya Education Commiesion to stem the tide, the government was forced into a pusi ve role. The foundation of new H8.rI!~bee schools often went unknown to the government. The mot:tyes of the lOc~~ schc:>ol committees in undertaking the considerable financial sacrifice~l that went with the estaolishment of & Harambee school Y were.,...., . obvious: the 'ioutput of the country '8 primary schools had grown at a pace m;!lch faster than:\that of wage employment , with an ever increasing share of KPE certificate llolders remaining unemployed. By securing them two or 80 year~ ot additional \~dU­ cation, the parents hoped to improve their children I e chances in the ll\bor market or to later place them in a regular secondary sch~ql which led to the esc or HSC. An even more remote hope was that the government might eventually take over and maintain the Harambee schools. 14. However, t1:lis strategy proved to be correct ~y tor those who innovated it; for the great majority, it merely led back to the situation prevailing in primary education a few years before. Tire Labor Force Survey ~ data suggest that only three out of tour ot the 1966 Form II leavers found urban employment; for the 1967 and 1968 cohorts, thia share must have dropped atill further ,"~ perhaps to as little &8 10 to' 20 percent. In the ca*!! ot the logistic functions, the solutions obtained were mean- ingless, yielding negative· values tor C, the upper limit tor the enroll- 1 ments. The Gompertz curveatitted the put obaenations tairly' well but implied unattainable 8rovth rates (leading to IIOre than ODe million Form I enrollments in. 1975). The main reuon tor theae ditticult1es v .. the. abaence ot • distinct slowdown 1n the recent growth process; only fram 1967 to 1968 did a slight intlection occur. 2/ See Chapter III. .j f. ., - 174 - 15. There are two possible reactions. of the Harmnbee school coD1l14ttees to this situation: they can either try to expand their schqo1s up to, Form IV and enter their pupils for the esc (which a number ot them;have already managed to do), or else reappraise their earlier decision. ' The first- optiOl)' cannot p~ovide for more.than a temporary solution; it has became clea~ by now that the "movable surplus" is about to reach the esc level. -The 6,econd <ertlative would mean a reallocation ot resources from education. to areas such as rural water supply, livestock improvement, or better farm equipment, or from Harambe~ schools to other kinds of education. 1/ Whether th~a more realistic assessment of the gains that could be reaped from education would lead to a rapid decline of the Harambee school system or merely to, a ~tag­ nation and gradual decrease in enrollments is an open question. Y ';\ 16. \0 In the present stugy, a rather optimistic projection has been mo.de., as'sumin~(a growth of Form r:enrollments. in government maintainecl. SJld _sisteq. schools \:9f the order of the 1963-67 increase, i.e. at 10 percent annually, and annui.:\ decreases of 1,000 in the Form, I enrollments in' unaided schools after 1968) (brought about partly by an absorption ot some Harambee schools into the ~Jystem of aided secondary 8chools, and partly by a net decline in the demBn;(4). The transition probabilities are assumed to be 95, 85, and 75 percen~ throughout the projection peried (which correspond to the average coefficien~s~ for the 1963-68 period). The results are shown in Table 8.3, ~d ~e us~~)d in part of the Chapter IX labor market projections. . ' ~ , 17. PrOjections of Forms V and VI enrollments have been based on lo- gistic and Ganpertz functions fitted to the 1959-68 obs.ervationa (columns A, B, C, and D i,n Table 8.3). Our pre terence for the more conservative esti- mates derived from the latter type of tit is explained by the heavy costs Qf !I The Kenya E1iucation COIIIII1ission, :1:n taking a stand against the uncon- trolled spread of Harambee schools, sugg~_,.ted that sui table obj ects for local initiative in the field of educatid;b "could be found in primary education,;"ln lim! ted contribution. towar!,:~ the building of maintained secondEJ.ry' sphools, or in adult education"l~ (Republic ot Kenya, Ken"ya Education Commission Report, Part II, Nairobi 1965, p. 24). 2/ The history. of the $econdar.y modern schools in the Western Region of Nigeria ia an instrUctive example. The!se had developed after 1955 from ;:l a similar surplus situatiqD in primary schooling. Like the Harambee schools, most, of them were locally organi 2ed and financed. The parallel to the case of Kenya goes even .turther in that the government made the same unsuccessful initial attempts to channel this development. In 1957 alone, 254 new schools wi th 30 ,000 pupils appeared on the scene. However, after 1961, the impetus ot this movement was gone, and enrollments dropped from a peak of nearly 75,000 in 1963 to about 45,000 in 1965. (Source: S. Weeks, Diver ance in Educational. Davelo ment: The Case of Ke a 8ild Uganda, New York, 19 7. The author thinks. it is well possible that the Harambee schools will meet the same tate as Western Ni~eria's secondary modern schools, a view we are inclined to share. t ., Table 8.3: !ez,~ t .EhroJJ..mel1ts and Outflow of~Secon_~a:r-:( .2E4. Post-Secondary Schools oy 1'Orp1, 19b9-75 ~ Projected Enrollments 1969 1970 1971 1972 197;2 1974 1975 Form1:Y 36.7 37.4 38.3 39.4 40.7 . 42.2 44.0 Form 2: .34.3 34.9 35.5 36.4 31.4 38.1 40.1 Form 3: 27 .. 0 30.7 31.2 31.8 32.6 33., 34.6 Fora 4: 18.1 23.9 27.1 27.5 28.1 28.7' 29.6 Form.5 (A~: 2.2 2.5 2.9 3.3 3.8 4.2 4.7 Form 6 {B~= 1.7 2.0 2;4 .. 2.8 3.2 3.7 4.2 ~ ::Form 5 (c : 2.3 -'2.7 3.1 .'. Form 6 (n)2l: 3.5 3 •.9 4.3 4.7 1.9 2.2 2.1 3.1 3.5 3.9 4.2 Estimated Net Outflow of Pupils I :t From forms 1 and 2: · 5.4 ;>. 6" oJ 5.6 5.8 5.9 6.2 Fro. forms 3 and 4: 18.7 24.6 27.5 27.4 21.8 27.9 From ioms 5 and 6m 1.9 2.1 2.5 2.9 3.3 3.7 Fro. Poet-5econdar,y Institutions: 0.7 O·fl, 1.0 ~ J..~ 1.5 1.8 .'. , I f-' -.J \J1. 1/ FOrDlI enro1laJts are based on the aSSlDIption that enrol.lJarts in aided schools will increase by 10:"percent per year - between 1969 and 1975, _ereas enrollJllalts in unaided schools will decrease by 1,000 annually •.... ,Progression pro- ~bi.ll.ties are assWled to be 0.95, 0.85 and 0.75 throughout the projection period.··~ .. " 2i Derived iro. Y • e 3.427 - 4.991 • O.944 t ; t 1. -1959 - 3.553 - 5~890 • 0.942t ; tr'- 1959 J;.. 3/ Der! ved lro. Y • • .~ - 6.552 1V Dsr1 ved il"Qm Y • . 1 + 28.688 • e - 0 • 250 t J t l .- 1959 21 Derived from Y • 5.653 ; t1 - 1959 1 + 53.197 • e - 0.296 t . Source I Table.3.2. ~~ ~~h ". "1"\ .... ': \. '.~~''': - 176 - establishing and ~~ning senior secondary classes and by the possible emi- gration of non-African pupils who otherwise would have qualified for Form V. 18. The demand for post-secondary education is very difficult to evalua.te~partly because of the border nature of some specific institutions or types of education (such as Sl tea.cher training, some courses at t.he Kenya Polytechnic and the Mombasa Technical. Institute, the Egerton College of Agriculture, etc.) partly because of the absence of information on over- seas students (more than 3,600 in 1966), many of whom are certainly not pur- suing university studies in the narrower sense. An additional difficulty relates to the uncertain future role of the ethnic minorities who still account for a sizeable share of the demand for secondary and post-secondary education. After weighing all these questions, it was decided to take re- course to a very global estimate by assuming that 75 percent of li'orm VI leavers would receive SOOle uni versi ty or other post-secondary training, and that on average, they would become available for the labor market three years after leaving secondary schools. The resulting supply figures are included in Table 8.3. - 177 J' IX. THE FUTURE DEMAND FOR EDUCATED LABOR 1. Introduction: The Projection Techniques .' 1.. OUr approach to the estimation of the demand :for labor with a given education level drops the assumption made by the traditional manpower (\ approach that demand is completely inelastic with respect· 1;,0 wages ('ind is related to output by: technically fixed coefficients. We a~telDpt rattLer to estimate this demand by relating the quantity of each skill demand,c-d to its relative price, i.e., to-the wages ot workers with the mioUnt and type of education that is associated with. this skill. We also m8ke use of. the labor supply projections of Chapter VIIle Our purpose 1s to get a pictUl-e of Kenya's labor market in th~ year 197~: '" . 2. Two types of estimates of demand for labor are carried out, the one oased on wage, employment, and output time series (Section 2) and the other derived,trom p~oduction,func~ions which include as variables capital and ,sevex:al kinds of l,abor inputs (Section 3)-. ' The first approach requires time series data. on rela.tive wages and a breakdown of the labor~ torce by level of flducation. From these data, it is possible to estimate demand directly, using a cru~e demand,equation for each type of labor: ." ;. = d 'Wx ( t ) wr{ t ) (9.1 ) a + b 'Wx(t) + c GDP(t) + + e W:y(t) w~(t) where· ~(t) = the qU8D~ity of workers with x education in the ~ . labor force in year t, = real wages or salary paid to workers with x education in the labor force in'!year t, = real wages or salary paid to workers with next highest level ot I education, y, in the labor force in'year t, Wz(t) = real wages or salary paid to workers with next lowest level ot education z in the labor force in year t, and GDP(t) = gross domestic product in prices ot year t. The coefficientwb is an estimate for the slope ot the, demand curve for labor as a tfUnction of real wages ~ with domestic product and relative wages held constant. if the regression is run 6S a log equatio~, then b represents the -,J , II. o ' .... - 118 - price elasticity of demand and c the income elasticity of demand for labor with x education. 1/ The coefficients d and e relate to the wage 'stnacture. 3. If time series data are Dot available, then 1 t is necessary to de- rive demand from some kind ot production function estimates for the economy.: as a whole, a. much more complex method. It requires cross-section estimat'es of production functions in which labor itt,'no longer treated &S a sing1e input but· rather as a collection of difterent kinds of labor, each charae-h terized by different amounts of education, and each consider.edaa a separate input in the produc~1on process. Using a Cobb-Douglas produc~ion function as an example, the demand for a gi ven. ~ype ot labor can be derived i,n the following way. Equation (9.2) is the production function for a given industry, :t.orexample, 'Wherr L1 ( i ), L2 (i)' and L3 (i) are different skill levels of labor, K( i ) is e:apt tal, and V(i) is value a.a.ded in the industrY. ~ : 6 Veil 6 L11.(1) wbere 6 V(i) is the marginal. productivity ot Ll(i), or, d.n a perfectly 6 L1 (i) competi ti ve si t:uation, wages ot L1 (i) '. Similarl.y', the demand curves for other types of labor as functions ot ,output ot the industry or of the economy 1/ Since th~ supply fUnctions S • t(W/p) il not specified and the demand ana. supply equations are not solveds1multaneously, the absolute v:alue of b would be a downward biased es·t1JDate ot the price elastic! ty of demand. The higher the price elut1city ot supply, the greater the bias in b,and vice 'Versa. We would therefore expect th'at theesti- mates of b tor labor with prilii.ai7 schooling would be more biased then the approxim~tions for second&ry' school or Wliversity educated labor, since the elasticities ot, supply ot the two latter categories are lower than that ot the first groUp. Our '1914 eIIpl~nt estimates tor a given wage level ot primary school leaversare thuS underestimated. Corre- spondingly, to bring about a drop in the rate ot return, the increase in the output of t:p.e education system by 1974 would have to be grea.ter than estima.ted below. However" there is no bias in the, estimates 'which we make Qf unemployment in 1974 under th~ ~sumption ()t constant ,re,al, wages. " - 179 - 8lld of wages can rpe /p.eri ved. Other kinds of functions besides the Cobb- Douglas, such as ~ttie CES (constant elastici ty l'~t aubstitu~ion) tunction, can also be used to assess demapd for labor. \1':'1 \ \ f In Section 2, we de~ive demand functMiJns 'of the equation (9.1) t' ',\ I 4. t type trom proxies for the requil':'ed data on wa,g~s t employment, and output by education group. In Sections 4 - 6, several ot these demand :f'unctions are used to make labor market projections for three sChoolin~levels: primary school leavers (Section 4), second~ school leavers (Section 5), &~d upper secondary school and university leavers (Section 6). For each of these levels, two types ot projections are made: (i) projections of emplOY1J1ent and real wages, assuming that the supply of labor with a given level ot education is'" a function ot the rate ot return to investment in that ley,el, and that rates of return tis1,\ investmen~ in a:ll leyels ot schooling should be equal to lOpercent (parts 4(&), 5(a), and 6(a»; (i1) projections of real. wages t assuming that the supplq of labor is given by school output between. 1968 and 1974 as determined in Chapter VIII, ~d postulating that tull employment is to be obtained in 1974; alternatively, projections of surpluses or shortages of labor with ditferent educational qualifications in 1974, .given a fixed level of real wages (parts 4(b), 5 (b ), and 6 (b) ) • These two kinds ot projections are made in very similar wqs; the main dif- terence between them is that the tormer implies that "~he suPPlY of labor varies wi'tt.~ the retur~ to education, while the latter)' takes the supply of' labor as gi ven by the projections. ~.n Chapter VIII. ' , 5. Our attempt to estimate',. demand for labor using production functions in Section 3 is UDsuccesstul, tor. two reasons, both connected with data prob- lems. First, we have to use two 'groups ot cross-section data on Kenyan firms to obtain all the necessary information: :. .'t, (i) value of output by tirm (1966), replacement value ot capital by tirm (1963), total number ot employees by tirm (1963, 1966, 1967) for all 126 manut,acturing tirms participating in the Kenyan Government Manu- tacturing Surveys, and (ii )' wage and employment data tor the 20 tirms participating in o\~ Labor Force Survey (Jan.-Feb. 1968) which em- 'ployed persons in the three labor categories we decided to use. However, the confidential natur'e' ot the government employment/output data required that we use it without being able to identi.f1 individual firms; therefore we could not matchrthese data with the individual firm employment/ .alary tigures from our survey •., Second.ly, the firms in both samples are too , " ,. - 160 - heterogeneous for successful cro3B-section analysis, and data were, not ~va11- able to us to do & longitudinal analysis using time-series tor one :rirm or a group of similar firms. 2. The Demand for Labor Estimated Directly from Wage!=, and ~p1oyment. Data 6. Although the time s~ries data on employment and wages o~ workers by year. of education which are specified in equation (9.1) are not available tor Kenya, there exist time series on salaries and employment of Kenyan males by ethnic group and economic sector which we use as proxies. Our data cover three ethnic groups (African, Asian, European) and three sectors (commercial. agriculture and forestry, private industry and cODmlerce, public sector), which can be canbined into nine ethnic-sect~r groups (see Annex Table 9.1). We then use these groups as proxies tor education groups by assUll1ing that Africans in commercial agriculture and forestry have an aY~rage o~ one year of schooling; that Africans in private industry and commerce haTe an average of four yea.rs of schooling; that Asians in all three sectors have an average of seven years, and that Europeans in all sectors have nine years. This en- ables us to develop nine equations ot the form of equation (9.1), some of which are used in the later sections to make projections to 1974. 7. The nominal earnings data for the ethnic-sector groups from 1957 to 1966 are ~hcnm in Annex Table 9.1 and; the employment figures for the same period in Annex Table 9 .2~ In both commercial agriculture and private in- dustry end commerce, nominal salaries have groWn steadily during the ten-year periocl end employment has fallen. :i:n the public sector, Africans' earnings have risen and ao has their employment, particularly since independence. 8. Th.roughout the ten-year period, gross domestic product (GDP) and prices haTe also been increasing. GDP and the cost-of-living. index ~or Nairobi are shown in Annex Table 9.3. The Nairobi cost ot living index is used to deflate all salaries and GDP. 80 that ·both are expressed in 1957 prices. Since prices probably rose more during this period in Nairobi than in other towns end the countryside. the index tends to give a downward bias to real wages and GDP. The estimated coetticients relating employment to real wages and real GDP therefore are probably biased upward. 9. Adaptin~ equation (9.1) to these data, we use equations of the following torm. Y log N1 ,J,k ~ Ri,j + bi,j log (Wi,J,k/Pk) + Ci,j log (GDPj,k/Pk) where N -= employment (in 'thousands) W/P • nominal salar1es(in KE) deflated by the N~irobi consumer price index to 1957 prices, and GDP = grOBS domestic product (in million KZ), deflated by the Nairobi consumer price index to 1957pr1ces. 11 The inclusion ot the variable Wi/Wj, Which relates the wages ot Atricans in one sector to those ot the same group in another sector. .howed little success and 1s therefore not deacribed here!', 'I /' - ~f8l - The subscripts i, j, and k indicate ethnic group, economic sector and year of observation, respectively. The regression estimates are made tor the para- meters a, b, and c by ethnic group and sector. In the ease of the government sector a dummy (I) is added which takes a value of zero tor years before independence (1957-1~63) and a value of unity' tor years thereafter (1964-1966). 10. The results ot these regressions are given below (the figures in parentheses refer to the t-vaJ.ues of the coefficients )11, and are schematized in Table 9.1. AGRICULTURE (9.5) Africans log N :: 7.5062 - l.~i301 199 (W/P) + 0.6724 log (GDP) t,·{~.296) (2.a56) 2 " " R :: 0.931 (9.6) Asians • log N = 7.4541 + 0.150 log (W/P) + 1.6712. log (GDP) ( 0 • 76'4' ) ( 4 •398 ) 2 R = 0.787 (9.7) Europeans • • r ~,,~Q629 log (GDP) • log Ii = 12:.0392 - 1.0008 log (W/P) - I"~ (-1.884) (1.761) 2 R == 0.601 1 PRIVATE INDUSTRY AND COMMERCE (9.8) Atricans log N ~ 5.2204 - 0.7408 log (W/p) + 0.6648 log (GDP) (9.~65) (4.ioo) R2 a 0.950 1/ In the following equations, all subscripts indicating year, ethnic group, and sector have been omitted in order not to complicate unnecessarilY the formulae. VaJ.ues not significant at the 0.05 level of probability are starred. . - 182 - (9.9) Asians log N • 1.0340 + 0.1257 log (W/P) *+ 0.3042 log (Gnp) • (0.267) (0.948) 2 R .-: 0.498 (9.10) Europeans log N • 9.9751 - 1.4844 log (W/p) + 0.6341 log (GDP) * (3.075) (1.098) 2 .R II: 0.836 Africans , (1.701) . log N = 3.7256 + 0.3500 log (W/P) - 0.1182 log (GDP) (0.371) . 2 R :II: 0.766 - log N = 4.7610 + 0.0960 log (W/P) *- 0.0802 log (GDP) *+ 0.1382 I (0.533) (0.334) (2.528) R2 = 0.887 Asians log N = 14.8484 - 2.1438 log (W/P) + 0.3688 log (GDP) * (4.181) (1.121) R2 = 0.885 log N .-: 6.2136 - 0.7924 log. (W/P) •+ 0.3932 log (GDP) *- 0.3303 I (1.472) (1.823) (3.212) R2 = 0.956 (9.13) Europeans log N = 9.7945 - 1.5780 log (W/P) - 0.2354 log (GDP) • ( 3.002 ) ' ( 0 .394 ) R2 •. 0.816 . log N ~ -6.2400 - 0.3386 log (W/P) *- 0.5304 log (GDP) * (0.738) (0.901) - 0.4002 I (2.437) R2 -= 0.907 Table 9.l: Kenya: Values of Regression Coefficients in Demand for Labor Equations Based on Employment, Wage, and Output, 1957 - 1966. "'= Form of Regression Eguation!' log N :II a + b log (W/P) + clog (GDP) +'<\ d (1) where N - employment (in thousands) W/p = nominal salaries (in KIs)deflatedoby the Nairobi consumer price index to 1957 prices GDP = gross domestic product tn millions of Kb), deflated by the Nairobi consumer price index to 1957 prices 1 - independence dummy, for government sector only; \) equals zero for 1957-63 and unity for 1964-66. Values of Regression Parameter~1 ( t - values in parentheses) Source: Sectors and Equation No. Ethnic Groups a b c d 2 11 (interce~t) (real wage) (GDP) (1nde2endence) Agriculture I (9.5) Africans 7.5062 -1.2301. (7.296) 0.6724 (2.856) (9.6) 093 J-I Asians 7.4541 o.~oo. (0.764) 1.6712. (4.398) .79 ~ (9.7) Europeans 12.0392 -1.0008 (1.884) -1.0629 (1.761) .60 I Private Indust!l & Commerce (9.8) Africans 5.2204 -0.7408. (9.465) 0.6648. (4.100) .95 (9.9) Asians 1.0340 0.1257 (0.267) 0.3042. (0.948) (9.10) .50 Europeans 9.9751 -1.4844 (3.075) 0.6341 (1.098) .84 Government (9.11) u Africans; no (1) term 3.7356 0.3500. • (1.701) -0.1182.'* (0.371) .77 (1) included 4.7610 0.0980 (0.533) -0.0802 (0.334) 0.1382 (2.528) .89 (9.12) II Asians; no (1) term 14.8484 -2.1438. (4.:181) 0.3688. * (1.121) .88 (1) included 6.2136 -0.7924 (1.472) 0 .. 3932 (1 .. 8l3) -0.3303 (3.212) .96 (9.13) u Europeans; no (1) term 9';7945 -1.5780. (3.002) -0.2354 • (0. 3:il~) .82 (1) included 6.2400 -0.3386 (0.738) -0.5304 (0 .. r.~Ol) -0.4002 (2.437) .91 a/ See Equation 9.4 in text. bl Values ~ significant at the 0~j)5 level of probability are starred. ... 184 - 11. The regressionestiJllates re1a.ting African employment to average salaries end gross domestic sectoral product yield significant coefficients fer both independent variables in bath caamercial agriculture and private industry and commerce. The salary and GDP variables explain 93 percent of the variance ot African employment in agriculture and 95 percent of the variance of African employment in private industry and commerce. As expected, the price elasticity ot demand is higher in absolute value for the agricul- tural workers than foX" industrieJ. workers, probably because ot di:fferences in schooling: industrial workers have, on average, "about tour years ot school- ing .compo.red to perhaps one year tor Africans in coaercial agriculture. 12. The coefti.cients of sectoral GDP are almost exactly equal for Af'x-icans in agriculture and private industry a:nd commerce. For a one percent ~ increase in sectoral GDP, African employment increases by 0.67 percent, sal- \ I; aries being held constant. This imp1r~es that real sectoral GDP would have to grow 4" 5 percent per annum for the·se two sectors to absorb a. 3 percent annual increase in the labor torce in the two sectors 1/. It sectoral GDP does not grow at 1~.5 percent per arJ.n\Ull, but the population growth rate re- mains at the 3 percent level, the two sectors will not be absorbing a propor- tional sh~ ot the labor -- other sectors, $uch 88 the government or sub- sistence agriculture would haTe to take an increasing proporti~q of the labor force if lmemployment rates were to remain at the seme level. Y GDP in both sectors grew only about 3 percent per annum in real terms during the ten- year period, so they could have been employing an add! tional 2 percent an- nually in their reapective sectoral labor forces, had se.laries remained con- stant 9 Since average sal.~ie8 in real tems, however, have also been in- creasing, African employment in the two sectors has remained essentially un- changed between 1957 a.nd 1966 .~\ 13. The estima.ted parameters tor Asians and Europeans, (eq,uations 6, 7, 9, 10, 12, and 13) are considerablY less consistent than those for Africans. The former reflect changes in employment which stem from factors other than salary increments or changes in sectoral GDP. Thus, the coefficients of determination (R2) are considerably lower tor Asi611s and Europeans than tor Y If population grOW's at 3 percent annually, and the percente,ge ot popu- lation that enters agriculture and private industry does not change, em- ployment in the t~o sectors would elso grow at 3 percent per annum. gj This rather pessimistic conclusion is supported by the findings of a recent study on employment and output in a number of selected African industries: c. R. Frank, Jr." "Urban Unemployment and EconOllJic Growth in Atrica", Ox1:~ord Economic Papers (New Series), Vol. 20, No 2, July 0 1968, pp. 250~274. The author's regressions of output growth on employ- ment growth, based on time series covering an average of' 15 years, show that substantial increases in output are necessary in order to compensate for the negative impact of the intercept term in the equations. The "breekeven grow't.h", or growth in Qutput required to keep employment from. falling, ranges from 1.7 percent tor Nigeria's tin industry to 6.7 p~r­ cent for the East Afri,can Railwqs .. - 185 - Africans. 1/ In the four equations rela.ting to non-African private sector employment - (9.6), (9.7), (9.9) and (9.10) - there are only two estimated coefficients whicll are statisticeJ.ly significant at the 95 percent level, namely, the real sectoral GDP coeffieient for Asians in agriculture, and the real salaries coefficient for Europeans in private industry and CQmmerce. Equation (9.6) indicates that with a one percent increase in real sectoral GDP, Asian employment in commercie.1 agriculture 'Would increase by 1.67 percent, salaries held constant; it also shows that With sectoral GDP held constant, Asian employment is very inelastic with respect to reaJ. salary. T.t.~~ est:l,- mated parameters indicate that Asian labor in commercial agricul-t'Urt~ II w'hich probably has an average level of educations.l a.ttainment of as mud) as 6-8 years of primary schooling ~ is far more sensitive to changes in !"r,H~.l ~et~toraJ. GDP -than African labor but not nearly as senei ti ve to changes in rf.:~~11 wages. This seems to be a plau.sible result: more qua.lified la.bor, be.ing a more spe- cia.lized input in the production process, would tend to have a. lower price elasticity and a. higher income elasticity of demand. 14. In this context, equations (9.7) and (9.9), which show estimated parameters insigni.ficantly different trom zero for both sala.ries and sec- toral GDP, and equation (9 ~10), which shows a rather high price elasticity and a rather low in~ome ela.sticity for Europe.ans in industry and commerce, do not seem 'to make much sense. In the equation (9 .10) , it could be argued tha.t the estiDia~e reflects Africanization and independence effects which are highly correlated with salary increases. A large number of Europeans left Kenya in the ten-year period, thereby decreasing European employment in pri- vate industry and commerce. The fall in employment was not the result of in~reasing salaries, but was highly correlated with such increases and pos- sibly even helped eauae them. The high price elasticity (-1.48) of demand is therefore partially a real price elasticity and partially the result Qf non-price factors. 15. The government sector equations reflect I~ s~Jllilar situation in the Asian and European cases - (9.12) and (9.13) - whel~ the independence dummy is not taken into account. In both ~quations, estjLmated real salary coef- ficients are high and sectoral GDP coefficients not significantly different from zero. The effect of Africani zation is particularly relevant in the government sector, and, indeed, introducing the independence dUJnDJy' eliminates the effect ot salary variation in equations (9 .12) and (9.13). The d~ has a small posi ti vecoefticient for Africans and somewhat higher negative coefficients for Europeans and Asians. The results of the government sector demand estimates thus only indicate that the effects of Africanization out- weigh the other independent variables in explaining employment in that sector. - 16. The regression results are con~istent with two hypotheses, coo- cernip.g the relative wage and price elasticities of skilled and unskilled labor, which we have occasion to refer to when estimating elasticities for 1/ Parameters were estimated for data covering total employmen.t of Asians and ~Uropeans in all sectors (excluding sUbsistence agriculture),'overall average salaries'j/ and total GDP. The results were no more inf'ormati ve than the above equations. - 186 - the projections in Sections 4 and 5 ot this chapter. The first ~ypothesis is that unskilled labor is more easily lubsti tutable with other inpu:ts than is skilled labor, i.e., that unskilled labor is more price ..elastic than skilled labor (equations (9.5) and (9.8». The second is that employment of more skilled labor 1s leas sensiti ve to wase changes and more sens·j,ti ve to GDP changes than less skilled labor (equation (9.6». ' 3. The Demand tOi: Labor Derived from Production Function E8timate~. 17 • The Kenyan Goverument' s Manutacturing Surveys fuJ;"llish us with data on the vaJ.ue ot 1966 output ot ],26 manutacturing firms, the replace- ment value of capital in 1963, and the total number ot employees by firm in 1963, 1966, and 1967. TheBe data constitute the elements necessary for es- tablishing a simple production function. In order to obtain a replncement value estimate for 1966, the following formula has been used: (9.l~) where Ni ( 66) :: total employment i11 firm 1 in 1966, N (63) II total employment in firm i in 1963, i .. • replacement value of ~capital in K (63) 1 firm i in 1963, Ki (66) :. estlaated replacement value of capital in firm i in '1966. l8~ A number of stmplit,y!ng assumptions are embodied in this formula. Firstly , it was postulated that ~ach tirm' s cap!tal. stock grew proportionately to its growth of employment petween 1963 and 1966, and that this occurred in the same wtq' astor the manufacturing sector as a whole. Specifically, this amounts to assuming that the growth of cap!tal stock was 23.5 percent above that Qt employment. 1/ Secondly '- the replacement value figure of the 1963 Manufacturing Survey and the capital formation data tor 1964-66 as published in the Economic Survey tor 1968 have been combined to obtain the capital stock growth figure without having sufficient evidence tha~ the two are strictly comparable. It t e. g., the latter are gross rather than net figures, o2~ estimate would imply an overstatement ot the capital stock F,rowth. Thirdl.y, it was n8ces8ary to make the 1963 total employment' figure comparable with the 1964-66 series which apparently v .. based on a different concept. Fourthly, the fact that all' capital and output figures are eXpressed in cur- rent prices leaves open the possibility ot distortions ,stemming from a diver- gent movement ot factor and product prices between 1963 and 1966. !I Source: Calculated from growth of 'employment (Manufacturing Surveys) and capital formation (Economic 'Survey tor 1968). - 187 - a. Simple Labor Component 19. Fitting the 1966 data for these 126 firms to a Cobb-Douglas pro- duction function 1/ yields the following result. log V = -OD9856 + 0.7567 log L + 0.3101 log K (7.794) (6.053) R2 ::: 0 . . 71 where V = the value of 1966 production, L = employment in 1966, and K = the estimated replacement value of capital in 1966. The employment figure in~ludes all levels of skill and all ethnic groups. The production tunction shows that the share of labor in manufactured product is ab()ut 71 percent (76/107) and that of capital. about 29 percent (31/107). 2/ A de~~nd curve for labor can b9 derived from this function by calculating the partial derivative of ~~utactured product with respect to labor: (Xv 0.157 V - : L L where (J1 6 V :II: the producti vi ty of labor. Assuming that productivity equals real wages (W), we get L = 0.151 V. W In other words, the\ demand for labor amounts to roughly three-quarters of the ratio ot \\ total 1roduction \0\ to average real wages. I; I{ b. Weighted L~b,or Crojec,tioDs in line with growth estimates for these two sectors ot the Kenyan economy, ve retain the assump- tions introduced in Section 2 to justifY the use of ethnic-sector groups as proxies for education categories. That is, we assume that equation (9.5) -- demand for male African labor in monetary agriculture -- represents the demand function for male African.s with 0-2 years of schooling, and. that equation (9.8) -- the demand for Africans in private industry and commerce -- can be used to estimate demand for male urban Afrie811s with 3-5 years of schooling. We vill henceforth call the lower education group schooling level one, and the higher, schooling level tvo. 33.. Although in Section 2 we developed demand equations in these two sectors for Asians (equation, (9.6) and (9.9» and Europeans (equations (9.7) and (9.10» as well as Africans t we del) not deal with non-Africans in the projections for two reuons. First, while tor Africans we ~ reasonably assume tha.t a difference in sector ot employment can be used to proxy a difference in level ot education, no 8uch assumptions seem reasonable for non-Africans. Thus, we assumed in Section 2 that Asians in all sectors hnve an average of seven years of education, and Europeans of nine years. ~'here­ fore ~ only in the Afri.can ethnic group do we have demand equations for more than one level of education. Secondly, in any case, African males constitute by far the largest and most important part ot the labor force. 34.. Our first step in using equation (9.8) to make an employment pro- jection for the higher of the two education levels (3-5 years of schooling) requires that 'We fix the value of GDP in private industry ad commerce in 1971~, so that we ~ then deal with possible COMbinations of real wages and employment in this sector in 1974. For this purpose, we give equation (9.8) the following form. 1/ log N2 (1974) ;:: 2.2671 - 0.7408 log (W/P)2(l974) + 0.6648 log GDPl(1974) , , The subscript ~efers to schooling level two (3-5 years of schooling). 35. We project' GDP on the basis of annual growth rates. The real rate of growth of the total monetary sector during 1963-67 averaged about 6.8 per- I? cent, 8.lld that of monetary agriculture, 2.1 percent. 2/ The growth of the . 1/ For convenience's sake, the intercept ter.m in equation (9.8) has been changed from naturaJ. to decimal. log. 2/ EconOlllicSunel t 1968 (Table 1.2, p. 6) shows GDP growth in current market prices equal to 7.6 percent between 1963-67 •... 'Prices increased by about 11 percent during this period, reduc11i.g the rate ot growth to 6.8 percent ,t;;t real teras. - 193 - latter over the longer period 1956-64 was 4.0 percent in current prices, and about 3.6 percent In rea.l tel"1l1S.. While the planned rates ot growi;h for the period 1964-70 of 1.1 percent 11 overall and 6.2 percent tor monetar,r agri- culture would appear too high in the light of the foregoing figures, the 2 .. 1 percent growth rate in agriculture ref'lects partic\u~ly adverse condi- tions during 1964-67. \ole use intermediate figures of 7.0 percent. tor the whole monetary economy, 4.0 percent for monetary agriculture, 7.1 percent i'or private industry and commerce, and 7.5 percent in private in~lu+=l i;:ry a..."1.d government together. 36. The assumed 7.1 percent growth rate of GDP in. private industry and commerce yield..a a 19'r4 GDP of ~235.9 million (in 1.957 Pl":t~~{,9): GDP1974 .. 118.8 (1.071)10 = 235.9 million KE Substi tuting 'I.;hi,s value of GDP1974' equation (9 311) n becomes: log N (1974) ~ 3.84115 - 0.7408 lQg (W/P}2- 2 37. This last equation shaws the empl~nt level in private industry and commerce as a function of the OeTelopment ot real wages in that sector, UD.der the assumption of e. 781 percent annual real growth iii .ector&! output. It is possible to derive from it immediately, tor exuple, the aalar)" impli- cations of a given employment target, or the employment conaequences ot a certain sala.ry policy. Under certain .implityillS uIUllPt1on., Uld in combi- nation with the age-income protiles d18cuaaed ill Chapter V, thia kind ot equation can answer the queat101l: how much doea the equilibriua eaployment level of persons vi th a given level of education change ri th a certain in- crease or decrease in the rate ot return to in'Y8.taellt in that level ot educa:t ion '1 38. Let us assume that, after taking into due ~~ount the risk, non- pecuniary benetits, and externalities (i .. e., ettects which extend beyond the individual. investor) of formal schooling, the Itdeairab1e" social pecuniary rate of return is tound to be 10 percent per year t ad that thi. ia au eq,uili- briUlll rate applying to all levels ot education. The queation can then be posed: what tuture output of the Ichoo1 qste. . .eet.' this condition? 390 'l'he solution can be der.lTed using the to~a below. We are limiting the reasoning to two skill leve18, which cor~8Pond to the two schooling levels tor which we have demand tunCtioD8: level (1) haa 0-2 years of schooling, level (2) has 3-5 years. We aSlume that the difference in 't-ta.ges between two skill levels ot labor yields the benetits of investing in add! tional schooling. The shitt ot wages between 1968 and 191iJ 1s as- sumed to be constant over the whole age-earnings protile ot each ak111 level.gj 'J:! Kenya DeveloPlD;ent Plan, 1966-70, Table 1, p. 83. 2/ The bue year tor calculating wag~ shifts 18 1968, to conform with data trom our Labor Force Survey. \ , Ii - 194 - 40. We start with the general equation which relates the equilibrium value ot the internal rate of return (r ) to costs (e ) and bene:f1ts (b.) associated with the investment: 2 l '3. )\ (( ,l fi le! =- t i=5'+1 If 'We assume that the cost and benefit streams shift over time because of a. shift in the age earnings profiles, the relation becomes: S' n bi + 6. W2t (9.31) 0=- i=l I: L i=8'+1 (1 +%'2)1 where C and b = the expected costs and benefi'cs, respectively, 1 i of taking the additional years or schooling needed to move from skill level 1 to level 2, each level being defined in years ot schooling {earnings foregone are assumed to equal 0 in primary school, so 6 WIt = 0 for 1 5 i~ S I (see Appendix F, paragraph 3). fj, WIt = the shift in the age-income profile ot skill level 1 in time period t, to be assumed constant OTer the entire age range, (8'+1) i n; ~W2t = the corresponding shift in the age-income profile ot skill level 2 in time period t; r = the rate ot return to the additional investment in 2 schooling required to move from skill level 1 to skill level 2; and S' == the number of years of schooling (3) required to move from level 1 to level 2. 41. In order to find the number ot school leavers trom a certain level necessary to f'ulf'ill the condition r2 = 0.10 in, sq, 1974, we must determine the changes in WI and W2 between 1968 and 1974 which meet this condition. In order to reach a single solution, a turther assumption fixing the wage change for one of the two groups has to be made; we choose to work with WI- To fix a solution for Wt, we assume that the substitution pattern represented by equation (9 -5) for agric:.~ultural employment holds for the 0-2 years ot school- ing group, just as we assumed that equation (9.8) represents the demand for .t - 195 - the 3-5 years ot schooling group. And, analogously to our restatement of equation (9.8) as equation (9.29), equation (9.5) becomes: Y (9.32) 42. Continuing the anal8,gous treatment, we apply the 4. 0 perce~t growth rate for agricultural GDP to the 1964 GDP gi of 41.72 million ~ and a..n-ive at a sectoral product ot 61.83 million ~ in 1974 (in 1957 prices). Equation (9.32) thus becomes: log N (1974) 1 = 3.2598 - 1.2301 log (W/P)1(1974) + 1.2044 = 4.4642 - 1.2301 log (W/P)l(l974) 43. Using this form of the equation, the average real wages ot worker. with the lowest skill level can be determined under numerous plausible as- huaptions about the likely development of employment and wages of this group. We consider only two: (i) agricultural wage employment does not increase until the target year (1974), (ii) it increases at the same rate as the population (2.9 percent a year), or, what amounts to rougbly the same thing, it increalea to an extent which leaves the real income level ot that group unaftected t i. e. t accor,ding to (9.5) and our assumption for the growth ot agri- cultural GDP, at 0.6724 x 4.0 percent t: 2.7 percent annually. While the first case resembles most closely tbe past ten years (see Annex Table 9.2), 'JJ the other alternative is probabl1' more acceptable from a social policy and employment point ot new. 44. Using equation (9.32), the average real wases (W ) ot workers with the lowest skill level can be determined for each ot the t.o alternatives: (i) Wl (1974)· ~55.8 (i1) W1 (1974)· KE45.3. 11 As before, the intercept ot equation (9.5) haa been converted trom natural to decimal log. . 21 !)Th1s is the last year tor which the old income series, on which our time series analysis was based, were used. :JJ The real annual wage for level (1) (agriculture) was 45.3 Kli (1957 prices) in 1964. We assume thi.real wage remains constant through 1968. Real wages in agriculture did not change between 1964 and 1966. - 196 - The corresponding wage changes over the ten-year period 1964-1974, would hence be: Y (1) 6. Wl ( '64-' 74) lIS Ia:lO.5 (11) 6w1 ('61j_'11j) • KJ: 0.0 This represents a 23 percent increue in real. wages over the!' ·1968 vage in the first case, and a zero percent increase in the second case,.in which the labor force in skill level (1) incre.. e. at a rate of 2.7 percent 80- nuaJ.ly. We can now apply these percentages tb" the average observed monthly' income ot the urban labor torce with 0-2 years ot education· (Ani1&X Table 5.2) • We then:tind an absolute ch8llge in wages ot the latter betWeen 1968 and 1974: (i) 6 Wl('68-'7~) • 320 Kah/JIOIlth t1m~s 0.23 - Klli4.4,/year (11) 6W l ('68-'74) • Km. 45. Using the data on the co.'t. ad benetita ot inftst1'Dg; in Standards III, IV, and V (3-5 years ot priJU.rT school) trom Table 5.1 t COltUlln one (Unadjusted age-1ncoJle protiles) .. T&l.ue tor 0i and bi in equftton (9.31), letting r2 == 0.10, ad taking the 1968-191" wage change valuea trc. above, we get the following alternative solutions tor the absolute chadge in wages between 1968 and 1914 tor the higher skill level: 2/ (i) 6. W2 .. -81j~:¥1 + 41j.1j = -124.2 + 44.4 • -Kt79 .8/year -124.2 +0 • -KZ124.2/year. Thus, under the assumption ot a cODstant level (1) employ1lient tigure, by 1974 '! wagea ot level (2) skills (W2) wOUld haTe to tall by 019,.8, or 22.5 percent of mean income (see Annex Table 5.2), in order for the UJladJusted social rate ot return to investing in the third, ·fourth and fifth yeara of schooling for urban male Africans (equal to 16.4 percent per ennUJliin 1968) to fall to 10 percent per annum. Similarly, in the second alternative, where level (1) employment in~reue. between 1968 ud 1974, W2 anuat tall bt KZl24.2, or 35 percent Qf mean income, tor r2 to equal 10 percent in 1974. 1/ In order to mesh the base year ot ~ Wl with the year ot the L~or Force Survey t 1968 was chosen ae the bae. Theretore, the cOmputed . WI is calculated by taking the ditference between (W/P)1974'~d. (W/P, 1968, assuming, tor the lIfaent, that the latter equals reaJ.vligffs ..tor Africans in wage· agriculture in 1964 (1951 prices). gj See Appendix F tor ecaplete solution tor (j, W2' - 197 - 46~ In order to estimate from equation 9.29 the increase in level (2) employment which these decreases in real income represent, we apply the percentage decreases to average wages of Africans in private industl~ and commerce. In 1964, the average wage (in 1957 prices) in private indus-crJ and commerce for Africans "ras KEl12. 3 annually. Employment. in 1964 cor- l'Oected to the 1963 measurement of employment 'Was 129,200. Assume that the ::ea.l wage decrease described in the previosu paragraph can be extra:pola.ted lirlearly to 196h. '.rhus, the 22.5 percent decrease in wages betwetH~, .~ ;;[)3 and 19'71~ becomes a 37 percent decrea.se between 1964 and 19''{4 and th~ 3;; ~O percent. deCl'ease becomes a 58 percent decrease between 1961l and 191'4,) lror the t'wo alterna.tives these tr811s1ate into (l957 prices): (i) 6 W2 = -41.6 KJj/year and (ii) 6w2 = -65.1 KE year, and (9 .. 29) becomes~ for the two aJ..ternatives: (i) log N2 (1974) ::::: 3. 8445 - 0.7408 log (70.7) = 2 .l~745 N2 (1974) = 298,200 and (ii) log N2 (l974) = 3.8445 - 0.7408 log (47.2) N2 (l974) = 402,300 tl~eae figures indicate tha.t if the social rate of return to i"nvestment in inco~plete primary schooling were to attain 10 percent per annum, real in- comes of thoBe with 3-5 years of schooling would, have to fall and employment 'V.tould have to increase sharply. 47. If the social costs and benefits adjusted for socio-economic 'val-iablee ('fa.ble 5.1, column (2» are used instead of the unadjusted pro- files, 6, W2 and N2 are essentially unchanged. Since the social rate of re- turn to the third, fourth, and fifth years of schooling is almost the same when. ca.lculated on the baa:i.s of adjusted rather then unadjusted income streams (16.7 versus 16 .1~ percent), the real wage reduction necessa.ry to bring the rate to 10 percent is only slightly greater. By the same token, employment of those with 3-5 years of schooling is also about the same, b. !:!tplol!!lent, a.~d W~~ Cha.n6e~.2 Labor SU}?Pg ExogenouslY Detenn:i.ned 48. In Part (2) the discussion of employment and real wages in the monetary economy assumed that the future ~upply of manpower with different educationaJ. qualifications is e. function of future wages and rates of return" In this section, supply projections as estima.ted in Chapter VIII are used as the point of' departure, and the wage implications of' alternative employ....; ment targets fo;r primary s chaol tra.ined labor are analyzed. - 198 - 49. According to the supply projections of Chapter VIII, there will be a. gross increase of 1.7 million in the supply of labor with I-T yeart:; of schooling between 1966 and 1974" African employment in, the monetary sector of agriculture and private industry and COn~erce totalled. ~77,4oo in 1966. The a.verage education of this group is probably very close to the a.verage educa.tion of the African male population, i.e., between 2.0 and 2 .. 5 yea~~s of schooling. 50. If' we also assume a working life of 33 yea.rs, i. e. an annual de- crease of 3 percent in the labor stock, the residual stock in 1974 is 286,800, a decrease of 90,600. 'l.'heretore, if it is desired to employ a.ll new entrants into the monetary sector at this level, more than 1.6 million new jobs will be needed. Even under the extreme aSsUDlption that "rage employmen:t!. will 'be sought by males only, the increase in the labor supply of approximately 0.8 million can not be absorbed in the monetary sector by any plausi'Ql:e, com.bi- na.tion of increased output and decrea.sed 'Wages, as the pre.ceding s.e.ction makes clear. 51. A more realiatic employment assumption for the 1966-1974 period is that only about 20 percent of males with various amounts of primBr.y.' education find work in the monetary sector (both agricUlture and private industry and commerce). ~~ere ~ be reasons Yh~ males with less schooling ma~'be em- ployed in ~riculture, where physical: ab1li ty probably counts for' more than addi tiona.,l schooling, but probably' the maJor! ty of males employed. will have comple be Pl"imary education. The remaining available men with 7 or. fewer years of schooling are either voluntarily or involuntarily unemployed, or find 'Work in the SUbsistence sector. If the rate of growth of out.put in the pr.ivate monetary sector is set equal to 6 percent, (weighted averaGe of 4.0 and 7.1 percent) and the male labor force employed in 1974 equal ttt 502,000, we find that average real wages in 1974 (1957 prices) 'Would have to be !r,i.ues.., assuming that the structure of real wages of the variously qualified teachers does not shift, and that teachers with only primary schooling are not com- pletely phased out of teaching. This m.eans that abOl..lt 550 ,000 A.fl~icalJ.s with 7 or fewer years of schooling will be employed in commercial agriculture, private industry and commerce, and as primary school teachers, if real wages in the former two ,stay constant, an.d real wages of primary school teachers relative to more highly trained teachers stay about the same as in 1967. The estimate also assumes that all Africans employed. in commercial agriculture and private industry and commerce fall into the primary schooling category. 53. If we further assume that the GDP elasticity of employment in gov- ernment of Africans with primary schooling equals unity; tha.t the wage elas ..• ticityequals zero (i.e., that average government wages do not change), and that all high- and middle-level African manpower in the 1964 manpower survey data are employed by the government, then approximately 97,000 non-teaching Af'ricans with primary schooling were employed by the government in 1964. 2/ If '\fe project this figure to 1974 by means of a 6 percent growth rate, the employment of those with complete or incomplete primary schooling in 1974 in government riser; to 180,000 and tota.l employment in all sectors rises to 732,000, of Which 340,000 represent the addition to the labor force between 1/ The average, salary we use here is an avera.ge of the salaries given for the P3, p4, and KPE categories of primary teachers in Table 3.13, each weighted according to its share in the total number of P3, P4, and KPE teachers in 1967. Projections of teacher requirements are given in Appendix G. 2/ ~~e fi~lre we use is a residual. Government employment totalled 173,700 in 1964. It is assumed that the 12,800 Asia.ns and Europeans in public service in that year all had more than primary schooling. 'lbe 1964 man- power survey shows 70,800 high- and middle-level manpower in Kenya out- s j. de of agriculture. If we subtract from. thi s figure the 12,800 non- Africans in public aervicep1uB 25 percent of the 28,000 Asians and 10,100 ];urope(,Uls in privateindustry and conunerce, we arrive a.t 41, 000 with more than primary schooling who are Africans. This is probably some,\ihat of an overestimate, since skilled manual la.bor and technical and semi-professional labor is all a.ssumed to have more than primary schooling. If' all 41,000 were employed 'by the government, this leaves a.bout 120,000 Africans with 0 -7 years of schooling employed by the government in 1961~, 23,000 of which are primary school teachers, leaving 97,000 non-teachers. - 200 - f 1966 and 1974, including replacement of retirees. Even if we assume;, tha.t only one-half of those becoming available at this level (1.1 million:) in the eight-year period vish to enter the monetary labor force (850,000), about one-half million people will not find jobs in the monetary segment of' the economy Ullless real wagearbll rather sharply, an unlikely occurre.nce. 5 • The Future Labor Market for African Urban Males with Second!n S,chooling a. Equ111bri um SUPEly ot Labor-I-- Rate of Return to Schooling Equal to 10 Percent 54. As in the case ot prim.a.ry schooling, we ask what ~ecre.ase· in wages is required to reduce the rate of ret.urn to secondary schooling: to. 10 percent annuaJ.ly. According to the Labor Force Survey, average earni!l8i~. inti. 1968 for employees with seconda.ry schooling were about RE400 annually !{ andL for those with completed primary schooling, ~40 annually (see Annex Tab.le 5'.~~, 7-AJ.l category) • Costs are estimated from Tables 6.1 and 6.2. Using the; simplified rate of return f'ormula, which assumes that the average income dift.~.r.ence is a perpetuity .and that cost is incurred at one point in time (c c y/r), we find that the social rate of return to' 10.5 years of schooling, adjJ,lsted for socio-economic variables, is equal to: (9.33) r = J..- =; 400-240 ~8 II: 0.33 c . 0 or 33 percent, in 1968. This is an overestimate ot the rate of re:t.urn, as can be seen from a comparison with the more detailed discount fOl'Jllula (equa- tion (9.30». 55. As the a.verage real. income of persons with secondary e.duc:ation tends to fall towards the full employment equilibrium of Klb5 (s.ee. below), tnere will be substitution of secondar,J'. school leavers for persona.: with only primary education. This vill tend to hold up the wages ot se.condary- trained persons and lower the wages of priJIary leavers. If' av:eragei~ real wages of primary school lesvers 'lalla bY', s8¥, 25 percent, to KZl80. annually, income foregone would be decreased b~ 2~ percent to ~135 and total. costs to about 0435. This would mean that real wases ot Africans with secondary education should fall to about KE223 UIlually in order to lower. the;, rate of return on inves.tment in secondary schooling to 10 percent. As we show below, even at this real wage, a considerable number ot persons wi~h se.condary ; schooling could still remain unemployed. 56. Since no sui table estimates are ava:11able for the relatfonships among the employment of African school leavera from post-pr~ary' r~vels, 1/ The ~4oo figure is arrived at by assuming the average level ot school- ing in the 8-13 group to be 10.5 years -- which is also the aerage in the sample - .... and the average salar.y of those with 10.5 years ot. school- ing as Ksh 660 monthly, which is slightlY less than three-fourtlls of the difference between the average s~ary ot those with 9 years: of: s.!chooling and those with 11 years (see Anne~' Table 5.11:,). - 201 - their wages, or growth of sectoral output (especially in the ILon-agriculttU"al sectors), an attempt is made here to approximate these relationships and to assess from. them (i) the employment effect of a ,~onstant wage assumption for employees with secondary school training, and (ii) the Yage effect of a full employment assumption f.vr secondary school output between 1968 and J.974. 57 • We can reasonably expect that the wage elastic! ty of emplf';~nY.lent of both secondary school and universi ty leavere is lower than that of' primary leavers, and the GDP elasticity higher, as discussed in paragraph 16 of this chapter. We make the assumptions, probably' realistic ones, that the :p:!f-:lce. (wage) elasticity of employment for those with secondary school education equals -0.5 and that the income (GDP) elasticity of employment tor this group is 1.5. Our purpose here is to delOOnstrate this particular line of approach to the problem of demand projections for dif~erent labor categories~ Thus, log Ns = A - 0.5 log (W/P) + 1.5 log GDP(p+G). 58. We start from a 1964 figure for GDP outside agriculture (i.e., priva.te industry, commerce, and government) of Km.46.5 million in 1957 prices, (see Annex Table 9.3), and an average income of African employees with secondary school of about n400 annually 'in 1968 (see paragraph 54) f which corresponds to about ~320 in 1957 prices. Y If real wages are assumed constant between 1964 and 1968, 1964 wages in 1957 prices are KE320. 59. To solve for A in equation (9.34), we need estimates of secondary level employment in 1964. The 1964 manpower survey suggests that there were about 41,000 Africans with middle and higher levels of schooling (see paragraph 53, footnote 2). The number ot Ah'icaDS in Kenya who had uni- versity training was about 1,000 in 1961. Y It we presume that another 500 Africans underwent university training between 1961-l96~ (see Table 3.12), we arrive at a very crude estimate ot 1,300 Africans (excluding 300 teachers -- again a rough estimate) with uni v'erai ty training in 1964, and of 40,000 Africans with some secondary education. 60 u The data in Table 3.3 show that in i~he same year there were 4,800 teachers with some seconda.ry training teachinE~ in primary schoo1s(P2 + Pl + 81 + ese + :KSC teachers) of whom a maximum ot about 4,000 may ha"i'e been .llf'ricans. 'From Tables 3.9 and 3.10, we deduc.~ in a similar way that about 25 percent ot the 500 secondary school teache:lC's with secondary train:i.ng in 1964 :;fere Africans. The total secondary-i!railaed African 'teaching force in 196.i.'was thus ot the order of 4,000, which 1e.ves 36,000 secondary-trained Y The cost of living index (1957 II: 100) was approximately 125 in 1968 and l12.5in 1964 (see Annex Table 9.3). gj See Guy Hunter, Education fQr a Developing Regio~, London: George Allen & Unwin, 1963, p. 58. - 202 - Africans in jobs other than teaching. Using the GDP, wage and emR~oyment figures for 1964, we can solve for A in equation (9.34): A I: -0.4401 61. Applying the 7.5 percent target growth rate for GDP in p.ri vate industry and government combined of the 1966-70 Development Plan (\see paragraph 35), leads to a 1974 sectoral product of KE293.3 mi Ia..i on· (in 1957 prices). Consequently, (9.34) becomes: log NS (1914) = -0.4401 - 0.5 log (W/P) + 1.5 log (293.3) Assuming that real wages do not change, between 1968 and 1914 (Kf:32Q in 1951 prices), secondary-trained employment in non-teaching j obs est.imat'e~d from equation (9.34) would be slightly Over 100,000 in 1914. In addlt::ton, Ap- pendix G shows that 13,000 secondary Bchool trained teache.rs· will: be employed in 1974 if relative wages of teachers, remain about the same as i'n the period 1960-67. Y About 11,500 of them will.: be A:tricans. Total. emploYilRnt of Africans with secondary schooling is thus estimated to be about 114,000 if real wages remain the same. b• Em1210yment and Wage Changes". Labor SuppJ..y Exose:nously De.termined 62" We now look at the output trOll secondary schools.. Accor.ding to the projections of Chapter VIII, &bout 280,000 indi viduaJ.s from all ethnic groups with any amount of secondary schooling will becomel available for employtrtent between 1965 and 1914. The 1964 manpower surv1ey shows 64,500 employees in the high- and middle-level manpower category who- are: below the professional and top management level. If we assume that they all! have at least secondary schooling and that 2.5 percent retire 8l1Jl\18l.ly' trom the labor torce" a stock ot 48,400 will. remain in 1974. The 1~ota.l available labor force with secondary schooling' in 1974 vill thus be 328,000.• 63. We then assume that employment ot nOD-Afric:'~l.~s with, se,condary schooling in public service and private industry and commerce will remain constant at 30,OQQ between 1964 and 1974 (which is probably an overestimate) and subtract this figure from the total available labor torce, which gives' us a stock of secon'dary-trained African males ot about 300,000'. 64. We projected that 13,000 'secondary school trained teachers will be employed in secondary and primar,y school in 191~., If we assume that 2.5 percent ot the 1961 teaching force of 7,000 will retire each year, then 1,400 will have lett the profession by 1974, and 7,400 new teachers (1914 projections minus the remainder ot the 1967 atock) will be needed, all of which, it is assumed, will be Africans. Subtracting this figure. from the above total. of 300,000 leaves &bout 293,000 persons for whom jobs outside teaching would have to be found in- order to secure full employment of secondary school trained Africans in 1914. Y This figure does not include teachers used for teacher training. - 203 - 65. Under the 'elasticity assumption on whieh equation (9.34) is based, rea.l annual wages ot Africans with' secondary education will have to 1"&11 to 1 ~5 (in 1957 pric~s) to assure full employment. This wage would be lower ! than average real. wages of employees with primary sehooling, if real wages of the latter did not fall. We would consequently expec·t that employers would substitute labor with secondary schooling for labor with primary schooling if relative wages of the former fall. Total 1974 primary school employment in non-agricultural sectors, however, will be only 530,000, even if 'lire assume that the figure of 200,000 employed in monetary agriculture in 1966 stayed. constant until 1974. Full employment of the secondary school output of tht~ years 1968-1974 would mean that a substantial portion of p:r'irn~~ry school employment outside agriculture would have to be replaced by secondary school trained labor. 66. Alternatively, if real salaries of Africans yith secondary school- ing stayed constant at KE320, only 102,000 of the 293,000 seeking employment outside tea.ching couid be a.ccommodated under the assumed conditions. Clearly, our elasticity assumptions may be wrong. However, to guarantee full employ- ment for this category of labor under the hypo'theses ot constant real wages and a wage elasticity of 0.5, GDP elasticity would have to be about 2.0 -- i.e., for every one percent increase in GDP, secondary 'employment would rise by 2 percent -- a rather extreme pattern. If wage e1as~ici ty 'were claimed to be higher, GDP elasticity would have to be adjusted downwards to reach full employment. 6. The Future Labor Market for African Urban Males with University Training a.. Equilibrium Supply of Labor;, Rate of Return to Schooling Equal to 10 Percent 67 • We can make the same type of estimate as we do for secondary schooling in order to determine the change in wages nee'ded to reach a rate of retiurn of 10 percent annually" for investment in higher education. As- suming the average amount o~ schooling of university-trained people to be 15 yeELrS, the average cost of the two years of uni versi ty education to be ICI:3, 400, and the average difference in salaries between Africans with and without these two years of schooling to be K£340 annually (K~l,OOO minus lCt660 II see Annex Table 5.4), the estimated rate of return following the perpetuity formula 1s 10 percent. No change in wages is thus necessary to reach the target rate of return. 68. To project the demand for labor with university training, we assume that the wage elasticity of emplo.yment for university-trained persona is equal to that of those with secondary schQoling, -0.5, and that the income elasticity of employment is 2.5, higher than the corre8ponding elasticity for se~ondary level employment. As in the equations tor the primary and secondary level labor, the estimate of the demand function for university- tr~ined persons does not include an approximation for the elasticity of substitution with labor from lower education categories (which may be rela- tively high). The data available do not even give ,a clue as to the order 204 - . · ,"" .\ :' of magnitude for this elasticity. The simplified d.emand equa.tton for those \"\ wi thuniversi ty training, It then", is the followip.g: log Nu = At - 0.5 log (W/p) + 2.5 log GDP(P + G) 69. As in equation (9.34), 1964 GDP out$ide agriculture ,equals KE146. 5 million, and the average 1968 income of employees with universityenucatioD, . e:\'0ut KEaOO 11 (bot~ ~in 1957 prices). Assuming that ~eaJ. salaries have not C~~ged appreciably ~etween 1964 and 1968, the 1968 fl.gure, can be used for th~ 1964 demand equation. ; 70. We estimated in paragraph', 59 'above that there were about 1,300 Africans with universi ty traj.ning Qutside of ,te'aching in 1964. Usi'ng the employment, real earnings, and GDP figures for that year, we can solve for At . in equation (9.35')::r At = -3.8491 Introducing the 7.5 percent 'target growth of GDP in private industry and government ( as in paz:agraph 60), (9.35) becomes: log Nu (1974) = -3.8491 - 0.5 log (W/p) + 2.5 lO~, (293.3) 71. What will .. the demand for "Wl1vers! ty trained labor bf! if' we assume that real wages do not change betwe~n 1968 and 1974 (i.e. remai:nconstant at ~8oo in 1957 prices)? Employment estimated in non-teaching jobs in 1974 from equation (9.35) would be 7,300 in 1974. In act~1t1on, Appendix G shows a. total demand by the school system tor about 3,800 graduat~ teachers in 1974. If we aSsume that 50" of the 200 or so African grad\late teachers in Kenya in 196'4 will have retiredibJ' 1974, new demand for all graduate teachers will be about 3,700. If th, llon-A\f'rican graduate teaching force remained at the 1967 level of abQut i, 100, this would leave about 2,000 new teaching positions for African graduates (out ot a total of 2,,100). If real wages remained constant and the structure of wages for teachers behaved as assumed in Appendix a, total demand f9r university-trained Africans in 1974 would equaJ. 9,400. b. .Emp10plent and Wye Chans~~" ~abor Supp,l.y Ex,ogenously Determined 72. According to the Manpower Survey of 1964, 6,300 ,employees had' university training. Fifteen thousand of these were Africans: 200 teachers and 1,300 other employees. If the employment of non-Africans with university training remained const'ant until 1974, 4,800 non-Africans 'Would be employed in that year. 11 The average income in 1968 of those with university tra.ining (Klil,OOO) .is taken as a simple averl£ge of the mean salary of Atricansin the sample with 13 and those with 17 years of school (see Annex Tables 5.7 and 5.8). - 205 - 73. Accor4ing to the supply projections of Chapter VIII, about 7,800 university-trained persons will meanwhile become available, a high percentage of whom will be Africans. Assume that the latter's ahare were 75 percent, and that the 25 percent of non-Africans replaced leaving or retiring non- Af.ricans in the labor force. In 1974, there would thus be 5,800 Africans with university training available tor the labor force. Since we estimated in the previouB paragraph that som.e 2,000 Africans would be employed a.s tea,chers, the remaining 3,800 African graduates would be available f('>:'~ non- teaching jobs. Adding the 1,100 or so university-trained Africans Temaining from the 1968 stock outside teaching (1,300 minus 200 retirees between 1968 and 1974, assuming a 40 year working life), we arrive at a tota.l of J~ ~ 8no " 74. We estima.ted above, however, that the demand for univers:f.ty-trained Africans outside teaching, if their real wages remained cons'tan't!J 'Would be 7,300 in 1974, which implies a substantieJ. shortage at this level in 1974, given the supply figure of 4,800. In order to reach equilibrium at an employ- ment level of 4,800, real wages of Africans would have to rise to Klil,890 (in 1957 prices), or to almost twice the present level. 1/ 75. Like the secondary school demand estimates for graduates, the university approximations are very sensitive to changes in the GDP elasti- city of employment. ~'hus, an assumption of a 2.0 GDP elasticity and a 0.5 wage elasticity for university-trained persons yields an employment figure of 5,200 in non-teaching jobs at constant average wages, a m.uch lower figure than the 7,300 estimate with a GDP elasticity 01' 2.5. Similarly, 8. 2.0 as- sumption would imply a rise in real wages from ~800 to KE&43 in order to balance supply with demand outside the teaching profession. 76. Unlike the secondary school demand estimates, the university ap- proximation is very sensitive to the projections of teacher demand, and thus indirectly very sensitive to the projection ot anrollments in secondary education. If the figure ot 151,000 pupils in 1974 on which the teacher demand estimate in Appendix G were on the low side, the wages of university-- trained teachers would tend to rise even more sharply between 1968 and 1974. It, tor example t the rate of grolith of" secondary school enrollments in 1967-74 equaled the rate observed from 1960-67, there would be about 300,000 secondary school stUdents enrolled in 1974, and the demand for graduate teachers , given the assumptioDs ot Appendix Gt would equal almost 7,500 in that year, of whom 5,800 would be Africans. This .eans that only 1,000 ".. Y It should be noted, however, that the teacher proj.ections we use a.re 1mplicitly based on current relative wages, conceding only small. in- creases or decreases in relatiTe wages If lIowever, if wages in the non- teacher labor market rose by very mucb, either graduate teacher wages would also rise, inducing schools to lower the graduate teacher to per student ratio, or teachers would leaY'e teaching to take advantage of relatively higher wages ;l.n the non-teaching jobs. In the long run, this would tend to reduce wages outside 'teoohing. - 206 - Africans woul.d be available for non-teaching jobs. Even wi.th a GDP elasti- city of demand for university-trained persons equal to 2.0, average annual wagesot that group would have to rise to abo'~t Kr..23,OOO tQ baJ.ance demand and supply. ' 7. Results and Conclusions " 77. The methodology developed for educational planning in this study uses estimates of cUrrent weges of labor vith different amounts of formal schooling and current rates of return to investment in additional schooling as a base from which to project alternative :fUture wage and employment patterns. Two types of estim.ates are made. in section (e.): the caleul.ation of equilibri'llm. supply of labor with different amounts of schooling, assuming that the supply ia a function of rates of return to schooling, and that tbe desired level of rates ot return to all levels of schooliDg is 10 percent, and in section (b): estimates of' the real wage level consistent with full employment in 197~ for a g,i ven ~chool output between 1968 and 1974, as- suming that the supply of skills ....i8 inde- pendent of the rates of return, and, alternatively, surpluses or shortages of labor of ditterent educational qUalifi- catioDs given a cert.in level of real wages (this approach resembles customary manpower requirements calculations). , .. 78. Tbe results of thele 8sti,matel, while admittedly crude, demonstrate that projections of the ~e~d .for the.outputs of a school system are ver,y sensitive to cbanges in wages •. In tUrn, it wages are Dot flexible, rapidly increasing output -'trom schools c;an leave large numbers ot educated persons with.out employment. Even if::wag~s are flexible, it is possible that de- clines in real wages ~ have to be unrealistically large in order to achieve full employment; Unemployment of trained manpower could t under such cir- cumstances, be an inescapable result of ,rapid expansion of the present structure of the education system ~n a developing country. 79. Our estimates for KenYa'shOW that BlOSt of those Africans taking primary schooling in the next five years will by necessity remdn either unemployed or in SUbsistence tar.ming, even it a rapid fall in real wages were to come about. The rapid growth ot the number of" Africans with second.... ary schooling between 1968 and 1974 would tend either to lower considerablY their real income or to create 8ubstantial unemployment among secondary school leavers. cBecause of the large increase in secondary- school enrollment and the resulting strong demand tor university-trained Atr1cans to fill . teaching posts, ft seems possible tha.t tbere will be upw8.l',d pres8ure on the wages ot Afric8Dswith university training during this or an even longer period. - 207 - so. As a result of the downward pressure on the wages of secondary school leavers and the upward pressure on wages of university graduates, there is likely to be an increased recruitment of labor with secondarJ schooling both into jobs normally held by those with universi ty training (especially teaching posts) and, as we have noted before, into jobs now held by primary school trained people. In other words, the demand ana. supply situation would lead to an upgrading of many industry jobs l't.::G.v.iring a fairly low level of training, and a downgrading of other jobs (pr(uj~m,shlY largely in the public sector) requiring a high level of trainin~~ ~11~m, the eftects of the downward pressure on secondary wages ~ be partly dampened by the effects of a further downward pressure on primary ruchool leavers I wages, on the other hand, and of an easing of the upwa.rd p:reastlre on wages of university graduates on the other. 81. f~is downward pressure on the wages of those with primar,y schooling will tend to reduce present low rates of return to that level even further, giving it, from the point ot view ot society, essentially the charac·ter of a tlmeri t want" beyond purely economic justification. The rate of return to secondary schooling would initially fell, as the alternative wages to be expected when leaving school during or at the end of the primary c,ycle tend to fall, but the fall would be slow because the rate i8 supported by the substitution process described above. 82. After 1974, a continued rapid growth in secondary school outputs might result in a sharp decline in the corresponding wiges, and also pos- sibly in substantial unemployment at this level, causing the return to in- vestment to decline sharply. At a later stage, this de'cline might be re- inforced by the increased supply of university graduates and a reversal of the substitution process described above. The rate of return to investment in university training, however, is not likely to fall even after 1974, if' our forecast of a marked shortage of university-trained persons by that year is correct. Only it the universities' output were to catch up with and surpass the fUture demand for this category, would the rate of return to university training tall. 83. Whatever the outcome ot these speculations m.8\Y be, it should be clear trOJll this study that the state ot our knowledge about the :f'\mctioning ot labor markets is most unsatisfactory. The theoretical framework we have used here is not matched by an adequate amount of intormation. and we are thus unable to test, reformulate, and improve our tools. It thu8 appears that the procurement of relevant data should rank high among the tasks of education and manpower economiat., particularly since the turn ot events in the labor market ot an increasing number ot countries is drawing edu- cational planning away from the production of educated manpower and tow8.t'ds the problem of effective manpower deployment. ,.~ Table 9.2: Kenya: Growth Alternatives! 1966-1914: Social Rate of. Return, Wages and Employment for Two Labor Force Levels For Africans . ~ (1) (2) (3) (4) (5) 1968 Social Desired 1974 Change in Level 2 Alternatives Rate of Return Equilibrium Social Labor Force Average 1914 Level 2 Level 2 Employ- to Years 3-5 Rate to Years 3-5 wage to Attain (2) Average vage ment- Implied by (4) .. (Percent) (Percent) . (~erc~~8 Ctt~~j Over (IG,) ( thousands) U§!ng liladiuste~ates to find ~ and W2 f£ Alternativel 16.4 10.0 -22.5 70.7 298.2 . 13 Alternative ~ .~ 16.4 10.0 35.0 47.2 402.3 :.- UsiIlg Ad.1usteAa~~ .. _~ .. ~ Q) to :rind ~ and W ,,', . 2 A1ternati~ 1l2. 16.7 10.0 -23.0 69.3 302.6 /3 Alternative 2- 16.7 10.0 -35.5 45.8 4ll.) .... ~ '-·1 tl:. For socio-economic characteristics. ~.,- -- ~ Level 1 Labor Force does ~ grow 1968-1974. t a.. Level 1 Labor Force grows by approximately 3.percant annually 1968..1974. ~ • Table 9.): Kenya: Growth Alternativesl 1966':'74: vages and Employment £or Africans "ith Primary and Secondary School and Univer si ty Training Assumed Real Net Addition to Total New New Unelll- Africans in 1974 Unfilled Wage, Ia./Year Total Available Employment Jobs pI oyment Jobs 1974 (1957 prices.) Labor Fbrce 1968-74 1974 1968-74 1974 1974 With Primary School Trainin~ &netary sector, agri., ind. and comm. 78 502,000 Prima.ry school teachers 136 50,000 Government (non-teachers) no change, 1968-74 180,000 Total 1 1 7QO,cxx:I.l 732,000 :'" 340,000 500,000&. 0 /2 L "'~th Secondary School Training- 4 .. i'~~ I\) ~~ ..... '4,",-f~ <.. _ ExclUding teachers ~ .... ,~. (a) If real ~ge does not change (b) If full employment is desired 400 65 ( (. a80~JO) ) l02,000 293,00 Teachers Ba,'1le relative 7,oo0L2 13,000 7,000 wages as in pas~. Total H'~ 287 , pOO f 96 000) .. *' .,,~ .. " 287:000) WithUniver sity Traj ningll ExclUding teachers (a) If real wage do esnot ch~e 800 ( ) (b) IT £ul1 employment is desired 3,800) 7,300 6,)00 2,500 1,890 ( 4,800 3,800 Teachers 2;000 2,100 2,000 Total 5,800 ( 8,300) ( 5)800) - 210 - x.. ~ OUTPOtt-LE OF THE S~IDY 1. In looking back on the preceding chapters, the authors ca~ll\o·t 1l~ l:rf wondering whether the range of topics covered ill this stu.dy has: :not be€:~l $0 broad that the major reeults "become difficult to see clearly and. to :rela.te one to another. For the benefit of those readers whose main intere~t is in the practical uaef'Ulness rather than me'bhodologicel de'tails otthe eost- benefit approach, our experience has been put into a more conoise form in this concluding chapter. 1. ~'he S~cific Results a. The Profltabilltl of. Educatio~ i~.!f~&a 2. To recapitulate the line of our analysis 9 we set out by assuming that ad
  • etour Into Education Production Functions 20. T~~ development of earnings profiles by age and education group and the subsequent rate-ot-return analysis treated educational achievement 1/ It might be argued that SUbsistence farms are operating below capacity, and that the availability ot a highly desirable consumption good such as primary schoolIng m~ encour~e the rural family to increase its tarm output for the express purpose ot p~ing tor the good. ~hus, increased conswaption may actua.1ly be met by an increase in output, rather than inducing a shift from savings into consumption. The latter effect m~ also come about by an autonomous reduction in the demand for educatioll. The emerging of such a cobweb-effect cannot be ruled out completely', if indeed the .demand is dominated by economic considerations. - 216 - as an autonomous element, detached. from the education process that fosters it. However, the average cost-per-pupil figure entering the cos·t;-benefit analysis lumps together the coat of' a.ll inputs for a certain layer of the school system, and there is no reason to assume that all or any ·of the mixes possible within a given average per-pupil coat are strictly equivalent from a eost-benetit point of view Our analysis thus sh.ifts to the micro-ed.u- 0 cational level. W',e investigate the contribution of a number of' identifiable schooling inputs <d, once this contribution 115 expressed in monetary (add1N- tiona! earnings) terms, deter.mine partial rates ot return to selected inputs (see Chapter VII). 21. By and large, the results reflect the paucity ot our data. They cast doubt on, rather than support, some commonplace assumptions about the role of certain schooling variables in the education process (the teacher- pupil ratio, for example) and point indirectly to the import811ce of e:7.:tl"a,- education variables.. The rates of return to various schooling inputs are small to moderate (lea8 than 1 to 6-12 percent), in spite of numerous built- in upward biases. The only notable exception is expenditure to increase the percent of secondar.y school pupils who board, which seems ta entail a sizeable p~off (see T.ble 7.2). c. Future A1ternatives : The l!wloY!l!nt OutlQok for Educate,.g. Manpower ~n ~e~ 22.. Up throug.b Chapter VII, the analysis i8 in terms of the present situation (average). Since the education planners' i~terest lies in the future deviations from it (marginsl), Chapters VIII and IX present labor market projections to 1974. 23. The es't;imates of the :fUture demand for various levels of education in Chapter VIII are 8~ple extrapolations of time series, more elaborate projection methods haVing bee~ abandoned tor lack of adequate data. They cover the years 1969-1915; enrollments estimates for the last year of this period are roughly 1.5 million pupils in primary education, more than 150,000 in secondary education, and somewhat less than 15,000 in post-secondary institutes. 24. In Chapter IX, Y the asse••ment of the fUture demand for educated manpower starts with a regression an$lysis of time aeries on employment of the country's three main ethnic gl'C:>Ups (Africans, Asians, Europeans) in the three major sectors of the monetary econoll)" (coJDlllercial agriculture, industry and services, and government). Employment in each sector is analyzed as a function of average .real wages and output in the sector. Two of the demand 11 We mention in passing an attempt to develop Cobb-Douglas production functions' with the various kinds of labor subject to a constant elasti- city ot substitution condition. Unfortunately, the implementation of this apparently acceptable theoretical approach is toiled by data short- comings (see Chapter IX, Section 3). - 217 - equations tor ethnic/sector groups which we obtain, those tor Africans in agriculture and Africans in private industry and cODDllerce, are then put to use to project demand for the two education groups with which we teel they correspond, namely, Africans with 1-2 years of schooling, and Africans with 3-5 years, respectively (Section 4). Equations ot similar form are con- structed for higher levels ot education (secondary school and university, Sections 5 end 6 respectively) on the assumption that demand tor higher levels of skill is characterized by a relatively low price (wage) elaediici'ty and a rather high income (output) elasticity. Gi ven a certain real l"!o.t.e of sectoraJ.. growth, we examine: (i) the employment implications ot &. certain salary policy, (ii) the salary implications of a certain employment policy, (iii) the employment and salary conditions which would leeA to a certain (desired) level ot the rate of return to education. 25. The results can be summed up as tollows. The output of the primary school system over the projection period cannot be absorbed by the private monetary sector, except under extremely implausible assumptions of a radical reduction of real wages and/or of a real annual output growth much higher than the already optimistic 7.1 percent we adopt from the country's Develop- ment Plan. If the government sector does not provide large numbers of new jobs, the majority of primary school leavers will have to stay in the small- farm sector or remain completely unemployed. 26. 'l'here will be a stl"ong downward pressure on the earnings of sec- ondary-trained people as more and more of them start looking tor jobs; the full employment wage level for this group would be below the present average earnings ot employees with primary education. 27. The wages of university-trained people will remain high and be subject to further upward pressure, partly becaWie of continued African-' ization of high-level jobs, end partly becau8e of the large nUJllber of uni- versity-trained secondary teachers needed. 28. There is likely to be a two-sided substitution procesa of secondary school leavera tor the two other categories, relieving somewhat the upward pressure on wages for the highest manpower category, but aggravating the employment situation at the lower end of the Bcale. 2. The General Lesson a. Manpowe!_~9.uirements Appfoach: The Alternati ve to Cost-Benefit Analysis 29. Present efforts being made in comprehensive education8~ planning to go beyond the education sector and its peculiar planning pro'bleme, such - 218 - as proj ecting enrollments, or financial needs, and to deaJ. with the broader issues of formation and utilization of human oapi tal, rely entirely on the so-called ll18llpower requirements approach. Y 30. The reasons for the early popularity of this approach$mong educa- tional plann~rs are manifold, but its main appeal probably derives from the misleading impression of straightforwardness, simplicity, and exectness which it conveys. The resalts obtained can be immediately understood by adminis- trators!) policy makers and the public in general: we are told that certain investments in the various segments of the education system are necessary to provide a given number of people with the qualifications necessary to bring forth a gross domestic product of a given size and composi.tion by some target year. Furthermore,. the data required may be obta.ined rel.ati ,--ely easily, since the basic information for the initial manpower inventory is often col- lected in the course of a general population census~ Thirdly, in developing countries the acquisition of skills, their allocation over the economy, their contribution to production, and the interactions of these processes mar be less flexible than in industrialized countries. Such a situation favors the application of the manpower requirements approach, the very rationale of which is the implicit aasmnption of. such rigidity. b. £pncept'uel. Difficulties with Cost-Benefit Anallsis 31. Compared to the manpower reg.uirements approach ,the application of cost-1:Hinlefi t techniques leads to seemingly vague results. This has probably been one of the major reasons Why these techniques have never found theoreti- cal :r~!c.~ognition, let alone practical acceptance, among education and man- power planners. Another inhibi tillg force mq have been an !. priori reluc- ta.nce, particularly e.mong non-economists, to submit educational investments to any ,tes't which involves a comparison with investments in other sectors. Thirdly, date requirements are more difficult to aeet, a fact which, con- sciollsly or otherwise, mq enter into an educational planner t 8 attitude to- w~~~ this approach. Finally, the manpower ~equirements approach is likely to r:tnd more fa.vor fi"om a political. point of view, beca~e of what might be a _ _ _ _ _ _ _ _ • _ _ _ _ _ _ _ _ _ _ _ _~_ _ _ _- - _____ 1/ Apart from the cost-benefit approach followed in this study., which to O~~ knowledge has not been used before 88 a basis for educational plan- ning, or some variation ot this approach, such as a linear programming model to determine school input needs, there exists one other widely used. alternat1 ve -co the manpower requirements technique. In this ap- proach, social demand for education is often taken as the criterion for making educational investment decisions. The underlying political con- cept is that the role of the gOTernment is to meet, 88 rapidly as pos- sible,the demand for the different kinds and levels of fee-free school- ing. This impl.ici tly 88SwneS that the social returns to all levelfrj ot education are high and persistent. It is difficult to con.ider t17,is social demand approa.ch as comprehensive educational planning. FOi,r a de- tailed methodological discussion, see M. Blaug, A cost-Benefit Approach :Eo Educational Plannins in DevelQl)ing Countt.1!!., IBRD Economics Depart- ment Report EC-157, December 20, 19b7. - 219 - called its innate expansionist quality_ Even a bold educational growth pol- icy can be plausibly justified in this w-r, something which the more restric- tive nature ot the cost-benefit approach would not permit without falling back on a second line of defense, namely, that ofi indirect or non-economic· benefits. 32 - It is thus hard1.y surprising that a good number of arguments have been provided, and are still being put forward, against the use of cost-ben- eti t analysis in educational planning.!/ They are discussed in the .firs·t; chapter of this study and need not be repeated in detail. It will slli'f'ice to name them here and to recapitulate briefly the way in which we try to take account of them. The Argument of Ability or Other Income-Relevant Factors 33. It is shown that (i) under certain circumstances the tmpact of abili ty and other income-relevant factors can be more important than that of education, (ii) it is possible, hQWever, to deal. with these influences in a reasonably satisfactory we::j. The variables can conveniently be aggregate'd in three groups, the first encompassing socio-economic variables such as tribe or parental education, the second job-related variables (occupation, length of work experience, etc.), and the third, p.ersonal traits like ability in a specific or general sense. In this study, the assessment of the impact of each of these variables or group of Tariables through multiple regression analysis and the ensuing correction of earnings streams can be ter.med success- ful, altho~ the last group ot variables could only be dealt with by intro- ducing the proxy measure of exam pertormance. However , it is questionable whl!!ther the very complex phenomenon of ability can be measured at all satis- ;~8.ctorily by means of anyone single effort, suoh as an ad hoc survey, wi th- out extended experimentation and observation. This exerci.e&s a whole, apart from performing the prime task of separating the influence of education from that of other income-relevant factors, can shed light on, for example, the possibilities and limitations of a policy of equal educational opportu- nities. The Argument of Systematic Earnings Biases 34. The three kinds of distortions usually mentioned in this context are those related to the size of the firm, union membership, end affiliation wi th the public service. The influence ot these variables, aa & group, wher- ever it is significant t can be assessed and corrected tor in exactly the same way as ''for the above-mentioned three groups of variables. They can con- veniently be a~tributed to the cluster of job-related vari~ble8 and do not constitute a problem sui sene!~s. !I It m~ be noted in passing that a good deal ot these obje~tions, and other more serious ones, apply to the manpower requirements approach as well, although on. the whole they seem to be either ~ilently ignored or dealt with much more generously.. However, such a duplication of short- comings is hardly a satisfactory counter-argument. - 220 - 35. It, is worthwhile, however, to ,discuss in somewhat mor.e detail the case of salmes in the gOTernment sector. The public sector is usually the largest single employer in both developing and developed countries, so that a signiticant (usually positive) diverg~nce ot its salaries from the cor- responding s81aries in the private sector is cause for concem ..1The problem is not 80 much one ot identification but of interpretation: are -l;he "true ll rates of return those realized in the private sector, in which case private sector earnings might serve as a yar.dstick tor both education~ investments and eventual salary reform. in the public sector, or should a weighted aver- age of the two categories be used? The answer to this question depends a good deal on. whether one views the inter-sectoral earnings biases as a transi- tory or permanent teature of the labor market in a given count·ry. If they are believed transitory, budgetary ,and social pressure will lead to a down- ward adjustment ot civil service salar.ies; it permanent, itma.,v be necessary to ofter attractivesalaries to expatriate manpower (e. g., teachers), B1ld to respect the" strong political powers ot civil servants. 36. If these differences have indeed prevailed over a considerable period, an endogenous explanation -81' exist. The public sector Bubsample ot our Labor Force Survey, which unfortunately is non-random. and. thus cannot yield concluai ve results, appears at the esc and HSC Y levels t.o encompass a much larger share ot certitieate holders from schools which are considered, both by common consent and on the !rOunds ot exam results, as the country's best. If the statistical signitic~ce of this ditference could be established, i'l; could be argued that one ia reallY' dealing with two diff'.erent groups ot school leavers, because the elite schools (in the academic sense) ~~ght have provided quite a difterent experience and conveyed a difterent set of atti- tudes than the rest of the secondary sChools. Y In this case i.t would be justified to regard the two groups of schools as separate systeu, one open- ing access' to better paid civil .errice po.itions, the other to' private sec- tor jobs, and each wi th its own 'rates of return. The Argument ot §pillover Eftect. ~r Externalities 37 • As hu been pointed out earlier in the study, the spillover effects of education mq appear in l'I8.DY areas ot aD individual's and a ,society's life. These are occasionally straightforward, but more otten than not rather elusive. While it is certainly not alwqs possible to express them in mone- tary terms, if only for WIlDt of adequate data , it should be feasible to trace and quantity' them in SOUle form. or other. ' 38. The main reason why we neglect externalities in this study is that this subject justities a major study in itself. That the issue is extremely 1/ Tho :t'pl:lowing argument would not be Tery convincing tor HSC holders alone since s,condary schools with senior classes are almost by definition above e.yerage. 2/ \That is', abOTe and beyond differences which are reflected in the exam. 6\COreS ."d which have been otherwise accounted for in the analysis. 11 - 221 - complex Dl8¥ be illustrated by two examples. It has been mentioned earlier (Chapters I and IV) that the Labor Force Survey data as well as the 1966 Census figures show a negative .. sociation between educational achievement of aduJ.ts and the number of children they have. It is an open question whether a causal relationship is at work or not. Annex Table 10.1 Juatifies this caution. While the figures do not come as a surprise, they should dispel the illusion that spillover effects are easily discernible phenomena. A second frequently quoted example of externalities is the modernization impact of education. Again the picture is far from clear-cut, as can be seen from Annex Tables 10.2 and 10.3. 39 • On the whole, we feel that claims of spillover benefits should be viewed with same reservation. A first sound rule in interpreting them is to determine whether symmetry of external. -benefits and external costs (dis- benefits) is maintained. In other words, increased national cohesion should not be credited to education without also taking account of the possible disintegrating influence of large numbers of unemployed school leavers with frustrated expectations. Or, to use a more remote example: the increased familiarity with modern life of a country's ]i'w-abiding majority ot citizens ought to be confronted with the increased sophistication of its law-breaking minority. 40. Secondly, externality arguments should normally not be used to in- vert a ranking, either within the education system or of education versus other sectors of the economy, which has been arrived at by applying purely economic (income) criteria, in view of the implicit (or, as in Kenya, ex- plicit 1/) priority of economic criteria in many developing countries. 41. This is not to deny that there ~ be occasional exceptions to this rule, but they can probably be dealt with in more expeditious wa.vs. It, for example, a government is pledged to the achievement ot universal free primary education, it m8\Y as well respect its decision as a basic poli'· tical choice, rather than looking for real or imaginary externalities to justify it after the event. The Argument of Non-Marginality 42. While we contend that the study shows how the above-mentioned argu- ments can be dealt with, we are not sure whether it answers the last, and most important, objection. As haa been pointed out time and again, a single set ot rate-ot-return figures says a great deal about current conditions in the market for skills, the de.and and lupply ot educated labor, and hence, where additional educational expenditure would be moat rewarding under pres- ent conditi~n8. However, it tells UI little about what level expenditure on various kinds ot education would have to reach betore the rate ot return would equal that of the next profitable segment ot the educational systeln. The obvious w~ out of this dilemma is to link the differential earnings ot cohorts of.chool leayera to their differential education COlts over a period 1/ Of. Chapter I, paragraph 21. - 222 - of year.. From there, ODe might choose some form of extrapolation to de- scribe the l1~ely ~ture development of rates of return to schooling under ceteris. E.aribus assUmptions. 43. Untortunately, the data problem in most countries. regardless of level of econClllic developllent, virtually excludes the reconstruction of pa.s.~ rates ot return to education. This can be particularly disturbing when dealing with educational systems which are rapidly expanding from a small initial w1uae, and in a situation where there i. a premium on speed rather than precision. 11 In this study (Chapter IX, section 2), we use a substi- tute for the time series of rates of return, based on the fact that employ- ment probabilities and their changes over time (in this case, 1960-66) have by far a greater impact on rates of return to primary education than any plausible change in nominal. wages. Projections based on suchestimatles are probably pr.one to substantial errors, but so are almost any ·.macro-economic estimates; manpower requirements projections are just as prone tl) error, judging from.past experience. 44. The .substitute we use is neverthele8s not totally satisfactory, and during the course of the study .:.a number of more sophisticated pro,1ections of rates of return are attempted jthey are reported on in detail in Chapter IX, section 3. It is obvious that any attempt to refine rate-or-return cal- ~~lations beyond the stage of straightforward extrapolation of time series requires the .introduction of demand functions tor various categories of edu- cated manpower. The demand function we use is of the Cobb-Douglas type, with the aggregate labor term replaced by three broad manpower categories (distinguished by the amount oteducatiop received); manpower is supposed to enter the production process under conditions of constant elasticity of Bub- stit~tion. 45. Thi,s attempt i. largely ,unsuccessful, although this is due to data shortcomings (the data buis coneisted or a rather heterogeneous cross- section of firms ) rather than cODceptualerrors • However, even if' ·"the in- formation available were more adequate, the aggregation ot all skills into on~ three categories might prove too much of a simplification, whereas an increase in the number ot manpower categories would obviously render the data collecting and analytical pr~blems more difficult. 46. Alao in Chapter IX, we deal. with the estimation or .employment fig- ures corresponding to' certain rate-ot-return/wages combinations, and the wage implications of certain rate-of-return/employment combinations (sections 4-6). This is a somewhat modified realization of M. Blaug's 2/ proposal to 1/ One should caution against exaggerated notions of what constitutes an ,acceptable trade-ott between speed and precision. With a few exceptions, the cm the schooling-e&.1'llingl relationship. 'l'hisadjustment reduces the sehooling coefficient to 31.5 Kah per aonth per additional year .of schooling (Equation 3). The coetticient bu now been corrected for all those non- schooling variables which are unconnected with tormal education. 'l'he in- crease in wage-earners' lIIe, and the -added job experience that goes with it, is- not part of formal schooling, and the effect of age on earnings is therefore separated from the schooling-earninF,s relation by holding age con.- stant. Likewise, father's occupation, tribe, and literacy of parents are proxies for variables exogenous to schooling, such as influence on the job market, closeness or remoteness from the major centers of demand for mappower, and home training, and they are also corrected for. 5. The remaining two groups ot non-schooling variables -- occupation, size of firm, aDd sector of employment and lecondary school exam score -- are also brought into the regres8ion (Equations 4 and 5). As 'We explain in Chapter V with reference to the earnings profiles corrected for the latter groups of variables" we cOD8ider the decl~ase in the schooling coefficient when these 'variables are held constant to be an overcorrection, because their values are linked to the amount ot--schooling. A::.. ~ ~- ~ Appendix Table·C: l~enya: .rl.egression Joefi'icients of Schooling and Independent Dumnv Variables, . Dependent Variable: Monthly Earnings, African 1'1a1es.L 196.~~ Scnooling irom 0-11 lears t{umber of Independent Coefficients Coefficients Coeffictents Coefficients 00efficients ?ieans Variables Name iguation 1~ Eguation 2 ~9uati~ Equation 4 ~uation 5 of Variables 85 Years of Schooling 25 .3--:·~·· 36.2 31.5 18.8 12.3 0.512 5 Age ~ 14 -328.8* -342.1* -217.0*- -202.9* 0.001 6 Age 15- 16 -492.8.r. -466.1* -224 .. 4* -177.8* 0.000 7 Age 17 - 19 -278.9 -292.5 -234.3 -209.4 0.019 1=1 o 8 Age 20 - 24 -204.4 -211.7 -183.3 -182.2 0.176 < 9 Age 25 - 29 -101.4 -110.6 - 89.'1 - 87.1 0.216 10 Age 30 - 34 - 32.6 - 35.5 - 30.7 - 25.8 0.210: 12 Age 45 - 54 - 3.2* 2 .4~- 702* 5.~ 0.100 1 A. e + 17 .1~- 20.2 - 6.1 .1 .026 :>-t 21 Father illite 109. 11 .9* 0.2-~ o. 90 i E-I H 22 24 26 Father lit. Hother illit. 152.6 - 84.7 146.5 - 47 . 5 107.2* - 40.9 0.291 0.872 H Uother no ans. 75.0 ll7.4 76.6* 0.016 27 Yillruyu - - 75.3 - 54.9 - 52.4 0.316 28 Kamba - 69.3 - 54.5 - 52.5 0.207 !Xl 29 Luhya - 92.1 - 56.6 - 45.4 0.171 t:Q 30 Coastal - 79.3 - 84.5 - 95.6 0.067 ~ 31 Luo - 80.0 E-! - 34.1 - 22.3* 0.127 32 Kalen -in .5* - 8.5.r· .5* 0.01 Farmer 31.7 17.0*- 12. * o. 99 69 Prof. 1 l2.6* 24.9* 76.5* 0.029 70 Prof. 2 37 .9-~ - 30. 7~: - 17.9* 0.104 71 Techn. 28.3* 59.8~- 80.2* 0.059 rllH 6 72 Foreman 32.2* - 1.~ 4.0*- 0.108 -.8 73 Admin. 49.6* 17 .2~- 2.9* 0.'101 ffi~ 74 Clerk 66.5 30.0* 25.4* 0.132 ~O 75 Skill. 78.5 58.8 ~O 55.7 0.146 ~O 76 Semi-skill. 25.3* 27 .1-~ o. 0.216 Nairobi 15 l{ombasa 27.1* 2 . )~t. " 0.7 65.3 69.5 0.162 I 2 Public sectcr.c 94.4 8.0* 0.096 A:5 17 Firm size l-14 -75.5 -76.2~'- 0.006 :::>H l8 Firm size l5-49 08 -50.1 -50.6 0.095 a 20 Firm size 100 + -15.3-~ -ll3.4~~ 0.655 Anpendix Table C : (continued) o - UZ"rd:l.rS ~~ .• .,.;.t+ --, -.------- Coefficient.s Coefficients Ivjeans Independent .::;oefficie 3 ts ~oei'ficients -';oefficients ?Quation ? .!i:guation 3 Equation h Equation 5 of Variables Variables huation 1 -102.0* -20.1* 0.000 39 i'armer 99.5 0.012 109.6 ...-.. hO Prot. 1 209 .5~~ 247.3* 0.001 Cl hl Prof. 2 172.5 138.5 0.005 E-t h2 Technician f5 252.8 257.1 0.019 r) 43 ~'orema.n 954.7 813.0 0.015 --- 6 4h 45 Admin. Clerk 136.4 117.0 0.118 0.138 H -1.3* 21.3* E-t 46 Skill. -74.0 -55.1* 0.414 ~ 47 Semi-Skill. 28.8* 0.450 Manufacturing 30.2* {.) 49 3 .cr.~ 4.4* 0.070 o 50 Construction o 21.0* 47.8 0.159 51 Commerce - 2.1-:'. 16.4~- 0.131 52 Services ll3.2 122.8 0.695 53 Union -46.3 -12.2* 0.253 56 Primary public -49.6 -11.8 0.391 57 Primary private 72.2* 86.1* 0.137 78 Addl. educe 70.5* 0.849 79 lifo addle educe 55.3* 75.1 -12.3* 0.164 81 Addl. training - 8. 6-~ -96.2~- 0.824 82 No addle training 78.4 0,048 60 Sec. public general 0.002 218.1 61 Sec. public teacher training -64.7* 0.004 62 Sec. pri:a.te technical -56.5 0.032 63 Sec. private general 0.001 z -115.1~' o 64 Sec. private teacher training H -110.9* 0.003 E-t < 65 Sec. Harambee -33.9 0.619 (.) 66 Sec. not applicable 42.6 0.213 ~ 67 Sec. no answer ~s-:! Division I l. ,oJ..., 4 Lj,;)':>. I 0 .. 003 34 357.8 0.011 35 :.:;stJ Division I I esc Division I I I 293.0 0.013 36 231.2 0.005 37 ~SG G:,jE ;.l2 0.09 0.16 0.19 r·.37 0.42 L~tercept term 27D.3 299.0 32l~. 2 17£1.;] 269.0 ,Average earnings (~~sh/1ilonth) 392.4 !1vpr:i -:e ace (years) 32.9 * Not slgnif'1cant. Source I Labor Force Sample Survey, January/February 1968. APPENDIX D DERIVATION OF PROCEDURES FOR ESTIMATION OF ABSOLUTE EARNINGS FROM REGRESSION DATA 1. As stated in the text of Chapter V, the model we use in pertona1ng the regression ot age, socio-economic, and occupational variables OD income tor each schooling group is ot the tor.m: i = 1 •• c. II j - 1 ••• JIl k • 1 1 and tbe predictive equation JIll\Y be expressed in the notation both ot deviates (e.g., Xij - X ) and of the original variables (X ' etc.): 3 i3 Deviates Orisinal Variablea (1) 2i = Y+~bJ (X iJ - XJ ) +~Ck (Zik - ~) I i · be +7 b .1 XiJ +~ c k Zit where b 0 '" 0 • B X ij • age variables bj • §j Zk • aocio-econOllic, occupational variables c k a:: C ~ k - Y, - X, - Z • means of X, Y, and Z within a schooling group 2. Our objective ia to tind tor each schooling group ot African males with 11 or tewer years ot schooling a monthly incame figure that baa been adj~sted to eliminate differences in earnings due to differences, within the aeliooling group or between schooling groupe, in parents' li terac)", tather' s occupation, or other characteriitici which intluence eara.iql. The compu- tational procedure is derived gelow siaultaneoUlly in the deviate and original variable notations. The calculations in the text are in the de- nate tora; the original variables torm is an equivalent alternative for the computation. 3. F'irst, (1) is evaluated "hen Zik. • ~ for &1.1 i, that is, when the: Zik are let equal to their _an y&1.uelr. Deviates Orisina! Variable. ~il Z ik -· • Z k b0 +Lb j . ~ Xi" d APPENDIX 1) Page 2 4. Since the Xij are zero-one variables , Xj =: 1 for age j and zero otherwise. Restating {2) tor the case X a: Xj' we obtain: Y (3) iii Zit = - .. Y + b J ~i - Eb J Xj j Y1 I _- '" Zik. - Zk b 0 + b.1 + L ck Zit k X ==X-:- X aX J j 5. We now wish to adjust the Ii to reflect the extent to which the particular schooling group varies from the whole sample (A:t'rican males with 11 or rewer years of schooling) with respect_to the distribution of socio- economic and other variables. We designate Z*k as the means for the whole sample of the aoeio-economic and occupa.tional variables. Recalling that the Zk are' the meanfl' ot these variables for the individual schooling groups, we add the adjustment D: D = L: * Ok (Zk - Zk) k and this gives: The adjusted incomes are thus equivalent to those measured by the given regression equation wllere the Z variables are assumed to take on the sample mean values for all regressions to The e,q>ression (4) in deviate notation describes algebraically the computational procedure carried out in the Ex8Dlple in Chapter V, when the Z are the socio-economic variables only. It describes the procedures used to optain the Table 5.3, when the Z are the socio-economic and occupational variables, but when only significant values of Ok are used. 1/ A look at (3) shows in a particularly clear w~ the relationship between the 8 chooling grouP. meane in the deviate form (Y) and their relevant intercept terms in the original variables form (b o ). By setting equal the right hand terms of the tvo equations, we find: bo =y - (F b.1 X J + ~ cit Zk) which Illq be recognized 88 the usual tQ;rmula tor the least squares inter"", cept or, regroupj.ng: r. ':)y .. b o + E j b .X j j l' E k C k Zk APPENDIX E CORRECTION OF RURAL LANDHOLDER INCOMES FOR ACREAGE AID FAMILY SIZE DIFFERENTIALS BmwEEH AGE GROUPS AIU) BETWEEli EDUCATION LEVELS FOR GIVEN AGE GROUP The correction is made in eight steps. The data are :f'r01l1 Economic Survey of Central Province, Statistic. Division ot Ministry of Economic Planning and Deveiopment. 17 Farm size in acres (5) 1s estimated as a linear function of age ot household head (A ) within each schooling category. Schooling Category Equation S1eItlcance Level F t R2 Illiterate S • -1.147 + o.l460A 0.01 0.01 0.03 LIterate S • -1.645 + 0.1735A 0.01 0.01 0.16 1 - 3 S c -4.668 + 0.3048A 0.01 0.01 0.21 4- 8 S • -3.667 + O.3074A 0.01 0.01 0.13 9 plus S • -8.242 + 0.~548A 0.05 0.05 0.~5 GroS8 t~ Income per acre (IrIS) is estimated as a hyperbolic function ot fara size (S). Schooling Category Equation Siggiticance Level F t Illiterate YtlS • 169.1 + 107.2/8 0.01 0.01 0.10 LIterate Y /S • 102.6 + 273.0/8 Q.Ol 0.01 0.35 1 - 3 yt/s • 195.4 + 168.6/s 0.01 0.01 0.24 4- 6 t (ieither the F nor the t-value is significant; ) 9 plus (we &88U1le average farm income per acre to be ) (27l"., 3 and 219.0 reepectively tor the tvo group •• ) step 3. Total cost ot t_ily labor be.ides houaehol4 head is estimated by tir.t calculating tro. the Central Province Survey the number ot man-, Waa&n-, aDd ch11d-~s worked Ulnual17 on the tarm &8 a tunction of t&rlll 8ize. Y The data reter to three t&rID 8ize groups ( c 4.00 acres, 4.00 - 7.99 acrea, 8Dd :!8.oo acree): a 11, bY ~n!i9.nF •. "~,,.ell,·StUltord Food Research Institute. ~~~a ~~1~,,1;~.~ • . " .. ~ ,I gj The average cost ot hired labor i. only about 13 Ksh a year, and is gener&.tly restricted to larger farms. :" .. It 1. omitted trom thi. analysis. APPENDIX }l , to Page 2 M_-d!l- WO!ll!'l-~ay:s 2.33 165 249 57 5.62 208 318 81~ 13.45 250 452 164 FamilY size (F) is estimated 88 a function of f~ size (estimated ~0Dl DIeGS, not as a linear regression). Acres Eg,uation S ~ 2.33 F = 4.96 + 0.41 (2.33 - s) . ~;, 2'.39 c:: s· ~, 5.62 F 4.96 :ft + 0.41 (8 - 2.33) 5.62 S~. !: 13.45 F = 6.31 + 0.31 (8 - 5.62) 13.45 <: S F :z 8.76 + 0.31 S Fro. these results and the average shares of adult males and fe- males in the whole aamp1e, we arri ve at the number of men per tamily as a function of farm aize. Assuming 21.8 percent adult males and 27.3 percent adult females regardless of family size, we arrive at 1.08 aales tor a farm ot 2.33 acres, 1.38 mal.es for a far.m of 5.62 acres, and 1.91 males for a tara of 13.45 acres. The number at man-day_ ot work by Don-head ot household ~ales is thus 13 tor 2.33 acres, 57 tor 5.62 acres, and 119 for 13.45 acres. Weighting these man-d8f8 ot work by 2.19 Ksh/d~, and the woman- and chi1d-4ays of york trom. above by 1.60 and 1.00 Ksh / dey , respectively, we arrive at the total im.puted labor cost of non- houeeho1d head t ..ily labor .. a function ot farm size: Total Don-household bead Coat per Land rent Total cost acre per acre _ per acre_ Acres , labor cost iguted tallill 2.33 483.9 207.7 20 227·7 5.62 717.6 127.7 20 147.7 13.45 1147.8 85.3 20 105.3 To get total cost per acre at other tarm sizes, we interpolate between these acreages, extrapolate below 2.33 acres, and. assume that the cost ot 105.3 Ksh/aere st8JS constant above 13.45 acres. step 4. The imputed tarm income ot the household head by age (Appendif Table i, eolUIID8 (b) i. found by subtracting total cost per acre (Step 3) tillles the nUllber ot acree by age (Step 1) from family gross. t8l'Jll income aa a function ot education and age of household head (combined steps 1 and 2; see table, columns (a»). step 5. The Don-tara income ot the tem:J.l.y (Int) is estimated as a function ot ~e ot household head for schooling groups. APPENDIX E Page 3 Schooling Category Equation Siseificance Level F t R2 Illitteirt';;t,li:! Y f • 518.7 + 5.596A 0.05 0.01 0.01 Literate ynr • 755.1 + 16.06A n.s. 0.05 0.00 1 - 3 ynr • -1963 + 88.45A 0.01 0.01 0.11 4- 8 yDt • 295.2 + 66.4oA 0.05 0.05 0.04 9 plus Y:t • 3066 + 32.89A n.s. 0.01 0.01 Although the F-leve1s ot two equations are not statistically signiticant at the 5 percent level, the equations are used anywa.v as the best estimates ot non-tar.m income as & function of age. Step §.. The nllilllber of adult male and temale family members as a fUnction ot age of household head i8 tound by taking the relation between farm size and aize of f.-ily (Step 3) and inserting it into the :relation between tarm size and age (Step 1). Again, 21.8 percent of the family 'is asaumed to be adult males and 27.3 percent adult females • The Central Province Survey results show that 23.6 per- cent of adult ~es and 5 percent ot adult temales work ott the farm. which implies that a temale has about a 20 percent probability of working off the f'&l"Jll :relative to adult males. Females also earn 1es 8 than males -- it is assumed that they earn 73 percent the wages of mal.es (1.60/2.19). One adult f&rl1 female is thus equiva- lent to 0.146 adult f&l"ll m&1.es. UsiDg these weight., we find the total male equivalents on farms grouped by age and schooling ot bead of household. Years of Schooling ~ o Illit. a Lit. 1 - 3 4 - 8 9 plus .~~~: 17 1.17 1.17 1.09 1.19 20 1.22 1.23 1.18 1.29 1.12 30 1.37 1.43 1.50 1.62 40 1.61 1.55 1.60 1·77 1.87 1.97 50 1.64 1.74 2.03 2.11 60 2.33 1.78 1.86 2.27 2.36 2·71 SteE I· Under the USWIlpt10D that the _mbere of the tamily (in .ale equi va- lenti) are ot the .... DOD-tar.m wage-earning capacity as the head of household, we divide taily DOll-farm incoae (Step 5) by these equivalents (Step 6) to let the non~tar.. income ot the bead ot hous.hold him8elt. The result. are .hown in the table, columns (c). Ste!? .8. The total imputed 1nco.e ot heads ot household by age and education corrected tor acre&8e and tamily siZe difterentials is the sum of C01UllD8 (b) and (c) . . . shown in text Table 5.8. f Appendix T~bl~ E: Kenya: Gross Fami. d Non-Farm Income -64 ~ YEARS OF SCHOO-LING I 1 l i t era t e L i t era t e l - 3 4 8 9 plus (a) (b) . (c) .\ (a) (b) (c) (a) (b) (e) (a) (b) (c) (a) (b) (c) 11 I 334 -4 525 406 11 879 269 130 -422 423 39 1197 20 i 406 -20 511 460 23 875 ~8 93 -164 673 117 1258 186 -38 3325 30 653 -12 501 638 -61 865 1044 256 460 1508 674 1412 1183 357 2517 40 900 98 479 817 -10 874 1640 601 889 2341 1202 1259 2179 9b5 2174 50 1147 255 487 995 9 896 2236 956 1211 3174 1828 1147 3176 1654 1979 60 1394 302 480 n64 14 915 2832 1401 1472 bOlO 2458 1307 b172 2172 1823 ?Tote: Columns (a): Gross :family :farm income. Derived :from 4>pendix E, Steps 1 and 2. See step 4. (b): Inq>uted farm income of household head. See Appendix E, Step 4. ( c): I~uted non-fElrM income of household head. See .Appendix E, step 7. APPENDIX F SOLUTION FOR CHllOE IN WAGE LEV.F.L'lWO FROM DISCOUNT FORMULA (9.35) 1. We begin with equation (9.35): s· (9.35) 0 • -~ OJ, + adWl t (1 + r2}i +±: i • SI + 1 bi + AW2t -.6Wlt (1 + r2)i the expected discounted costs and lifetim benefits, • respectively, of tald.ng the addi tional.. years of school- ing needed to !lOVe from skill level 1 to level 2, each level being defined in years of schooling; . ' . the change in real annual wages (income) of skill level 1 in t1_ period t, to be aasUJled COll8tant over the entire age range, 1 sa S I + 1 to 1 • n. /1W2t • the corresponding chauge in real. annual income of skill level 2 in time period t; • the rate ot return w the add! tlonal investment in schooling required to move from skill level 1 to skill level 2; and S' • the number of years of schooling required to JOOve f'rom level 1 to level 2. 2. Regrouping this equation, we find: • • Appendix It' Page 2 3. When income foregone does not enter into the calculations, the equation and solution for 6W2 are sinq>lified. . .. .. ~ -Oi + b i ~ 6W2t - AWlt o • + ~ - ,~ ;. i CI (1 + r2)i i + 8. '+ 1 (1 + r2) i APPENDIX G TEACHER REQUIREMENTS IN KENYA IN 1974 1. The following table shows teacher-student ratios by category ot teacher training, level ot education for the years 1960-1967. There is a trend in 1960-1967 for the number of teachers per thousand pupils in primary school to rise (from 23.8 in 1960 to 31.0 in 1967), and for the ratio in secondary school to fall (from 55.7 in 1960 to 42.3 in 1967). At both levels there is also a decrease in the average education of teachers per student over this period. Primary Education Level of Forms.l. Schoo1in~ 1960 1961 1962 1963 1964 1965 1966 !2tl Universit.,Jl 0.54 0.31 0.34 0.41 0.12 0.12 0.11 0.10 second~ 5.07 5.38 5.36 5.85 4.70 5.31 5.78 5.89 Primar;/ 18.24 17.52 18.50 19.26 22.67 24.83 26.25 25.05 Seconda~ Education Uni versi t·.,Jl 38.81 41.89 39 .40 35.91 35.18 28.52 24.89 24.14 SeCondary!V 16.86 17.16 13.65 16.71 11.48 18.81 19.21 18.13 Notes: 1/ Includes both trained and untrained Graduate teachers. 2/ Includes 61 Pl , P2' HSC,and esc teachers. Since P2 teachers statistics are not separated out before 1964, it is assumed that P2 teachers are the same percentage of P2 plus P3 teachers in 1960-1963 88 the average percentage in 1964-1961 (.185). Since there is no trend in the latter, and the variation is small (.178 to .191), the error on this assUlllption is probably small. 3/ Includes P3, P4, IO'E, and "other" categories. Ti/ Includes Sl, PI, HSC, and CSC, and "other" categories. Source: Statistical Abstract, 1967, Tables 3 and 5. 2. In order to make projections, several assumptions have to be made about the continuation of these trends between 1967 and 1974: APPEIDIX G Page 2 (i) The nuaber ot universi ty trained teachers per thousand stUdeDts in priDlary school will remain conatant at 0.10. Thia probably' biases upward the estimate o'f graduate teachers iD prtmar,y schools in 1974. (ii) 8_ The nlaber ot teachers with full secondary edueatlon per thou"and students in primary- schools riaes at the .rate in 1967-74 as it averae;ed in 1960-1967. The ratie would thus reach 6.84 in 1974. (iii) The nlaber 0'1 primary school trained teachers, in keeping with the 1960-67 trends of ·80 increase in total teachers per thouaudstudenta and a decrease in education of teachers per thousand stndents·, rises at the a_e rate in 1967-74 as in 1960-67. The ratio would the.retore be 34.39 in 1974. (iy ) The ratio ot graduate teachers per thousand atudents in seconclal7 school remains cODstant at 24.14 between 1967-74. 'l'hiB probably overestimates the numbe.r of graduate teachers per thousand secondary school students. (T) The .ratio of teachers vi th tull secondary educatlon in seconclal7 schools continues to rise at the sillie rate as in 1960-67, bringing tberatio to 19 .46 in 1974 and reversing the trend toward fever teachers pers'tudent in secOllclal7s chaol. 3. These as8uaptious result in the follOWing projectioDs for 1974: romal Schoolill8 Level ot School ... Leftl. 0'1 Teachers, Primary Secondary;!1'OtaJ. Graduate 144 3,645 3,789 Secondary' 9,863 2,938 12,801 Pr~1 ~9,590 49,590 59.591 6,583 66,180 10 estimate iaincluded tor the number of teachers needed tor1;eaCher train- iDS. For er.:.ple, about .80 percent ot the teachers with fUll '.econd~ary edu- cati_ teaching ill pr1aary .chOC)l. in 1967 had two yearsot t-eacher training beyond their tormal schoo1iog. More thaD two-thirds ot the teachers with prblary educatica had two years ot addition8.l traiDing.. The teacher pro- .1ectiona, therefore, would appear to be underestimates ot the likely 1974 total:s. APPENDIX G Page 3 4. In addition, it should be clear that the projectiODs take as given certain trends in relative wages ot teachers. The &8sUIIled more rapid in- crease in the number of teachers with primary education relative to those with either secondary and university education aaau.ea tmplicitly that the \\ shadow price of secondary and graduate teachers will rise relatiTe to pri- m~~ school teachers between 1967 and 1974, as it was probably rising in =_1~~6~ to 1967. The wages of secondary teachers are &1so implicit17 assumed to rise less rspidly than for graduate teachers. • Annex Table 2.11 Kenya: Annual Rainfall and Population Density, 1962 NUJIlber of Annual Census Total 1/ Rainfall EnUlleration Population Total 11'ea Median Danai t1- (in inches) Districts ~ 1000) (1000 sq. lliles) (e-r sona per sq. llile) 5 3 215.0 48.4 4.1 10 4 113.6 43.5 5.8 15 .18 120.0 34.8 8.2 20 21 2.55.1 28.6 13.6 25 52 363.4 17.3 24.8 30 .57 .510.1 11.0 14.5 35 68 117.8 8.3 107.8 40 54 1079.9 6.1 128.4 45 10 1243.9 8.2 236.0 50 5.5 1104.1 4.8 297.5 55 38 185.2 3.0 331.8 60 31 634.2 2.4 340.0 65 18 587.' 1 • .5 387.3 10 11 660.3 1.6 469.9 15 --1i 62.8 0.3 287.0 510 8633.6 219.8 !I Excluding urban areas. Source: W. T. W. Morgan and N. Manfred Shaffer, Population of Kenya, DensitLand Distribution, Oxford University Press, Nairobi 1966, p. 6. Annex Table 2.21 .Ratio of ~le-FarIIl to Small-Farm Revenues tram ,Spec ric Products, 1963-67 .- 2/ 2L96)!1 1964- 1965 19661967 Maize ~)214B 61139 ,4:46 46:54 68,:32 Other Cereals 1~6 ,14 89:11 89111 81:19 81:19 Pyrethrum '71129 ,3:47 49:51 29:71 16:84 3/ Sugar Cane- 'iB2t18 861.14 84:16 89:11 66:34 ,Co'free 7'12, 60:40 58:42 50:-50 40:60 .Tea "9113 95:5 93:7 92:8 88:12 1I1:sal .,84116 87:13 95:5 98:2 94:6 "Er.u1t 321:68 32:68 35:65 27:73 .25:75 4/ ~attle- ;,6,:44 26:74 21u'76 23:77 22:78 4/ :Pilgs- 99:1 9614 8,115 90:.10 88:12 .:P.oliltry and .83 s11 75,:.25 97,:3 71:29 67:33 ~gs :Da±rY,Products 87113 72:28 70:30 67:33 65:35 1/ Cash revenue to~producer8. Coffee in .1963 .includes Jlbum, ..:and :fruit for 1963 -includes v8'getab1es,and nowara. :Y.1964-19671 Gro8s revenues. J!Includ~8 ·j:qge1.'y i'orl:arge ,farm. Jv For alalqlhter. ~Source8: F.9r196), .Sta,tistical .Abstr-ac;t,,1967; ror1964~1'>967, ,;Economic Survey, 1968. - Ann~~~Table ~.3 : Percent Distribution of Non-African Urban Emp10lees by CitizenshiE status, 1968 Tanzania Third Applied Total Number Kenya or Uganda Country for Kenya No of Citizens Citizens Citizens Citizenship Answer Observations Asian };Iales, Priv-ate Sector 30.9 2.8 58.2 4.8 3.3 395 Asian Males, Public Sector 24.5 51.0 24.5 49 Asian Females, Private Sector 29.8 16.0 45.7 4.3 4.3 94 Asian Females, Public Sector (27.8) (-) (61.1) (-) (11.1) 18 European Males, Private Sector 6.4 88.3 1.1 4.3 94 European !vIales, Public Sector (33.3) (-) (55.5) (-) (11.1) 9 European Females, Private sector 3.6 - 84.5 2.4 9.5 84 European Females, Public Sector (40.0) (-) (60.0) (-) (-) 5 m Figures in brackets have been calculated from small numbers of observations. Source: Labor Force Sample Survey, January/February 1968. Annex Table 2.11: Distribution of No~~African Urpal}..EJ!llQle,esin~ the Priva~. Secto.r .blAgeandGitizehS~p StatWJ, 1968 (in percent) Asian Employees European Employees Tanzania and Third Tanzania Th::t,rd Kenya Uganda Country Appli- No To- Kenya and Uganda COUli't~y Appli- No Citi~ens Citizens tal Citizens Citizens Ci:tize~~,~, - .Citizens cants Answer cants Answer Total is - 19 66.7 33.3 (3) (0) :20 -24 24.4 9.sY 65.9 (41) 83.3 16.7 (6) 1~5 ..- ~9 35.4 13.-)1 42.5 5.3 3.5 (113) 14.3 11.4 14.3 (14) iO ... 134 34.6 1.3 55.1 5.8 3.2 (156) 3.8 86.5 3.8 5.8 (52) '95-i44 34.5 2.3 50.6 5.8 6.9 1(87) 1.1 89.1 2.6 (39) '45 -·-54 '18.8 3.1 73.4 1.6 3.1 (64) 3.2 81.1 3.2 6.5 (31) =:55 ana B.o 4.0 80.0 8.0 (25) 2.8 88.9 8.3 (36) over iNC).~: The totals for each age-group are given in absolute nmnbers. Y 'DUe to the ihclusionofa .groUp oftraiheesfrom Tanzania. Source: Labor Force Sample Survey, January/February 1968. '0 Annex Table 2.5: Distribution of Non-African Urban EmElolees in the Pri vate Sector, by Ci tizenship Status and Occupation, 1968 (in percent) Occupations II III IV VII VIII IX - V VI X Asian Employees Kenya Citizens 38.9 15.8 21.41/ 13.5 39.1 37.9 31.2 20.5 25.0 Tanzania and Uganda Citizens 5.6 33.3= 6.0 Third "Country Ci ti zens 50.0 73.7 40.5 78.3 52.2 48.9 59.2 66.7 75.0 Applicants . 5~6 10.5 5.4 6.0 3.2 7.1 No answer 4.8 2.1 8.7 1.1 6.4 5.1 Totals (18) (19) (42) (37 ) (23) (182) (125) (39) (4) European Employees Kenya Citizens 10.0 4.3 1.1 12.5 12 . 5 Tanzania and Uganda Citizens Third Country Citizens 100.0 100.0 91.5 70.0 89.1 76.9 75.0 81.5 100.0 Jpplicants No answer 2.5 20.0 6.5 7.7 12.5 Totals (13) (1) (42) (20) (46) (39) (8) (8) (1) Y Due to the inclusion of a group of trainees from Tanzania. Source: Labor Force Sample Survey, January/February 1968. Annex Table 2.6: Distribution of' No.n-African. Urban ,_lo7e$8 in the Private- SectG~c1~.zen8h1~. s~tus'. and" S~~'e of; _10J1~:F!~',., 196 . (w'p!l'cent) Size of' Flirm, (Number o~~ Emj)~~eesJ .ian· Employees 1 - 14 15- 49 50 - 99 100 and more .~} Citizens, 20.0 24.6 45.0 26.6 ftDsama and Uganda Ci:tizens 0.7 6.7 5.9 'lJiif'd Country Cit.i:zen~ 60.0 62.0 38.3 59.5 _Ucants 20.0 9.9 5.0 0.7 .)Ahswer: 2.8 5.0 7.3 'T'·'~'···~"', (5) (142) (120) (289) -- • v ., ~ItiIit'_role_ r.nJ:8! Citi.zens 15.4 2.2 7.6 Tanzania and Uganda CLtizens Tllird Country Citizens 100.0 80.8 9.5.6 80.6 AliRlicantS: 3,8 1.7 lt~l ARswer· 2.2 10.1 Tbta.Ls' (2,) (26) (4.5) (119) lIo,te: Totals in absolute numbers. Source: Labor Force: Sample Survey, January/February 1968. 1& Annex Table ).1: Average School-Leaving Age of Urban Employees, by Ethnic Group, Sex, and Educational Achievement, 1968- Years of Education 0-2 3-5 6-7 8-9 10-11 12-13 14-16 17 and over - LA 14.0 15.6 11.6 18.4 20.2 21.4 20.5 26.4 African Males SD 4.6 3.9 3.0 2.7 1.4 2.7 0.5 3.0 NO 153 825 1061 161 213 31 2 28 LA l[:I---~O~ro:-7 17.3 1B.3 20.5 African Females SD 6.3 3.9 2.2 1.6 1.5 0.5 NO 16 62 116 23 31 2 LA 14.0· -13.4 15.0 - -16. 2 -18.2 - - - ---19 .-6 21-:-1----- -- 22.9 Asian Males SD 3.1 2.9 2.2 3.4 2.0 2.2 1.5 1.8 NO 3 38 101 50 139 24 8 20 LA 14.0 I5.ff-T1~u~-19.0 20.0--------23.1 Asian Females SD 0.0 1.2 1.3 1.6 0.0 0.5 NO 1 11 11 10 2 3 LA 13.0 14.1-- 15.8 --- -Ili.D - -16.9 - 18:1 ------20-:-0 --2ij:.2 European Males SD 0.0 2.5 2.4 1.4 1.6 0.9 0.6 3.1 NO 1 3 4 7 21 16 6 20 LA 12:-0--n:-O-~:9-~-18.() - --20.5 22.0 European Females SD 0.0 0.0 1.4 0.1 0.5 0.0 NO 1 2 56 22 2 1 Note: LA = Average School-Leaving Age, SD = Standard Deviation, NO = Number of Observations. Source: Labor Force Sample Survey, January/February, 1968. Annex 'l'a1tle.J.21: Distribution of Primary Schools by Lowest and Highest Stanciard Taught, and by Number of Streams, 1967 NUMBER OF SCHOOLS NUMBER OF SCHOOLS mGHEST STAND. LOWEST STAND. No. of AT THE lnGHEST AT THE LatlEST TAUGHT TAUGHT Streams LEVEL TAUGHT LEVEL TAUGHT 8TD I 204 5830 1 stream 5585 5150 STD II 293 18 2 streams 334 663 STn III 454 8 3 streams 31 120 STD IV 801 21 4 streams 6 22 STD V 232 61 5 streams 1 3 STl) VI 281 9 6 streams 2 1 STD VII 3694 12 7 streams TOTAL NUMBER 5959 5959 TOTAL NUM- OF SCHOOLS BER OF SCHOOLS 5959 5959 SourcP-: Ministry of Education Annual Report 1967- I'" ,; Annex Table 3.3: .n and Non-Citizens :Employed on Employed on Kenya Citizens ~Qca1 terms Overseas Terms Grand Description Male male Total Male male Male male Total Total PROFESSIONALLY QUALIFIED Graduate 2 55 7 16 30 20 7 73 80 U.K. Min. of Education Certificate 1 8 9 10 153 21 29 213 222 S1 3 10 13 18 44 10 3 75 88 PI 790 280 1070 164 382 45 23 614 1684 P2 3038 731 3769 7 20 1 28 3797 P3 11543 4470 16013 2 15 1 18 16031 p4 2042 776 2818 1 2 3 2821 Tech. Instructor (all grades) 124 14 138 3 2 1 6 144 Any Other 85 57 142 3 27 7 4 41 183 Total Qualified 17,628 6 2 351 23 z979 224 673 106 68 1.911 25.050 NOT PROFESSIONALLY QUALIFIED Graduate 2 8 10 14 8 7 1 30 40 HSC 17 9 26 18 17 10 45 71 CSC 602 90 692 29 69 13 6 117 809 KPE 6655 1718 8373 1 8 1 1 11 8384 other 1026 242 1268 5 31 5 9 50 1318 Total Unqualified 8302 2067 10369 67 133 36 17 253 10622 TOTAL QUALIFIED_ANP_UNQUALIFIED 221 930 8 1 418 34 1 348 291 806 142 85 1,324 35,672 Source: Ministry of Education Annual Report 1967. Annex Table 3.4: Percent Distribution of Weekll Hours bl 'SUb~ect Group, Primary and Secondary Education, 196 Lower Upper Primary Primary Second.ary Secondary (stds. I - III) (Steis. IV - VII) (Forms I - IV) (Harambee) English 13.6 22.2 22.5 26.2 Mathematics 17.3 17.8 17.5 20'.7 Mother Tongue '21.8 Geog~aphy 2.7 6.7 7.5 9.7 History, Oi vics, 1.8 6.7 10.0 10.7 Curr'ent Affairs Science 7.3 13.3 15.0 9.4 swahili. 8.9 7.5 5.9 Physical Education 12.7 6.7 5.0' 5.5 Art and Craft, Needlework, Domes.ti.c Science, Music 13.6 11.• 1 7.5 4.1 Religious. Education 9.1 6.7 7.5 7.8 <\ Number of' Weekly Periods' 35'(1.) 45 40(3) 41(5) Average Length of Period 30 )6(2) 40(41 n.a. (in minutes) (1,) 4.0 periods· per week in Std. III. (2) Two 40-minute periods' and seven 35-minute per.iods per day. (3) Preparation periods and extra-curricular' actin.ties excluded. (4) 4.,'-minute periods admissible. (5) Average. of 38 schools. Sources: Minis,try of. Education, Syllabus for Kenya Primary Schools:, N:airobi, 1967; Ministry of Education Circular' Letter No. INS/67-1., Nair.oDi, 1967. J.E.•. Anderson, Report on the. Conference of Harambee. S.cho:ol Headmasters, AUIDlst.10 -. 14,· 1966, Nairobi., 1966 (mimeo). . Annex Table 3.,: Cambridse School Certificate Examination ~CSCl Resu!ts z I~44 - i907 Candidates PassesY' Girls Total.s Year Boys Girls Totals Boys - 1944 n.a. n.a. n.a. 1.5 15 1945 n.a. n.a. n.a n.a. n.a. n.a. 1946 31 37 35 3.5 1941 n.a. n.a. n.a. n.a. n.a. n.a. 1948 n.a. n.a. n.a. n.a. n.a. n.a. 1949 61 61 60 60 19.50 n.a. nllla. n.a. n.a. n.a. n.a. 19.51 81 1 88 81 1 88 19.52 106 2 108 99 2 101 19.53 148 .5 1.53 139 .5 144 1954 165 10 175 153 10 163 1955 24.5 1 252 226 1 233 1956 368 16 384 283 16 299 1951 361 22 383 341 22 363 1958 .584 41 625 451 40 491 1959 746 53 199 605 49 654 1960 900 85 985 580 69 649 1961 1,009 99 1,108 144 84 828 1962 1,192 174 1,1)6 815 148 963 19632/ 1,314 229 1,603 1,014 199 1,213 196~/ 3,)45 1,592 4,937 2,)18 912 3,290 1965- n.a. n.a. 5,818 n.a.. n.a. 4,227 19662/ n.a. n.a. 6,455 n.a. nQa. 4,667 . 1967Y n.a. n.a. 9,2)0 n.a. n.a. 6,366 1/ GCE (0) passes recorded as failures g; African and Non-African pupils. Source: Ministry of Education Annex Table 3.6: Breakdown of Cambridge School Certificate (esc) Examination Results, African School Candidates &nly,-1956 - 1967 (percent) Boys Girls Division Division Division 1/ N~ber of Division Division Division 1/ Number of Year I II III GCE* Fa11- Candidates I II III GCE* Fail- Candidates 1956 17.7 36.7 22.6 23.1 368 25.0 56.3 18.8 16 l25:7 29.6 45.4 19.4 5.5 361 22.7 59.1 18.2 22 l~58 19.5 34.8 .22.9 22.8 584 19.5 48.8 29.3 2.4 41 1959 19.4 39.7 22.0 5.6 13.3 746 9.4 43.4 39.6 1.9 5.7 53 1960 15.4 29.1 19.9 13.1 22.4 900 21.2 41.2 18.8 12.9 5.9 85 J~9.61 18.3 30.7 24.7 9.6 16.6 1,009 26.3 36.4 22.2 9.1 6.1 99 :1;962 14.3 36.5 17.6 13.0 18.6 1,192 13.2 50.0 21.8 5.2 9.8 174 1963 17.8 32.1 23.9 6.0 20.2 1,374 15.3 40.6 31.0 4.8 8.3 229 2/ 1964- 16.1 27.9 25.3 7.3 23.4 3,345 15.8 23.3 21.9 9.0 30.0 1,592 2/ 3/ 196~ ,- 17.7 27.3 26.9 18.4 9.7 5,878 2/ 3/ 1966-- 18.5 27.2 26.6 17.3 10.3 6,455 2/ J/ 1967-- 16.7 25.8 26.5 18.0 13.0 9,230 * General Certificate of Education (Ordinary level}. 1/ Does not include GeE passes. 2/ African and Non-African School candidates. II Boys and girls. Source: Ministry of Education. Annex Table 3.7: Performance of Individual Schools in the Hse Examination, 1961-66 !I 1966 1965 1964 1963 1962 1961 1966 0.01 n.s. n.s. n.so 't:S Q) J.4 1965 0.01 n.s. n.s. n.s. ~ s:: 1964 196j 0.05 0.01 0.05 0.01 0.01 n.s. 0.05 n.s. ,Q) Ul as 't:S 1962 0.01 0.05 0.01 0.01 :a ta CJ 1961 0.01 n.s. 0.0, 0.05 0.01 0 I=: 1/ Schools are ranked by percentage of candidates who passed. The entries in the - Table refer to comparisons of two years. 0.01, 0.05, and n.s. mean that Spearman's rank correlation coefficient is significant at 0.01 (0.05, neither) level. The , upper half of the Table applies to the for.me~ African schools, the lower half to all senior secondary schools. Source: Ministry of Education data. Annex Table 2.8: students in Teacher Education Colleges, 1960-196.1 1960 1961 1962 19~3 1964 1965 1966 1967 d3l .(1 year) Male 7 4 7 11 Female 9 13 9 13 Total 16 17 16 24 51, (3 years) Male ,20 62 140 234 372 Female 11 63 112 163 216 Total 31J/ 125 252 397 58aY PI Male 439 480 391 393 371 248 386 530 Female 26.4' 252 243 294 151 79 121 163 Total 703 732 640 687 522 327 507 693 P2 Hale 291 h65 306 289 466 104 834 1162 Female ..105 195 171 142 159 184 323 495 Total 496 660 477 431 625 888 1151 1657 P3/4 (formerly K. 'r.3 and 4) Male 2292 1940 1985 2276 2331 2322 ~2133 1795 Female 961 ,883 1146 1185 1201 1309 1264 1108 Total 3253 2823 3131 3461 3532 3631 3397 2903 Total all Trainees Male 30?2 2885 2688 2978 3237 3418 3594 3894 Female 1430 1330 1560 1632 1583 1697 1880 2004 (j1J Total 4422 4215 4248 ;4610, 48201' 5115 '5474 589BW (\ 1/'15 technical teachers in training are ,omitted from this table. ,~ .Hot including 35-1e and 22 l'emalestudents in PI/51 upgrading courses • 31 Including a small :DuDlber on two-year courses. . ~ :Hot including 4 male and 2 temale'teachers for the"deaf. Source: 'Minist17of Education, Annual Summar.ies. • I Annex Table ~.9 " Distribution of Primar,y School Teachers over ~ne Salary Scale a 190Z (Kiambu and Nyeri Counties) Point on Scale Numbers of Teachers % Numb/ers d /0 Category (§.!.lary in Ksh) Kiambu Nyeri (Average) ~ PoS K N (Average) SI 582 1 0.02 P2 402 6 23 0.64 PI 348 14 10 0.53 420 3 12 0.33 375 11 32 0.94 438 1 4 0.11 ',.02 3 5 0.18 456 6 5 0.24 429 13 0.29 P3 162 163 100 5.77 456 6 0.13 168 186 98 6.23 483 5 0.11 174 142 95 5.20 510 4 7 0.24 180 248 90 7.42 537 192 142 82 4.91 564 1 0.02 204 7 0.15 $:91 216 109 91 4.39 618 2 o.oL~ 228 83 85 3.69 645 1 0.02 240 172 92 5.79 672 252 60 72 2.90 699 264 158 108 5.84 726 P4 120 3 0.07 126 1 0.02 747 1 0.02 132 138 780 ~ 2 0.04 150 816 156 1 3 0.09 162 2 1 0.07 855 168 1 0.02 174 2 0.04 894 180 130 163 6.43 P2 240 80 30 2.46 UQ 84 52 2 1.18 I 252 45 20 1.43 96 486 536 22.42 264 35 10 0.99 102 17 0.37 276 29 28 1.25 108 22 0.48 294 66 43 2.39 240 3 0.07 312 29 30 1.29 300 330 27 21 1.05 2623 1935 100.05 348 22 18 0.88 366 15 9 0.53 384 7 8 0.33 Source: Ministry of Local Government. ,. Annex Table 4.1: Distribution of Urban E~lolees bl Ethnic Gro~, Sex, and Age, 1968 in percerit) European Age Group African Males African Females Asian Males Asian Females European Males Females 15 - 19 1.9 17.4 6.2 17.9 2.0 6.0 20 - 29 39.6 54.9 32.3 58.9 13.7 33.7 30 - 39 36.2 19.8 30 .. 2 11.9 31.4 24.1 40 - 49 16.3 7.4 17.7 5.4 21.6 16.9 (i 50 - 59 5.3 0.5 12.7 19.6 16.9 60 and over 0.8 0.9 11.8 2.4 Totals absolute numbers 3560 408 434 112 102 ~ Source: Labor Force Sample Survey, January/February 1968. 1Imex Table 4.2. Ase Composition of Urban Population, by Ethnic Gr0l.lP an~ ,Sex, 1~62'<-'-r""~ :}.J.l 4,~ 5,.1 0.82 J~ 't':J, ~,~ ". ,1 4 " 9!l7 ~Q ,~ ,44 6.6 2.5 5.01 3.7 4.0 0.88 3.4 3.0 1.07 ,.S =49 4.3 l.4 $.19 SO~~ ?7 0,,9 ~.90 2.6 2.4 1.04 Sp ~ $9 l.~ 0.3 7.12 l.8 1.4 1.23 ,0.9, P,4 4tt73 1.9 1.6 1.13 6Q=~ 6~ ~ ;fJ9 0.3 O!l ~~P9 l·l 0.8 1,21 70 flld.Q"r 0.5 0.2 2.59 2.8 1.9 1.42 notsta~cl 1.1 1.0 2.10 0.5 0.5 0.84 Totals 100.0 100.0 1.81 100.0 100.0 0.94 11 l(al~b~>Cl~ • . YrQ~ ~~ ~,C;a.QPQP~~,Qn, ,~~~~~~ lrQ~e. . Sovrc, , J{~4 PRPyJ..§t-iQ~ ;O.q~u. 196~, YQ:J.. ;J:l~,t Mn~t9 pg1:l~@.~. RDI l~~~Jt! . ~" , • Annex Tab1e Size of Firm (Persons Plr1ploled ~ Age (years) 1-14 15-49 50-99 100 and more 15-19 1.0 20-24 1.0 2.1 25-29 1.0 2.1 3.5 1.9 30-34 2.1 3.2 2.9 2.8 35-39 3.1 6.8 5.5 6.2 40-44 7.0 10.6 7.2 8.1 45-49 6.0 12.5 8.4 10.3 50 and over 14.3 12.8 11.8 Source: Labor Force Sample Survey, January/February 1968. AmlexTable 1l'~8 .: of Urban Eaflot!eat" DJ-EthidcG'ioUp, Sex" and Age , 196 African African Asian Asian European European AP- (Years) Males F.ales Males Females Hales Females 15-19 0.0 0.5 20-24 1~0 1.2 2.0 25-29 1.7 1.7 1.8 1.1 3.0 1.0 30-34 2.4 2.3 3.1 2.5 2.7 1.3 3S-39 S.8 4.6 6.8 4.7 5.1 2.9 40~ 6.1 1.2 10.0 1.1 1.1 6.6 45-49 9.\9 11.6 U.8 3.0 11.8 9.1 SC> and over 12'.J 14.0 13.9 15.7 5.8 Source: Labor Force Sample Survey , J anuary!February 1968. Annex 'J.'able .h .. 9 : Avera e Educational Achievement (in Years) of African Hale Urban Emp 0lees,z bl Ase and Lensth of Service in the Empl0l!n~ Firm,z 1968 Len~th of Service (Years) !g~(Years) 1-5 6-10 11-15 16-20 21-25 26 and over 15-19 7.0 20-24 4.B 25-29 1.5 9.0 1.0 30-3b 1.1 1.0 4.3 4.0 35-39 6.1 5.9 5.6 5.6 5.0 40-44 5.5 5.0 5.6 5.6 B.1 B.3 45-49 5.3 5.0 4.5 5.1 6.1 7.4 50 and over 8.1 1.5 8.6 B.1 6.3 9.8 Source: Labor Force Sample Survey, January/February 1968. .AlIMa- Table 4.10: Interv:i.8wee'S Education (in Years) Ithnic Group/Sex 0-2 3-5 6-7 8-9 10-11 12-13 14 and more c-:: Ur1e~Hale8 'ather'sEducation 0.2 0.3 1.0 1.7 2.8 5.6 2.9 Mother I s Education 0.1 0.1 0.3 0.4 1.0 2.4 1.7 African Paules 'ather's Education 0.1 0.5 4.0 5.1 7.2 6.5 Mother's Education 0.0 0.1 1.9 3.0 4.2 3.0 Asian Hales 'ather's Education 0.3 1.7 3.8 5.3 7.8 9.0 11.1 "ot.h~r '8 Education 0.0 0.3 1.1 2.5 3.2 4.1 5.9 Aai8n.Fana1es 'ather's Education 7.0 7.0 9.5 10.4 13.4 Mother's Education 4.0 4.1 6.1 8.0 9.0 Euroeean Males Father's Education 0.0 4.0 12.0 11.5 11.9 12.1 Mother's Education 0.0 3.5 7.0 9.5 11.5 11.0 hroP8an .Females ~ii'ith8r' 8 Education 11.0 13.5 13.3 14.5 Mother's Education 10.0 12~5 13.1 12.5 Source: Labor Force Sample Survey, January/February, 1968. Annex Table ~.ll: Distribution of Urban Em~lOlees bZ Ethnic GrouE, Sex z and OccuEation z 190 Occupation I II III IV VI VII VIII IX X Totals V - African Males 49 28 35 67 23 481 630 1471 65 2850 African Females 2 50 1 68 4 156 3 284 Asian Males 16 18 1 36 27 165 106 31 4 404 Asian Females 7 40 1 59 3 2 112 European Males 12 6 2 11 38 12 8 4 1 94 European Females 1 39 -2 -2 34 2 1 87 Totals 18 61 167 121 94 819 748 1667 76 3831 I = Farmers VI = Administrative, Executive, and Managerial Worker II = Professional Scientific and Technical Workers vn = Clerical and Sales Workers III = Other Professional Workers VIII = Skilled Workers IV = Technicians IX = Semi-Skilled and Unskilled Workers v = Foremen and Supervisors x =< No answer Source: Labor Force Sample Survey, January/February 1968. Annex Table 4.12 I Occupation II III v VII VIII Age (Years) I ~ - IV - - !! - - -IX I 15-19 7.0 ,20..:24 7.0 25-29 11.0 8.3 8~"0 6.2 8.0 30-34 9.3 7.5 10.2 11.0 8.4 7.3 6.3 7.3 35-39 8.8 8.8 7.4 7.2 10.0 7.5 5.8 5.0 5.1 40-44 8.5 5.5 6.3 5.9 9.0 6.7 4.7 3.8 3.6 b5-49 2.5 1.0 7.8 3.4 3.0 7.0 :SOand oftr 9.0 3.3 2.4 7.1 I -Farmers VI = Admi.n1strative, Executive, and Managerial Wor,kers II - Prof'esScnal Scientific and Technical Workers VII = Clerical and Sales Workers .III = otner Professional Workers VIII = Skilled Workers IV = Teehclcians IX ~ Semi-5killed. and U~~~lled Workers V ! Ie Foremen and Supervi8.or s X =No answer Source: Labor Force Sample Survey, ci'anuary/February 1968. Annex Table u.13: Average Education of lvIale African Urban Employees, by Occupation and Size of l.i'irrn Size OI Firm Occupation: (Enployees): II III IV V VI VII VIII IX X 1 - 14 4.00 7.00 4.21 7.00 l5 - 49 1.00 12.27 7.26 5.10 4.94 5.54 50 - 99 9.09 8.33 8.70 9.40 8.04 5.89 4.78 4.79 100+ 13.75 13.96 6.77 6.23 13.29 8.96 6.03 4.65 6.10 Occupational Categories: II = Professional workers (scientific and technical). III = Professional workers (other). IV = Technicians. V = Foremen and supervisors. VI = Administrative, executive and managerial workers. VII = Clerical and sales workers. VIII = Skilled workers. IX = Semi-skilled and unskilled workers. X = No answer. Source: Labo.x Force Sample Survey, January/February, 1968. Annex Table 4.14: OccuEation I II III IV V VI VII VIII IX I Asian Interviewees 9.5 15.9 11.0 8.3 11.4 10.6 6.9 6.1 7.0 Fathers6f ,Asian Interviewees 6.4 12.4 13.6 10.5 9.6 6.8 8.5 3.7 4.8 4.6 Fm'opean Interviewees 15.8 17.0 13.0 11.1 13.3 13., 5.6 9.3 7.0 Fathers of European Interviewees 12.5 12.9 15.9 10.4 11.6 13.3 9.4 10.5 aO.9 I • Farmers VI = Administrative, Executive" and Managerial Workers I I - Profes.sional Scientific and Technical Workers VII = Clerical and Sales Workers III = Other Professional Workers VIII z Skilled Workers IV z Technicians IX a Semi-Skilled and Unskilled Wbrkers v• Foremen and Supervisors x = No answer Source: Labor Force Sample SUrvey, January!February 1968. Annex Table 4.15: Distribution of Urban Asian and Eur22ean ~~ by OWn Occupation and Father's Occupation Own Occupation Father's Occupation I II III IV V VI VII VIII IX I Totals - I 2 3 8 7 8 9 4 41 n 5 1 2 2 11 7 8 1 37 III 2 2 1 2 9 1 17 IV 3 1 4 V 1 2 2 5 2 1 13 VI 4 6 6 19 41 5 3 84 VII 7 3 1 12 8 56 13 2 102 VIII 5 2 6 7 21 48 12 1 102 IX 1 3 3 6 9 6 28 X ~ _7 1 -11 _9 ~ ~ 12 ~ 121 Totals 30 -1l! -~ 51 68 190 134 ~ .2 549 I • Parller s. II = Professional Scientific and Technical Workers. III z Other Professional Workers. IV = Technicians. V • Foremen and Supervisors. VI = Administrative, Executive arld Managsrial Workers. VII = Clerical and Sales Workerss VIII = Skilled Workers. IX = Semi-skilled and Unskilled Workers. X = No answer. Source: Labo':r Force Sample Survey, January/Febr.uary, 1968. Annex Table ".16: Cumulative Distribution, of Urban Employees by Length of Service, 19~3 and'19bB Length of Service in Years 10 and Number of Median Under 1 1 2 1 l! .2 6 I 8 2- over Observations (Years) Nairobi, 1953 (private sector) 48 68 80 86 89 92 94 95 96 97 100 31,200 1.1 Nairobi ,1968 (private sector) 6Y 21 31 39 48 54 58 64 70 76 100 2,545 4.4 Nairobi ,1968 (public sector) j.! 25 39 50 61 64 70 72 75 71 100 364 4.0 Hombasa, 1953 (private sector) 40 51 69 17 83 81 89 91 92 9.3 100 14,300 1.6 HQmbasa, 1968 (private secto?) 7Y 23 32 42 54 62 66 74 78 81 100 520 3.1 Non-Africans ,1968 (private sector, Nairobi) 7 25 40 49 57 62 66 10 72 15 100 467 4.1 Y Less than 6 months of service. Source: For 1968: Labor Force Sample Survey, 1968; for 1953: Report of the Committee on African Wages, Nairobi, 1954. Annex Table 5.1: Key for Variables Used Dummy Name of Number Dummy Meaning , 2 6 Public Sector: Age ~ lL: Age 1>-16: dummy for sector - public dummy for 1L years age group dummy for 15-16 " " " 7 Age 17-19: " "17-19 " " " 8 Age 20-2L: " "2C-2L " If rr 9 Age 25-29: " "25-29 n " II 10 Age 3C-3L " 11 30-3L " n " 12 Age L5-5L " "L5-5L " " " 13 Age 5L+ L± " "5L+ " If " Nairobi: dummy forcity - Nairobi Mombasa: dummy forcity - Mombasa Firm si ze 1 -lL: dummy forfirm size - 1 -lL employees Firm size l5 ... L9: dummy forfirm size - 15-L9 employees Firm size 100+ dummy forfirm size - 100+ employees Father i11it: dummy forilliterate father Father lit: dummy forliterate father Mother i11it: dummy forilliterate mother Mot her no ana: dummy formother's literacy - no answer 27 Kikuyu: dummy fortribe - Kikuyu, Meru, Embu 28 Kamba: " II " _ Kamba ~~ 29 Luhya nil" _ Luhya, Kisii, Kuria Coastal: ~~ ~i " II If - Mijikenda, Tai ta, Taveta Luo: " II " - Luo 32 Kalen "in: II " " - Kalen"in-s eakin tribes ~ 3 esc Division I: dummy for esc exam score - Division I pass « 35 esc Division II: dummy for esc exam score - Division II pass ~ ~.fkl 36 cse Division III: dummy for esc exam score - Division III pass 8 t18 37 CSC GeE: dummy for CSC exam score - GCE(O) pass ~ CI) 38 esc fail: dummy for CSC exam score - fail 1 39 Farmer: dumn)y for interviewee's occupation - farmer ~o Prof. 1: dummy for interviewee's occupation - professional (scientific and technical) Ll Prof. 2: dummy for interviewee's occupation - professional (other) Tech. : dummy for interviewee's occupation - technician Foreman: dummy for interviewee's occupation - foreman Adndn. : dummy for interviewee's occupation - administrative, executive and managerial worker L5 Clerk: dummy for interviewee's occupation - clerical and sales worker L6 Skill: dummy for interviewee's occupation - skilled worker • Annex Table 5.1: Continued Dummy Name of Number Dummy Meaning L7 Occ. semi-skill: dununy for interviewee's occupation - semi-skilled and unskilled worker ~8 Manufacturing: dummy for industry - manufacturing 50 Construction: dummy for industry - manufacturing 51 Commerce: II " " - commerce 52 Services: " " " - private services UN.lUN 53 Union dummy for union membership S6 Primary public: dummy for type of primary school - public 57 Primary private: dummy for type of primary school - private 58 Primary no ans: dummy for type of primary school - no r.-nswer 60 Sec. public general: dummy for type of secondary school - public general 61 Sec. public teacher training: dummy for type of secondary school ~ public teacher training Sec. private tech.: dummy for type of secondary school - private technical Sec. private general: dmnmy for type of secondary school - 'private gl9nera.1 Sec. private teacher training: dummy for type of secondary school - private teacher training 65 Sec. Harambee: dummy for type of secondary school - Harambee 66 Sec. not appli.: dummy for type of seconoary school - not applicable 67 Sec. no answer: dummy for type of secondary school - no 68 Farmer: dunuuy for father's occupation - farmer 69 Prof. 1: dummy for rather's occupation - professional (scientific and technical) 70 Prof. 2: dummy for father1s occupation - professional (other) 71 Tech. : dummy for father's occupation - technician 72 Foreman: dummy for father's occupation - foreman and supervisor 73 Admin. : dummy tor father's occupation ~ administrative, executive and managerial worker 7L Clerk dummy for father's occupation - clerical and sales worker ~nnex Table 5.1: Continued Dummy Name of Number Dummy Meaning 75 Skill: dummy for father's occupation - skilled worker 76 Semi-Skill: dummy for father's occupation - semi-skilled and unskilled worker 78 Addl. educ.: dummy for additional education taken ~ while working ~ 79 No addle educ.: dummy for no additional education ~o taken while working PZ ~~ 80 Addl. educe no ans: dummy for additional education - Hi ~E-4 81 Addl. training: no answer dununy for on..,.the-job training ~p:: 82 No addle training: dummy for no on-the-job training E-40 83 Addl. training no S answer: dummy for on-t.he-job trainning - ~ no answer • Variables~thdrawnbecause of linear dependence 11 Age 35-L4: dummy for 35-L4 years age group 16 Nakuru: dummy for city - Nakuru ~ 19 Firm Size 50-99: dummy for firm size - 50-99 employees 23 ,Father no ans.: dummy for fatherts literacy - no answer 25 Mother lit.: dummy for motherts literacy - Ilother literate 33 Other East African: dUmmy for tribe - other East African tribes or not identifiable or no answer 38 CSC fa1l= dummy for CSC exam score ~ fail L8 Occ. unskill: dummy for occupation - unskilled worker 52a Public services: dummy for industry - public services 58 Primary no ana.: dummy for type ot primary echool - no answer ,9 Sec. pub. tech.: dummy tor type ot secondary school - publicteehnical 77 Faocc. unskill: dummy tor father's occupation - unskilled worker 8e Addl. edua .no ans.: dummy for additional education - no answer 83 Addl. training no ans,: dummy tor additional training - no anewer Ll Mldpoint taken equal to 58 years of age. Annex Table 5.2: Kenya: Regression of Coe£ficients of Independent Dummy Variables; Dependent Variable: Monthly Earnings, African Males, 1968 f Independent Years of Schooling Variable, Xi c- 2 3 - 5 7 All 7F 7P 7Q 9 11 All 11 F 5 Age ~ 1L -281.2* -202.9* 6 Age 15 - 16 -LL5.2* -366.9* 7 Age 17 - 19 -!Ll.O -158.9 -2L8.L -218.0 -325.7 -511.2 -671.2 -967.L -999.3 8 Age 2C - 2L - 5C.0 -132.L -227.2 -166.9 -32C.6 -L5c.8 -5L1.2 -753.0 -831.2 9 Age 25 - 29 - L7.1 - 6L.L -125.8 -100.6 -2C2.3 -281.1 -LI7.L -5L2.L -6L9.2 10 Age 30 - 3L - 9.7* - 22.~ - 9.9* - 16 .. 9* - 22.~ -255.2* -266.1 -190.1* -328.9 * 12 Age L5 - 5L - 9.2* - 23.L* 110.6 23.3 289.1 -605.2 -159.1 -1,231.7 -1,263.6 13 Age 55 + - 21.C* - 36.~ l67.e-. 69.L* 2,102.L Intercept 335.0 389.9 505.2 L26.9 603.6 786.2 897.6 1,371.7 1,40).6 R2 0.01 C.05 o.lL 0.15 C.21 0.23 0.L3 0.17 0.18 Mean Earnings (Ksh/Month) 320.6 35L.7 LOl.3 3L2.l L38.5 L68.8 505.2 763.8 712.0 Mean Age (Years) 38.6 3L.o 28.8 28.6 28.2 27.3 26.8 25.L 25,.'L * .Not signi£icant. Note: The exam now given at the end of the seventh year is called the Kenya Preliminary Exam (KPE). A student can either .fail (7F), pass (7P) or qua1i£y (7Q) for secondary school on this exam. The eleventh year exam is called the Cambridge School Certificate exam (CSC) on which a student can score a Division I, II, or III pass, a GCE-C pass, or fail. (11 F) covers the last three categories, plus those who did not take the exam at the end of 11 years of schooling or who did not answer the question. Source: Labor Force Sample Survey, January/February 1968. Annex Table 5.3: Kenya: Regression Coefficients of Independent Dummy Val·lables. Socio-Economic Variables Held Constant; Dependent Variable:Month1y Earnings, African Males, 1968. Years of Schooling ___ _ __ Independent Variable !i. 0-2 l...:2 7 All 7Fl/ ~~L 7Q 9 11 11F --=- ;; Age-S' 14 -218.8* -200.3* 6 Age 15 - 16 -407.4* - )44.3* 7 Age l7 - 19 ~168.8 -135.0* -259.8 -218.4 -350.8 -496.8 -608.1 -1111.3 -1202.1 ~I 8 Age 20 - 24 -50,6 -128&0 -224.0 -175.4 -313.1 -385.1 -588.2 -111.1 -886.1 ~ 9 Age 25 - 29 -43.5 -64.9 -136.4 ·~91.0 -222.5 -289.0 -457.1 -519.8 -108.1 10 Age 30 - 34 -0.8* -24.9* -16.4* -24.5* -33.1* -164.2* -337.1 -186.2* -382.9* 12 Age 45 - 54 -8.1* -20.5* 125.0 40.~ 276.2 -542~1* -155.5* -1249.3 -1306.0 13 Age 55 + -18.9* -27.2* l85.~ l13.5* 1667.3 21 P'a tfier 111J..ter ate -66. 6* 134. 1* - 10. 8* =1i9 • 2* -6T.D* B73'. .5* 41.0* - 644~ 2*- 838.8* ~~ 22 Father literate 1.1* 123.1* 33.1* -42.8* 124.5* 914.3* 195.8* 146.5* 926.0* ~ ~ 24 Mother illiterate -44.9* -11.3 -14.3 -24.6 -118.0 -1.2* -15.5* 9.5* 14.5* 26 Mother - no answer -69.5* 81.7* -3.~ 137.2* 865.2* 90.9* 564.6* 790.9* 2TKi1ruyu--·- -35.~*- :'::33.2*- - -35-:-6* -137.7 - - 5.7"*- -220.4* -2-60~-2 ---- 22.2* 57.9* 28 Kamba -58.8 -20.3* 2.2* -101.6 64.1* -188.6* -312.6 -144.~ -11.1* ~I 29 Luhya -78.3 -44.7* -14.8* -122.7 23.~* -150.7* -356.3 -46.8* -89.3* S 30 ~oastal -116.8 -43.0~ -23.1* -97.4* 41.5* -117.1* 27.9* -187.5* -119.2* 8 31 Luo -85.6 -29.D* -16~~ -38.5* 15.1* -249.1* -274.0 11.1* 115.0* 32 Kalel~in 85.8* 22.5* -90.3* -269.6 -22.7* -16.1* 79.5* -13.1* 5.1* 68 Farmer- ---- -- -24.1 ~4.7*-- ----30.3"* ~8.3*- - 37.7* -83.8* 1.9.3*---09.0* 27.9* 69 Professional (Scien.) 262.9* 1.9* 2).5* 33.3* -91.7* -91.~ -71.1* 70 Professional (other) 0.9* 91.4* 94.3* -121.5* 110.0* 193.9* -138.2* -77.~ 43.9* 6 71 Technician 79.8* -27.4* 89.3* 34.1* 198.1* -612.3* -246.2* -195.~ ~ ~ 12 Foreman and supervisor 98.4* 3.3* 6.5* 120.5* -40.1* 394.5 80.1* 85.7* ~~ 13 Administrator 19.4* 144.8* 117.3* -61.4* 16.~ 342.~ -121.5* -115.~ -87.4* ~o 74 Clerical & sales work. 12.6 89.D* 42.5* -43.1* 74.0* 102.8* 0.9* -51.~ ~8 75 ~lled worker 45.9* 150.5 117.6 -70.4* 151.3 389.1* -2.3* 93.9* 29.0* 76 Semi-skilled 30.7* ge~ 19.2* 65.6* 25.1* 26.5* 95.5* 319.3* 434.1 Intercept 478.0 334.9 551.9 577.4 559.8 183.8 1034 . 1 644.3 502.5 R2 0.09 0.10 0.19 0.22 0.29 0.38 0.60 0.23 0.26 * Not si giii.fi cant . 1/ See Note, Armex Table 5.2 . Source: Labor Force Sample Survey, January/February., 1968. Annex Table 5.4: Kenya.; Means of IndepenaeI1l, Variables, African ¥ales t by Years Schooling, 1968 Years of Schooling Independent African Variables 0-2 3 - 5 7 Alia. Trl-2 ...1L -'1!i.. ..:L -1L llF All No. of Observations 1079 881 1109 302 496 60 174 220 173 3464 Average age (years) 38.6 34.0 28.8 28.6 28.2 27.3 26.8 25.4 25.4 32.9 5 Ag e s14 0.00 0.00 0.00 0.01 0.00 0.00 0.00 0.00 0.00 0.00 6 Age 15 - 16 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0000 0.00 . 7 Age 17 - 19 0.01 0.01 0.03 0.04 0.03 0.05 0.06 0.01 0.02 0.02 QI 8 9 Age Age 20 25 - - 24 29 0.05 O.ll 0809 0.20 0.26 0.30 0.26 0.28 0.29 0.32 0.30 0.40 0.44 0.23 0.50 0.37 0.51 0.35 0.18 0.22 10 Age 30 - 34 0.20 0.28 0.21 0.21 0.20 0.13 0.10 0.05 0.06 0.21 12 Age 45 - 54 0.22 0.09 0.03 0.02 0.02 0.02 0.02 0.00 0.01 0.10 1;2 A~e 52 + 0.07 0.02 0.00 0.00 0.00 0.00 0.01 0.00 0.00 0.03 I 21 Father Illit. .90 0.75 0.56 0.S7 0~54 0.57 o.lio 0.30 0.34 0.69 ~I~I 22 Father lit. 0.09 0.24 0.42 0.43 0-.45 0.42 0.57 0.60 0.54 0.29 24 Mother illi. t. 0.97 0.92 0.82 0.85 0.79 0.88 0.69 0.59 0.59 0.87 26 Mother no ans. 0.00 0.01 0.01 0.00 0.01 0.02 0.01 0.10 0.12 0.02 27 Kikuyu 0.27 0.27 0.35 0.37 0.36 0.38 0.36 9.49 0.43 0.32 28 Kamba 0.27 0.23 0.16 0.18 0.1$ 0.15 0.10 0.13 0.14 0.21 29 Luhya 0.14 0.21 0.17 0.16 0.20 0.22 0.24 0.13 0.1$ 0.17 ;1 30 31 Coastal Luo Kalenjin 0.10 0.10 0.01 0.05 0.14 0.01 0.06 0.14 0.02 0.07 0.14 0.01 0.04 0.14 0.02 0.03 0.1 'Z 0.02 0.03 0.14 0.01 0.03 0.10 0.02 0.03 0.12 0.02 0.01 0.13 0.01 ~2 8 Farmer 0.44 0.46 0.49 0.li8 0.52 0.53 0.47 0.53 0."50 0.47 69 Prof. 1 0.00 0.00 0.00 0.00 0.00 0.00 0.01 0.00 0.00 0.00 70 Prof. 2 0.00 0.01 0.02 0.01 0.02 0.03 0.03 0.03 0.02 0.01 :z; 71 Tech. 0.00 0.00 0.00 0.00 0.01 0.00 0.01 0.00 0.01 0.00 -~ (1)0 72 Foreman 0.00 0.01 0.02 0.01 0.02 0.00 0.02 0.01 0.02 0.01 ~~ ~o 73 Admin. 74 Clerk 0.01 0.01 0.01 0.01 0001 0.03 0.01 0.03 0.01 0.03 0.02 0.00 0.01 0.05 0.02 0.04 0.02 0.04 0.01 0.02 ~g 75 Skill 0.01 0.03 0.05 0.05 0.05 0.05 0.03 0.06 0.03 0.03 76 Semi-skill 0.06 0.09 0.J!3 0.1 0.12 0.10 0.1$ 0.10 0.09 0.10 Nairobi 0.7- 0.7 0.7 o. 0.77 O. 0 0.71 0.82 0.79 0.75 6 15 l-bmbasa 0.17 0.15 0.16 0.11 0.14 0.17 0.22 0.15 0.17 0.16 ~ 2 Public Sector 0.02 0.04 0.12 0.11 0.14 0.25 0.18 0.51 0.46 0.10 ~ 17 Firm. size I - 14 0.-60 0.01 0.01 0.01 0.00 0.07 0.00 0.00 0.00 0.01 o 18 Firm size 15 - 49 0.07 0.12 0.09 0.11 0.06 0.10 0.13 0.06 0.06 0.09 o o 20 Firm size 100 + 0.62 0.68 0.67 0.66 0.68 0.58 0.56 0.71 0.66 0.66 .Annex Tab~e 5.4: Kenya: Means of Independent Variables, African Males, bl Years of Schooling, 1968 (cont'd) Years of Schooling Independent Al-r:1.can Variables 0-2 3 - 5 7,AllY lFY ..1L J.£.... L .,- 11 I1F All 39 Farmer 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 40 Prof. 1 0.04 0.05 Os07 0.03 0.08 0.05 0.09 0.06 0.07 0.06 41 Prof. 2 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.01 0.00 ,.... 42 Tech. 0.00 0.00 0.01 0.02 0.01 0.02 0.01 0.02 0.02 0.01 "tt ..p, 43 Foreman 0.01 0.02 0.02 0.02 0.02 0.03 0.05 0.05 0.06 0.02 44 Admin. 0.00 0.00 0.00 0.00 0.00 0.02 0.03 0.04 0.03 0.00 5' 0, 45 Clerk 0.02 0.04 0.18 0.14 0.23 0.23 0.34 0.60 0.57 0.13: ~ 46 Skill 0.11 0.18 0.25 0.25 0.25 0.27 0.21 0.12 0.11 0.18 ~ 47 Semi-Skill 0.80 0.70 0.45 0.52 0.39 0.35 0.26 0.08 0.10 0.59 H 49 Manufacturing 0.50 0.49 0.42 0.34 0.4,' 0.30 0.39 0.27 0.28 0.45 ~ 50 Construction 0.07 0.09 0.07 0.06 0.07 0.13 0.03 0.01 0.01 0.07 ~ 51 Commerce 0.14 0.15 0.18 0.25 0.17 0.20 0.24 0.11 0.14 0.16 o 8·, 52 Services 0.19 0.15 0.11 0.19 0.07 0.08 0.03 0.01 0.01 0.14 5i Union 0.73 0.7~ 0.67 0.66 0.69 0.60 0.67 0.44 0.49 0.70 5 Primar,r public 0.05 0.2 0.39 0.34 0.L3 0.L3 0.39 0.42 0.4S 0.25 57 Primary private 0.10 0.63 0.51 0.51 0.50 0.47 0.44 0.32 0.31 0.40 78 Addl. edUce 0.03 0.05 0.21 0.16 0.26 0.37 0.41 0.42 0.43 0.14 79 No addle educe 0.96 0.95 0.78 0.84 0.73 0.62 0.57 0.49 0.46 0.85 81 Addl. training 0.05 0.09 0.25 0.18 0.32 0.35 0.43 0.36 0.32 0.16 82 No addle training 0.94 0.90 0.74 0.82 0.67 0.63 0.56 OQ55 0.57 0.82 34 CSC Ili.vision I 0.05 0.00 0.60 35 CSC Division II 0.16 0.00 0.01 36 esc llL vision III 0.20 0.26 0.01 - ~ H 37 csc GeE 0.08 0.10 0.01 ~ 60 Sec. public general 0.55 0.49 0.0$ g 61 Sec. public teacher training 0.00 0.01 0.00 ~ 62 Sec. private tech. 0.02 0.03 0.00 63 Sec. private general 0.17 0.20 0.03 64 Sec.• private teacher training 0.00 0.01 DaDO 65 Sec. Harambee 0.01 0.00 0.01 66 Sec. not applic. 0.01 0.62 0.02 67 Sec. no answer 0.18 0.21 0.20 Average earnings (Ksh/mo.) 320.6 354.7 401.3 342.1 438.5 468.8 505.2 763.8 392.4 712.0 Y The 7 All category includes 7F, 7P, 7Q, plus 251 observations who answered the KPE question tlnot applicable" or did not answer the question. The average income of this group is 383 shillings per month, somewhat higher than the 7F categor,r. ~ See Note, Annex Table 5.2. Source: Labor Force Sample Survey, January/February 1968. Annex Table 5.5: Kenya: Coefficients of Independent Variables, All Variables Held Constant, Dependent Variable Monthly Earnings, African Males, 1968_{Ksh!month) Years of Schooling Variable Group Variable !i 0 - 2 3 - 5 TFY 7P 7 All 9 11 5 Age ~ 14 -llO.3* -179.2* 6 Age 15 - 16 -216.4* -200.1* 7 Age 17 - 19 -126.3 -122.8 -131.4 -275.5 -223.3 -375.2 -884.4 AGE 8 Age 20 - 24 -54.8 -112.2 -123.1 -273.8 -200.3 -498.3 -540.3 9 Age 25 - 29 -48.4 -59.1 -59.5 -186.3 -121.1 -370.6 -414.9 10 Age 30 - 34 -2.1* -33.9 -10.8* -37.7* -19.9* -301.2 -33.8* 12 Age 45 - 54 -6.8 -3.8* 105.8* 311.6 167.8 -147.9* -516.5* 13 Age 55 + -7.7 ..-28.8* -3. 7* -5.~ 1269.4 21 Fath~r flIrt. -otr.-6-----'· a. 9* -60-:-0* _u :"~2~* -- -':'55.6* ---Ii9-=-3~--42r.13* LITERACY 22 FattM9r lit. -16.9 -2.1* -51.4* 20.4 -27.9* 40.8* 497.9* 24 Mother illit. -20.4 -34.D* 18.0* -73.8 -38.~ 28.9* 32.5* 26 l-1other no ans~. -61.8 16.4* 420.9* -180.7* 470.5* 334.9* 2r-KiIroyil --:sr.-9------=.5(.lr--...57-.-2*----~7·~·5*- - ...6.6*- ---=270~7--- -17.5* 28 Kamba -48.0 -42.0* -37.9* 49.1* 8.2* -306.1 -132.3* TRIBE 29 Luhya -61.3 -60.1 -41.4* 43.7* 15.9* -30a.3 44.4* 30 Coastal -66.0 -45.0* -43.2* -64.1* -44.0* -55.5* -280.~ 31 Luo -37.3* -40.7* 37.7* 11.4* 12.9 -233.8 -120.5* 32 Kalenjin 83.1* 38.3* -234.7 -2.3* -61.~ 289.3* -148.5 68 Farmer 23.2 1.5*--~. .3* - ---~4-:9* 17.9* 4.6*---- ~~o. 69 Profes. 1 168.2* 7.4* 29.7* 1.~ 96.9* 465.8* FATHER's 10 Profes. 2 -24.7* 51.5* -57.1* 13.8* 32.0* -189.2* -10.3* OCCUPATION 71 Tech. 38.5* -91.0* 134.1 209.8* 165.1* -569.6* -193.8* 72 Foreman 7.3* -17.8* 6.1* -88.5* -31.5* 225.8* 16.7* 73 Admin. -17.5* 105.7* 7.1* 6.3* 122.9* -72.8* 2).8* 74 Clerk -11.7* 58.4* -25.4* 12.2* -13.1* 4.9* 174.1* 75 Skill. 45.3* 113.0 4.1* 108.7* 83.3 44.~ 32.0. 76 Semi-skill. 22.5* 6.2* 70.5 29.~ 20.1* 8901* 325.9 14 Nairobi 18.5* 32.5* 21.6* 38.7* 39--:-0*----- -):1*- -- - [2.2* 15 Mombasa 21.1* -16.1* 3.3* 110.3* 89.4 36.3* 514.1* OCCUPATION 2 Public sector -238.3 -210.3 -124.6 35.3* 36.6* 145.1* 792.6 17 Firm size - 1-14 -96.8 -28.8* -54.1* -277.9* -124.7* 18 Firm size - 15-49 -55.9 64.6 -85.4 -99.3* -90.2 20.1* 39.4* 20 Firm size - 100+ 44.0 74.3 3.7* -105.6 -51.9 2.~ 63.6* Annex Table 5.5 (Cont.): All Variables Held African Males, Years of Schooling Variable Group Variable !i 0-2 3 - 5 7F 7P 1 AllY 2 11 --"! 39 Farmer 159.9* -73.1* 40 Profes. 1 7.7* -28.7* 121.8* 134.5* 80.9* 563.9 -155.5* 41 Profes. 2 22.9* -375.5* -238.9* 166.9* OCCUPATION, 42 Tech. 220.1* 227.9* 222.7* 117.8* 304.0* 453.5* 43 Foreman 160.8 23.~ 524.3 229.9* 212.3 738.1 182.8* cont'd. 44 Admin. 180.7* 364.8* 482.5 1162.7 337.8*, 45 Clerk 104.4 45.8* 213.9 156.3* 85.~ 478.2 -183.3* 46 Skill 52.2* 13.1* 166.7* 35.7* -23.1* 399.0 -189.9* 47 Semi-skill. -47.1* -64.~ 49.4* -44.7* -115~7 258.1* -212.5* 49 Manufacturing -69.5 -101.0 -68.~ 22.~ 42.1* 132.4* 390.6 50 Construction -118.5 -121.4 -102.4* 43.2* 13.9* 207.2* 41.9* 51 Commerce -47.~ -88.3 -28.4* 17.3* 42.2* 167.7* 401.5* 52 Services -73.4 -125.3 -138.6 43.5* -12.9* 74.8* -253.0*- 53 Union 86.2 67.1 60.9 80.2 111.4 204.5 111.7* EDtreATIOO 56 Primary public 14.1* 14.5 -13.1* -86.9* -3.0* ·~47~9* -ltj4.b 57 Primary private -11.0* 25.8 -41.3* -120.4 -27.9* 89.5* -127.4* 78 Addl. educe 187.8 105.7 19.3* 134.7* 162.4* -201.6* -146.2* 79 No addle 35.2 65.8 -27.5* 110.9* 128.2* -259.4* -104.3* 81 Addl. training -130.7 -8702* 31.2* 433.8* -299.3* 479.7* 389.8* 82 No addle training -238.3 -125.8* 346.6* -365.7 493.~ 115.4* 34 CSC - Division I 122.8* 35 cse - Division II 58.1* 36 esc - Division III 48.8* 37 esc - GeE 128.3* 60 Sec. public gener:al -513.4 61 Sec. public teacher tr. -187. 2* 62 Sec. public tech;. -697.6 63 Sec. private general -187.8 64 Sec. private teacher tr. -905.0* 65 Sec. Harambee -568.3* 66 Sec. not applic. -708.8* 67 Sec. no answer -536.3 Inte. ~cept 658.7 49~.~6 393.4 115.3 709.1 34.1 416.5 R 02. 3 . 0.56 0 . 44 0.39 0.78 0.55 * Not significant. 2/ See. note~Ann.ex -Table 5.4 1/ See note, Annex Table 5. 2 . Source: Labor Force Sample Survey, January/!"ebrnary 1968. Annex Table 5.6: Years of Schooling 1/ 1/- Age o- 2- 3 - ~~ 7 9 11 15 .- 16 301 292 17 - 19 291 347 522 20 - 2h 457 650 710 25 - 29 688 1203~- 12487} 30 - 34 1053 1103* 15801t- 35 - 41.1 1501 1374 1464 45 - 54 115t,', 1610~t- . 1605~} 55 + 103Q1t- 950* 981* 11 There are no ~igni£icant coefficients in the regression estimates for the 0-2 and 3-5 schooling categories. The number of observations is 42 and 45 indivi- duals , respectively. * Not significant. Source: Labor Force Sample Survey, January/February 1968. Estimated as in Table 5.1. Annex Table 5.7: _K_eny_a_:~~~~~_ _~_~~~~~~~~~_... .. lvIeans of Independent Equation Equation Equation Variables Variables Xi (1) (2 ) (3) Af;rican Hales Age (years) 27.9 l\ionthl earnin 1121.9 Hale Asian .1~~- 27 .J~I- 0 IJlale European 2783.0 3098.7 H Female African 447. 7~~ -121.91(' ~&.i -, ... t Female Asian -681. 2~~ -119.11*" c-l t:::::l \1) r-,~ Female Curo ean • 7~~- 35 • 8~~- 1012.0* Age 17 - 19 -97 . v ;\ -1019. ~~ - 91.9~~- 0.00 Age 20 - 24 -859J~ -1108.1 -1055.1 0.47 Age 25 - 29 -602.3 -693.9 -698.9 0.22 rrl ('j Age 30 - 34 -'<75.91;: -223. 8~*" 308.CH'c- 0.11 ;{- H"d Admin. 198.8->;;. 0.06 8- .:;x:.+> Skill. 1h.4~~ 0.06 ~ § Semi-skill. 40. 3~~- o C) (.) Unskill. -2211.2~~ 0 Primary public -lh3. 31~ 0.h4 Primary private -587.9 0.22 Sec. public general 377 • 3-l~ 0.53 Sec. public teacher training -515 .l~!- Sec. private tech. Sec. private general 518. 51~ 0.11 Sec. Private teach training 462 • 6~~- Sec. Harambee 6 H Sec. not app1ic. 72.1 E--I Sec. no answer 398.5-l{- 0.33 (3 369.8* 0.36 Addl. education 8 r.cl No addle educt Sh8. 3->;~ 0.36 Addl. training -59 .l~~ 0.25 No addle training ~~60 .6-l~ 0.47 z I Citizen other E.A. rLl "'J Citizen non-African -320 .S~i- HP-t Applied Kenya -1.\.95 .4~~ ~~ ou) Citizen no answer -69 .J~:- Intercept 1655.3 1637.5 384.3 R2 0.64 0.69 0.77 Number of observations 135 36 Source: Labor Force Sample Survey, January/B"ebraury 1968. ~*" Not significant Anne~ Table 5.8: Kenya: Regression. Coefficients of Independent Dummy -------- Vari.ablesj Dependent Variab1elMonthly Earnin~~ All ~e~-ethnic_gFouPs with 17 Years of Schooling, l~ J _ .and Means of Tar:.re.T:>les for Arrican ~les Means of Independent Equation E9.uation ~uation Variables VariabIes Xi (1) (2) 3) African Males Age (years) 30.4 Monthll Earnin~s (Ksh) 2429.4 Male Asia.n -936.2* -946.2* -185.6*- I~ Male European 3409.0 2842.0 2576.8 Female African ~I~ P::I Female Asian -1907.1* -1951.~ -577.9* Female European 737.2* -801.0* -388.0* Age 17 - 19 6.00 Age 20 - 24 -1303.1* -1313.0*- -1615.9* 0.33 Age 25 - 29 -1472.5 -1412.~ -482.8* 0.40 Age 30 - 34 -791.0*- -750.7* -285.4* 0.43 ~I Age 45 ~ 54 5.2* 394.1* 829.9* 0.00 A~e 55+ -1495.3* -2410.1 -18.9* 0.00 Father 1~ t .• -78.8* -296.8* 0.50 I I>-t Father no ans. 4815.7* 5240.1* 0.30 ~Iul Mother lit. -505.6* -728.7* 0.40 ~ ~ Mother no ans. -5383.8* -8065.6* 0.30 Farmer -1755.0 -1094.6* 0.40 Prof. 1 1501.6* 803.4* 0.00 ooe5 Prof . 2 48.6* -1384.1* 0.07 - H ~8 Tech. 0.00 Foreman -1601.~ -2215.5* 0.03 ~~ <0 Admin. -1241.1* -125L.8* 0.03 r:x..g Clerk -1625.3* -1301.4* 0.07 Skill. -846.6* -2424.4 0.07 Semi-skill. -497.6* -1446.2* 0.13* Union 1181.0* 0.10 Public sector 8096.2* 0.90* Mombasa 932.8* 0.07 Nakuru -937. 6~~ 0.00 Firm size 1 - 14 -4430.9~E- 0.00 Firm size 15 - 45 -5842.0 0.07 Firm size 100+ -6604.7 0.87 Manufacturing 4915.8 0.07 tS Construction 6334.8 0.00 ~ Commerce 6235.7 0.03 « Services [?j 9642.7 0.00 () Farmer 0.00 8 Prof. 1 -1634-.5* 0.17 Prof. 2 383.7* 0.50 Tech. 148.1* 0.00 Foreman 5199.6 0.00 Admin. 2439.6 0.10 Skill. 0.00 Arulex Table 5.8: Ke a: Re ression Coefficients of Inde endent Dumm (Cont'd.) Variab es; Dependent Varia e:Monthly Earnings, All Sex-Ethnic Groups with 17 Years of Schooling, 1968. and Means of Variables~for African Males Means of Independent Equation Equation Equation Variables. Variables: ~ (1) (2) (3) AfrICanMales 8m-skill, 0.00 UneJd.l1. 000 Primary public 1321.6* 0.30 Primary Pl'iTate -1855.7 0.37 Sec. public general 1315.9* 0.50 Sec. public teacher 782.0* 0.00 training Sec. private tech. -1412.7* 0.00 Sec. private general 2063.9* 0.13 Sec. private teacher training 0.00 Sec. Harambee 0.00 Sec. not app1ic. 3163.44* 0.00 Sec. no answer Cit zen other E.A. Citizen non-African -1114.08* Applied Kenya -982.63* Citizen noans. m ve~,s :y 'N ro 25,88.52* 0.07 Univer~Jity other E.A. 143.22* 0.13 University no ans. -16)0.20* 0.07 Faculty soc. sci. 14)0.10*- 0.67 Faculty medicine 2495.1~ 0.00 Faculty engineering 3305.49 0.10 Faculty agric. Faculty arts -664.09* Faculty other 1450.80* Faculty other -4427.78* Faculty other 31,29 88 a 17 In~rcept 3404.7 4746.2 1044 •.5 R 0.44 0 •.56 0.89 Number of observations 104 30 Sources Labor Force Sample Survey, January/February 1968. Annex Table 5.9: Kenya - unadjusted Age-Earnings Profiles bz Years of SChoOliDf' ill 5ex- Ethnic Groups, 1968 Kshlmontli) Years of Schooling Age -1.L 15 - 17 17 - 19 1,353* 20 - 24 1,473 2,836* 25 - 29 1,730 2,667 30 - 34 2,O5~ 3,348* 35 - 44 2,332 4,139 45 - 54 2,997 4,144 55+ 1,694* 2,644 * Not significant. Source: Labor Force Sample Survey, Januar,y/February 1966. Annex Table 5.]0 Kenya: Wage Employment and Average Earnings by Major Sectors ,1961 Employment Average (in 1,000) Earnings CIa, p.a.) Agriculture and foreatry Lar ge farms 159'.2 66.5Y Farmers' co-operatives 12.5 Settlement schemes 18.3 Public sector element 6.9 Small-holdings outside 'settlement schemes 28h.7 11.9 Private indus.ia'y and c,ommerc.e Mining and quarrying 2.4 291.7 Manufacturing and repairs 58.1 296.0 Building and construction 21.4 141.2 Electricity and water 3.8 368.9 Commerce 43.6 495.4 Transport and Oommunication 21.5 334.9 Other services ~l. 7 228.2 Non-agricultural rural occupations 60.6 44.6 Public sector East African Railway and Harbors 2S.i4/ 3~1.B) Other government sern cas 159.9~/ 275.B) AllQwance f,or ,under-enumeration 33.9 106.2 Total wage eil1ol!ent it}16 •.6 151.8 {of which rican 922.5 2/ 11.a. ) (of which urban 281.8~/ 31B. 3) (of which urban African 236.63. n.a. ) 1/ Applies to total of first foutsub-items. 2/ 1966 figures. Sources: Republic of Kenya, Statistics Division, Mini5~ry of E;conomic Planning and Development, Statistical ,Abstract, 1967, Economic Survey, 1968. \ \ Annex Table 2. .11: Kenza: Mean Annual Gross Rural Fanti.lz Income bZ A~e and Schoolins of Household Head, 1963 A. Income in Ksh Years of Schoolins Age o Illit. o Lit. +,,-3 4 - 8 ~ 1/ 19- 833 833 833 833 833 20-29 1201 1182 1168 3069 4297 30-39 1414 1820 2111 3311 6162 40-49 1724 1980 2904 6721 4971 50-59 1876 3922 4014 5784 ) 8820 9692 ~ 60-69 1791 2364 3733 ) 70+ 1901 2054 4142 B. Income as Pe~cent of Mean s~oolinB GrouE Income Years of Schooling Age o Illit. o Lit. 1 - 3 4 - 8 ...2.!-. 19 0.49 0.33 0.25 0.20 0.13 20-29 0.71 0.46 0.35 0.75 0.68 30-39 0.88 0.71 0.63 0.80 0.98 40-49 1.01 0.78 0.87 1.63 0.79 50-59 1.10 1.,4 1.20 1.40 ) 60-69 1.05 0.93 2,64 1.54 ~ 0.91 ) 70+ 1.11 0.80 1.24 Y The means at; all educ,ation levels taken equal to the mean of age 19 of those in the sample with 4-8 years of schooling. • Y Mean incomes in thi s table taken as a ratio of tlgross family income" for the same level of schooling shown in Table 5.6. Source: Survey of Central Province. • Al"..nex Table 6.1: Kenya: Average Delay (in Years ) Between Leaving School and Starting \'iork of African ~1ale Urban ~loyees, by Level of Education and Year of Leaving School Education (in Years) (parentheses contain nTh~ber of cases) Year of Leaving School 0-2 3 - 5 6 - 7 8 - 9 10 - 11 12 - 13 14 and more 1966 1.0 1.4 1.1 0.9 0.7 0.1 (3) (37) (24) (56) (6) (7) 1965 1.0 1.7 1.1 1.0 1.0 0.0 (1) (64) (18) (33 ) (1) (1) 1964 2.0 3.1 1.9 1.4 1.3 0.0 0.0 (2 ) (11) (70) (11) (23) (2) (2) 1963 2.0 1.9 2.2 2.1 1.2 1.0 0.0 (1) (7) (61) (9) (14) (2) (2) 1962 3.0 1.4 1.9 1.3 1.4 1.0 0.0 (4) (8 ) (57 ) (15) (11) (1) (2) 1961 2.9 2.7 1.7 1.4 0.0 0.0 (18) (51) (10) (15) (1) (1) 1960 4.7 3.0 2.4 1.8 1.5 (3) (20) (55) (10) (15) 1955-59 5.6 3.0 2.6 2.3 1.8 2.j 0.3 (9) (148) (291) (16) (16) (3) (3) before 1955 4.1 3.4 2.7 2.1 1.8 2.7 0.3 (134) (583) (346) (40) (12) (3) (3) Source: Labor Force Sample Survey, January!February 1968. Annex Table 6.2: Kenza: Income Tax z Surtax and Graduated Personal Tax in Relation to Income and Fami1l Size, 19b8 (Ksh/month) Fami1Z status: In~ Single (5 ) --- Married no Ch. M. 1 ch. M. 2 ch. M. 3 eh. M. 4 ch. + ~ 160 2 2 2 2 2 2 161-240 6 6 6 6 6 6 241-340 9 9 9 9 9 9 341-359 13 13 13 13 13 13 360-520 13+0.125y 13 13 13 13 13 521-700 20+0.125Y 20 20 20 20 20 701-800 30+0.125Y 30 30 30 30 30 801-860 )0+0.125Y as (5) 30 30 30 30 861-1000 40+0.125Y as (8 ) 40 40 40 40 1001-1200 50+0.125Y as (5 ) as (8 ) 50 50 50 1201-1400 50+0.125Y as (5) as (8 ) as (s) 50 50 1401-1600 50+0.125Y as (s) as (5) as (s) as (5 ) 50 1601-1666 50+0.125Y as (5 ) as (s) as (8) as (5 ) as (5 ) 1667-3J32 50+0. 275y as (s) as (s) as (s) as (8 ) as (s) 3333-4999 50+0. 375y as (5 ) as (s) as (5 ) as (s) as (s) 5000-6666 50+0.475Y as (8) as (5) as (s) as (8) as (S) 6667-8333 50+0.,25Y as (s) as (5) as (s) as (8 ) as (s) 8333-10000 ,o+0.575Y as (5) as (5) as (s) as (s) as (8) 10001-11666 50+0.625Y as (5 ) as (s) as (5 ) as (8 ) as (5 ) 11667-16666 50+0.675Y as (5) as (5 ) as (5) as (5 ) as (6 ) 16667-25000 50+0.725Y as (5 ) as (5) as (8) as (5 ) as (5 ) OVer 25000 50+0. 775Y as (6) as (5) as (5 ) as (5) as (s) Note: If single per,sons are entitled to child allowance, the amount is 720 Ksh/month. The $ame allowance applies tor men olde~ tnan 64 and women older than 59. Ohild allowance only up to 19 years (except lfh~n receivil1g full-time education). No allowance if chilct's income exceeds the gaunt of the allowance. Source: Republic or Ke:PYQ1l .}fi~etr. y ot O~I'(lfj lUld Industry: A G\liae to Industrial In'Vestaent, Na1I'ODj., l'9f>B. Annex Table 6.] Kenya: Averase NUmber of Children for Urban African Males a I9~B Education ~in years) Age 0 1 - 5 9 - 12 13 + 25 - 29 1.200 0.1,55 0.0)0 0.072 30 - 3L 0.973 1.2,58 0.695 0.2L3 0.095 35 - 39 3.030 3.112 2.586 1.93$ 0.87$ La - LL L.895 L.97L L.951 5.632 3.000 L5 - L9 5.703 5.7Lo 6.3L5 7.333 (2.500) 50 + 5.9L6 6.576 (lL.ooo) (2.000) Source: Labor Force Sample Survey, January/February 1968. Figures in brackets refer to single observations. Armex Table 6.4: Ken a: Avera e Taxes for Af'ricanMales by Age and Years of Schooling, 1968 (Ksh month) Age 0-2 3 - 5 7 7F 7P 7Q 9 11 13 --11 17 - 19 6 6 9 6 9 9 6 63 20 - 24 9 9 9 9 9 9 13 97 130 25.: - ~9 9 9 63 9 63 76 73 134 182 589 3.0. - 34 9 13 13 13 13 20 20 185 233 750 35: - 44, 9 13 13 13 20 30 40 222 257 1.327 45' - 54 9 13 20 13 40 30 40 50 688 1327 55 ok 0 13 20 13 40 30 40 50 688 ·1327 Not.a:; Categories 7F" 7P' and. 7Q. refer to those who fail, pass, and qualify, respectively, on the Kenya PrL"Ilary Exam. &ource: Table 5.1 and Annex Tables 6. 2, 6.3, and 6.5 Armex Table 6.5: Kenya: Average Taxes for Afri~an Males by Age, 13 and 11 Years of SChooling, 1968 (in Kenyan shillings per month) ~ Years of Schooling 13 17 20 - 24 130 25 - 29 182 589 30 - 34 233 750 35 - 44 257 1,327 45 - 54 688 1,327 Source: Annex Tables 6.2 and 6.3 Annex Table 606: Kenya: Average Rates of Return to- Higher Secondary and University Education, African Males, 1968 Private Social Years of Adjusted Unadjusted Schoo1ins Unadjusted fo~ Taxes and Adjusted II - 13 23.8 22.9 14.7 13 - 17 27.4 19.9 8.8 Sources: Tables 5.5 and 6.,1 and Annex Tables 5.9 and 6.5. A.nnex Table 6.7: Kenya: Model Life Table Functions for the Africa.n Population of Kenya Life Table Age - Specific Probability of Age PO]2ulation Mortality Survi val of a 15 Year Old 15 1,000 20 960 0.0082 0.96 25 907 0.0113 0.91 30 856 0.0111 0.86 35 804 0.0124 0.80 40 750 0.0140 0.75 45 690 0.0161 0.69 So 621 0.0209 0.62 55 540 0.280 0.54 60 449 0.0310 0.45 Source: Calculated from Kenya Population t:enslls, 1962, Vol. III, African Population, Nairobj., 1966. ". Annex Table 6.8: -- Kenya: Average Differential Private Rates of Return to Schooling, by Yea~ of Schooling, African Males, 1968. When Incomes When Incomes When Incomes Unadjusted for Adjusted for Adjusted for Years of Variables other Socio··economic all Schooling than Age Variables and Age Variables Primary ? - 4 25.6 15.6 negative 2 - 7rl 17.8 10 ..5 0.3 2 2 - 7 all 32.7 18.5 10.0 1 5- 7F 10.2 6.1 16.2 5 - 7 p3 57.6 27.7 64.4 5 - 7 Q4 64.3 50.8 2 5- 7 all 55.1 21.7 Secondary 8 - 95 7.3 13.6 8 - 96 23.6 36.1 7.1 8 - lla 7 27.3 31.5 8 - 118 36.1 36.8 10 - lla 9 43.8 32.1 10 52.2 39~·2 23.7 10 - 11 Notes: 1. Represents rate of return to those with 7 years of schooling but failing the KPE over those with less schooling. 2. Represents rate to those with 7 years of schooling, including all those who failed, passed, qualified, or did not answer the KPE question, over those with less schooling. 3. Represents rate to those with 7 years of schooling who passed the KPE over those with 4 years of schooling. AnnexePAhle 6. 8 .1 '8Qya;Average Differential Private Rates of Return to SChoolin~, by Year of Schooling, African Males, 196 - Continued 4. Represents rata to those With 1 years of schooling who qualified on theKPE over those with 4 years of schooling. 5. Represents rate to those with nine years of schooling over those with 7 years mlo qualified on the KPE without continuing their education. 6. Represents rate to those with 9 years of schooling over the average of those with 7 years. 7 • Represents r'ate to those with 11 years who passed CSC below Division II over those with 7 years who qualified on the KPE. 8. Represents rate to all those with 11 years over all those with 7 years. 9. Represents rate to those With 11 years who passed esc below Division II over all those with 9 years of schooling" 10. Represents rate to all those wi th 11 years over all those with 9 years. Source: Column (1): Table ·$ ..1 and Table 6.1 Coltmlll (2)-: Table 5.2 Column (3): Table 5.3 Annex Table 6 •. 9.~ Kenya: Average Differential Social Rates of Return to Schooling, by Years of Schooling, African Males, 1968 When Incomes When Incomes When Incomes Unadjusted for Adjusted for Adjusted for Years of Variables other Socio-economic all Schooling than Age Variables and Age Variables Primary 2 - 4 16.4 12.2 negative 1 2. - 7F 12.9 negative 2-7Al1 2 21.7 B.8 5- 7Fl 8.6 10.4 5 - 7p.3 38.8 22.7 44.0 5- 74 Q 46.1 38.7 5-7Al1 2 38 .)~ 18.0 13.9 Secondary c 8 - 9;) 5.0 9.0 6 8 - 9 16.3 20.}~ 4.8 8 - 11a 7 18.2 17.5 8 - 11 8 23.6 2h.2 11.4 9 10 - lla 28.3 10 - 1110 33.5 17.8 Notes: See Annex Table 6.8. Source: See Annex Table 6.8. Annex Table 7.~. ~enya; Salary Scales for the Teachin~ dervice (Ia, per year) Category Old SealeY Actual Average Ne:;: ScaleY Ungualified without KPE 84 .. SalariesY 90 84 Unqualified with KPE 96 10).!! Unqualified with KJSE .108 117 9r# Unqualified with CSC 240 252 108 Unqualified with HHy 240 one PP- 300 330 two p~ 350 366 n.a. n.a. Qualified PH and AT! II without KPE 120-6-18<0' 135-6-177-9-231 179 P3 and AT I I I with KPE 162-6-180-12-264 180-9-216-12-360 204 P2 and TI I 240-12-276-18-456 264-12-300-18-480 301 PI and SATI 348-27-726 378-27-756 41171 Sl, AEO and Tl 582-24-750-25-800-30-830-35-1110 684-27-981-30-1011-36-1119 582- STl 1110-40-1230 1155-42-1239 n.a. '1}I 750-25-800-35-1110-40-1350-50-1450 810-42-978-45-1158-48-1446 n.a. Graduate Teacher 804-43-976-44-1108-46-1292 810-42-978-45-1158-48-1446-51-1548-54-1710 n.a. AL II 8~~43-976-u4-I108 ..1.16-1292 810-42-978-45-1158-48-1302 n.a. ALI 1154-46-1338-52-1442 1158-48-1446 n.a. Lecturer 1494-52-1598-56-1710 1494-51-1548-54-1710 n.a. Se...~or Lecturer 1839-75-1989 1839-75-1989 n.a. 1/ In force from April 1, 1964. ~/ As proposed in Sessional Paper No. 11 of 1967. 3/ All teachers employed in the primary school's of Kiambu and Nyeri counties in 1967. 4/ PP = principal pass. ~/Read: Kls120 increasing in steps of IG,6 to rn.80. "5/ Not included 17 teachers wi th K1J.02. Y One teacher only. ATI = assistant technical instructor. TI = techn. instructor. SAT! = senior techn. ass. instructor. STl = senior teehn. instructor. n1 technical master. AL = assistant lecturer. AEO = a~sistant education offices. Source: Ministry of ]ilucation--and Ministry of Local Government. ,; Annex Table 9.1: Kenya: Average Annual Earnings. 19$7-1966 (in Ki.) !2ll ~ ~ 1960 ~ 1962 !2El ~ ~ 1966 1951-100 1963-100 Camnerciel. Agriculture and Forestry African 34 35 35 37 39 39 46 51 52 54 159 117 Asian 400 400 500 500 667 667 571 571 714 624 156 109 European llll llll 1176 llO5 1313 1214 1143 1417 1545 1583 l42 13B Total 1a 43 43 46 49 47 55 61 62 65 155 UB Private Industry and Caamerce African 80 83 86 92 l~ no 132 124 138 148 185 112 Asian 446 452 454 457 460 485 514 523 569 628 141 122 European llOS 1147 1150 1179 1265 1321 1422 1480 1594 1719 161 125 Total. 189 198 205 212 235 239 276 244 263 287 152 104 Public Service African 85 91 94 102 113 121 138 172 196 212 249 154 Asian 560 566 559 559 607 608 65B 756 810 892 159 136 European 1196 1228 1256 1270 1506 1538 1574 1638 1667 2405 201 153 Total 177 189 192 200 223 222 232 239 259 289 163 125 All Employmellt African 60 63 64 68 76 80 95 103 124 133 222 140 Asian 478 485 486 488 511 529 561 575 622 680 l42 121 European 1143 1177 1194 1213 136.5 1399 14.52 1516 1612 1940 170 134 Total 125 131 133 136 152 153 172 178 192 212 170 123 Source: Calculatad from data or the statistical Abstract, 1967 Nairobi, 1967 pp. 150-151. J Note: Extreme fluctuations in the annual earnin,gs of Asians and Europeans are mai..1'l1y due to rounding of ~absolute amounts. Annex Table 9.21 Employment by Ethnic Groups and Major EconClDd.c Sectors I 19.57 ... 66 (in 1000) 1957 19.58 1959 1960 1961 1962 1963 1964 1965 1966 .Agriculture and Forestry African 2.51.1 247.2 249.4 269.1 249.8 243.5 217.6 206.8 202.8 204.4 .Aaian 0.5 0.5 0.6 0.8 0.6 0.6 0.7 0.7 0"7 •• 0.8 Eqropeap 1.8 1.8 1.7 1.9 1.6 1.4 1.4 1.2 '~1.1 1.2 Total 2.53.4 249.5 251.1 2n.8 252.0 245.5 219.7 208.7 204.6 206.4 Priv~:teIndustry and Commerce African 156.8 149.6 148.0 151.1 134.1 133.2 121.8 169.2 169.0 173.4 Asian 25.8 24.8 25.1 25.6 25.0 23.5 24.3 28.3 27.6 29.0 European 1l.4 11.6 12.0 12.3 11.1 10.6 10.2 10.2 9.6 9.5 Total 194.0 186.0 185.1 189.0 170.8 161.3 156.3 208.3 206.6 211.5 hPUc Service 4Crican 146.9 131.0 140.0 140.1 145.9 141.2 139.5 161.8 161.2 113.6 Asian 10.9 10.6 11.1 11.8 12.2 12.0 11.7 8.2 1.9 7.4 ElWopean 9.2 9.2 9.0 8.9 8.9 7.8 6.1 4.1 4.5 4.2 Total 167.0 151.7 160.1 161.4 167.0 161.0 151.3 174.1 119.6 185.2 All El!Plo1Dlent African 554.8 534.1 531.4 560.9 529.8 523.9 418.9 538.4 539.4 551.0 Asian 31.2 35.9 36.8 38.3 31.8 36.1 36 .. 1 31.2 36.2 31.2 European 22.4 22.6 22.7 23.0 22.2 19.8 11.1 16.1 15.2 14.9 Total Note: Over time the coverage has ehanged and therefore the figures are not strictly comparable. The rule between 1963 and 1964 indicates a break in the series caused by a marked improvement in the coverage of private industry and commerce when an approximated 42,000 employees were added to this sector as a resul t%'f survey of establishments in rural areas. In the analYSiS, the 1964-66 employment figures for Africans in that sector have hence been reduced by 40,000. Source: Statistical Abstract 1967. " AImex Table 9.3:- 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 Agriculture and Forestry!! 33.5 34.6 35.6 40.0 38.7 40.0 44.4 48.0 44.4 50.9 Private Industry and Commerce 100.5 100.4 104.6 110.4 111.1 112.9 119.2 130.1 139.9 51.9 Government Sector 20.2 24.5 21.5 24.9 27.0 28.1 28.8 34.1 38.6 40.7 Gross Domestic Product 154.2 155.5 161.8 175.3 1.76.8 180.9 192.4 212.8 222.9 243.5 Cost ofLi ving Index;i NairobiY 100.0 100.0 100.1 101.4 103.8 109.4 nO.1 112.5 lll.h 119.8 1/ Monetary economy only. 2i Excluding rent. ~ource: Statistical Abstract 1967. Annex Table 10.1: Atlti tudes of East African Pupils towards Fami~ p!anninS, l~O;:OO (figures indicate percentages) 1/ Positive Ambivalent Negative Total Number of Subpopulations Rank- Answers Answers Answers IesEondents Male primary pupils 24.5 15.2 60.1 24.7 1150 Male secondar,y pupils 12.5 19.0 60.2 20.7 469 Female primary pupils 22 15.8 59.1 25.1 708 Female secondary pupils 5 22.4 54.9 22.7 321 Primary pupils under 15 19 17.3 59.2 23.5 676 Primary pupils age 15 and over 27 14.4 60.0 25.6 1182 Secondar.y pupils under 15 1 29.4 54.9 15.7 51 Secondary pupils age 15 and over 7 19.7 58.4 21.9 739 Respondents in Kenya 12.5 19.0 51.5 23.5 9$2 Respondell ts in Tanzania 29 13.2 60.4 26.4 915 Respondents in Uganda 10.5 19.1 59.9 21.0 121 Primary pupils in Kenya 19 11.3 51.9 24.8 560 Secondar,ypupils in Kenya 6 21.4 56.9 21.1 392 Primary pupils in Tanzania 30 13.1 59.5 27.4 751 Secondary pupils in Tanzania 28 13.8 63.4 22.8 224 Primary pupils in Uganda 21 16.8 61.8 21.4 547 Secondary pupils in Uganda 2 26.4 54.0 19.6 174 Kikuyu males 4 22.5 56.0 21.5 200 Kikuyu temales 3 26.0 50.1 23.3 142 Gancia males 15 18.5 59.1 21.8 427 Ganda females 8 19.3 62.0 18.7 111 Luo males 24.5 15.2 63.0. 21.8 302 Loo females 9 19.2 53.9 26.9 161 Sukuma males 32 8.1 64.0 21.3 231 Sukuma females 31 12.0 64.9 23.1 208 Rural areas 26 14.6 60.1 25.3 708 Semi-urban areas 11 17.4 58.8 23.8 1419 Urban areas 10.5 19.1 59.0 21.9 461 Catholics 23 15.7 62.2 22.1 1251 Protestants 14 18.8 58.8 22.4 383 Muslim 16 18.4 55.3 26.3 315 Others 19 11.3 55.9 26.8 693 !I Percent positive answers (in descending order). Source: A. von Momos, Attitudes towards Family PlAnning in East Africa, Munich, 1968 (IFD-Insti tut fur Wirtscha.f.'tsforschung, Afrika-Studien, Nr. 26), pp. 161-172. Annex Table 10.2: ~citied Local Conditions by: Forei~ ExDerts National Experts Regarded .. Greatest Hindrance in: LAY Mit!-.- A"t-JiAS!!! LA!/ Mit! AFJi ~ Traditional practices 32 26 19 25 15 11 19 9 Low level ot general education 21 18 14 25 22 18 31 36 Absence ot progressive leadership .5 16 12 16 15 1 8 Lack ot coordination among agencies 4 3 13 9 11 20 6 9 Cbntent.-nt of local population 5 9 11 3 1 1 3 9 Beliet they cannot improve 5 1 10 9 13 3 Inadequate natural resources 9 4 4 2 4 1 3 Internal conflicts among villagers 7 6 3 2 7 4 19 Inequi table 01lJl8rahip 4 3 3 7 4 Distrust of non-nationals 1 2 No opinion or no 8DBver 9 10 10 6 11 1 8 36 Total. nUDlber of iDtODUUlts 57 68 lh4 57 27 45 36 11 !I Latin AlEricm survey countries: (l)lombia, &:uador, MexLco.• ~/ Middl.e East survey countries: Morocco, Turkey, UAR. JI African survey countries: Malagasy Republic, Nl.geria, Senegal, Tanzania. l!I A81an survey countries: Cambodia., Iran, Pakistan. Source I Annex Table 10.3: Expert's Qeinions as to which SEecified GrouEs are Antaso- nistic to DeveloEment'Pr0Jects a bl Major Resions (figures indicate percentages) Foreign Experts in National Experts in Group LA ME AF AS ALL LA ME AF AS ALL Older Men 60 66 65 L7 61 63 71 6L 73 67 Older Women L9 5L 52 35 50 59 56 61 73 60 Younger Men 1 3 7 3 11 3 Younger Women 1 2 2 2 3 1 The Richest The Poorest The Literate The Illiterate 60 7 2 37 38 7 6 35 L9 23 8 L , L2 11 30 L7 8 .~ 29 63 15 37 53 16 7 51 39 8 8 L7 6L 18 36 52 12 7 L5 All of These 1 /1 None of Thece 11 9 11 16 11 L L lL 9 8 No Answer and no Opinion L 5 5 L 5 2 3 2 Total Number of Informants 57 68 ILL 57 326 27 L5 36 11 119 Number of Groups Mentioned on Average by Each Expert (2.3) (2.2) (2.2) (2.0) (2.2 ) (2.L) (2.6) (2.6) (2.7) (2.6) /1 Less than 0.5 percent. Note: For the meaning of the abbreviations and the details of geographic coverage, see preceding Table; ALL a all survey countries. Source: H. H. Hyman, G. N. Levine, C. R. Wright, Inducing Social Change in Developing Communities. An International Survey of Expert Advice, United Nations Research Institute for Social Development, Geneva, 1967. hsh/rnonth 5oQ) • fit'" • I ' , ! . • I , f'" i : ... I '. ... t I : ' ~ .. ~ ~ , , I I ! I • I ~ : : 7~ . ,(! / / / .• ; I I ., I I ttl I t 17r'"(: ~ 'j i-I 1 I , I '/':, :! • I' • i~ ·1·J.·-'· t i t"1 ./ /. I • I J' ; I I 'I /; I i • : I t • .i . ," i , I t .~ I r ' 200 1 i I • I t I I ' I • I • , I ' I I:. I I • I I • I • • t • , , I 1 t • I , I . ,.. I " • I i. !: I • , I I I , 1 I 5000 ·. 4 -'~ '''T .• ,> \. ~: :. • I I • .... , . • •• ., • I • , • • I ; I i i " I • • , I- _ I I :., ~.. ~ ~:.L' .', .... I I "::;'00 ~. '. l f + ..1 1 I I ..... j I ' i· • I • 1 , i j . I t ...... t- ~ : I i I I I • . -y J .. • 1 • I • : ! t : i "1 i l·· I I,' I ; I ::, • t.. • rooo 2. -I,!", .... I • • !, ~ l.j,~ • I' •• ~. I ' •• : I ·!. I • , , . 1 • I • ; * I I I I ... I. t I + ~ f f I 'I"!' : I f If" I I , I Itt< it l- I I , • t .I It. I I • I I I 1 ' I , I , t .. i ~ ! ~ t I • I I I • f f I 500 l' T I 'I I • i I , ; • I • I I I I • I I! '1 11'-+.··t. +·f ! r· I : t I 400 ... I .' ! . ~. t t!· ·f· , . I ,i l' i i ! I ~ .... ~ '11 ~.1 .. .1 t .~.; ,1.-\ .; .; ! ; f I r!." !. I • 300 I i I ., 1 I I , Ii' 1: , / , I ; I I i; r : t 1 • i ! t~ + :.; 1- -:""1" ! j .... : .... I" I. ,·.1 t·, I , ; I It· ,I I ', I I :. " Ii ;: 1 200 I I , • ! I I . I I I . :, ...L I ;: l-~ +.! . I I ! I ' • I I t t t. ! • : I • i I' , ) l t I t '-1 ~ ,-,~ , .. " ~ " I .... • ··t ",.,.., I • f • I I I •. 'I I I • I I i ·I I: · • I " , I I .. t ~ I . .... ~ I .. 'I I I • j I i I ......,· •.J...·t ~ .; -I • .j•.• ' • , I I t ,,, . , • I j I.·; ! ~;. 1 J ! I • I I .I : t j , I I ;. : f I l ,I I j i 'I I ·! , I I ' •. j r .. t· ~ I ., :'l; " ~ 'r .~ .; :' : I l I " . " . .., I I i I I · Ii ! I I' i I I • I I • : I : ' ii, I: , I ; : I I I I I ,I I • I I 100 • I " I I " • t' ... ~ f. '"' .~".l I J J f. i t · I I H·i 1 ! :. ! ! ' • , ; , f : t I • I. t: I •• ,: Sour ce II Tab!J..e 5.l~ , I . I I I I ! l j , I • , I I • I I. ! • I I' • + • I I I I l' , .' I I • I I I I I I • , I I I I l ::! I ! ! I , I I I • I I . I • I , I I ' I I I j I d , ft· , t i I .I I·; I • I i I • I , ,, I • i ' I ,I 50 I • 20 )0 ho 50 60 Age (Years) 5000 ': #~H~-I --1- --t L: i- : 'lH-tmWfHtH~~! "'~'r +.lTrr!li-' ~ ~;f:; if;"; . -~}!~~~ ~~~ ;;~~ -~~::f;::n ': :--r<~~ -=ttt ;,T ·-,F '-"r-,.,."'~:-'-":"" ,T· -,,~.;:lJ< Tl ',,-,", ::~.':-.;.- Hi·-f .+ I ,.! T T t , . ,'7 ·Sr.2"~' ~AigJ3Lltarnil:lt!8-' ~'~~Wl~~ l~ .~.. \0: . ~or' .S~C ~'~~'~~-",!~: ~~~l-.. . . ':: ~~7' 8 ill-;' -, :+- r- . ~.-- .' .' . l·l.-·l'~' rl rtJr! i' .' I, flEe' ~ -. ~aili!:oF\l.wl 1ll 1"1i- ~ . ;;,.:: .- J - :.".'" .~ .. r-~'FF':··" - ".' .'.-" ' .. .. ,,c f--'c-c- 7 --: 11: -,r''':!'f+' I ' ! ,!I ;' '1+" I'~' ! t""'I'l- I r; -+l.-~L·' '., J ,.. ", 1'1' : .... I~' '1~ '~I .. :. ~~"::'I'j -;. t, t ! i t' <, f' -. '1' • " I . . . _'¥;:-~ I . '. ,-:-", :' ,~' . "',1511'..:": ~~~ ~~ ~~., --~ 1--'··---1···· I i " l' •• '~f'::~ . I ,. /' ,'" ! :,' , .:- : . " I:!: 1 ! I.:! i , .. .J . ' 't-''-'', :", ': .. ;., I ! t t 11 i" ;"., - I '.- .,"-~ - -::~~:'.:.' ~.~ t, '--~:-:~=-I--,:'" : :; ; 1: I : I : 1 f 1 ; t ! I I I:' : ~: ,I ' ,'" " !1 '.l :'w-' I J t ,I i ! ~ ~! ~ ! ;. 1 !' l' : 1 ;" t . t ; ! '"; t: l";' r ri .f rt· ' '1- r t i~ T: 1':-'1 il'-~'- i ':'~lt -l'rr'I--l~"l' r i -]},' j 5"f: +'"~~-1 1 ~ t ~,.-+ ~'f'~·iW . i-+. ~·t-t \'1 ~ ~-J'.~ 't+ 1·!~++_i-t4-tt-HT11- r l 1 '/. "'X i 't ':'-H~t~-:'-""':"":'I 't , ::" .... ;t. ! : :;I, :;:: ...... : ~ ; J !:: i :: i : t t !;I fit ~ ,,~-I I ! t t i :; t i : 1 I! :! 1 , . , . . * -\- ., .... ; i : : 1.., ~,,. .1,-\ !: : I, :: -,- ,'- : i !. 1 t. t i ' i ' ~i t . ":' , , L '1 L l :;: "-'1'1i- -t--1- 1 ii"~ -+-. -, -fl II' :rj , ;~rt -:-- - -+-l-f- -r-.--J U L! -. p"'I ,:~,- -I ,--\- itt t --r--- ri:ft · 1.\ , ..I . ~ -~ ~ f .. ... • · ., . '7" -r-j'"T-~-f'; .. : : r': r t'T ..-t-;-:-:- . 'ti~:-i'. i'--" ~,.l f -;-...;.,.... f f 'I'. I ' , , , • ... •• J p. t' t ~ t' 1 . l'" I ~r t !,.. '1' t· .!! I I - ... - .•' • : I l'~' It lt f I Tt I' ·:-:r-tf:· t '1 t : ,..'r' .. ."f "t +t' ., J t t : ~ :: • , . t ;: : : ; , .: ~: " • , • • 1 • , • t I •• r ... •• ~ - t... I, , , • •••• • ~4 -I • ; ~ :; : 1 ~! !::: !::: .., . ; . I. t ' !-., • • t t ..,. • ' t ""! 1ft ,., t .' I I ,.. .,. I I "1' .. t--r- -- t'1-r I I ' . ,. I" I "" t . 1 ' t 1:1 I". I ! ~ 1 "" ,: f l ' , ., t, 11'1 t I" I ! • -l-L t~, I, t· ,.,. 1" .. I--i.~ 1 I I ' · ! t· ·t-t- -4--+-1 t i + ; I -\.. -'r I - • I • I !', , : , - 'I!' 1 +- ,. I. '1' ,. , I I f - - I I I 1 1 I ",_.' -I -.. I I I j ,. ,f I I, 1.1 - • j f I I t I t+ -'r I I , • " t fl.. •• 1 • I ; ' , ' . ' , ' I •• t If' I t-.~l- ~ +--t-+-++ .4_~ .....i...,. : -;-.. I ••• I., ,.,' I I I' I I I + •.• ' . . - .... -+-, ' ' +- '3 ' • • • .1 • • .• I . • • , • /' i " , •• " • I • ••• +-.. • t ,I, ..._ 1 .!. .-, - • • t . . . . . . - ,. -\ --I ,,,, ;. I--+- I I ! I !!' I .' I ., ,.!t I • IIll! f t · ! t· I 1 I I ,I • " " I I f I I I t .. i ..I-f, -1-. I i I Itt I I •. t I i . , I . , •• \ • , I· I .. I I ,I I I j. I ! il I I + I I I I. I i • i • · i • ! ' ' " !, I;. i + • ,; I ,I t , • .·1 I I I I ~ : ",-+-+"+,,, t! ''1'" ! ,• I .,. ! I I -1' "I·j .. I • I It! . -+ .. i' f j' '''1-'1-' iI' ,J.. . . . I ~ t ,. t" !.' i 1--,··t- '! 'I · I t I •• ,!' . , I , 'I • . . , I • I • ! • • I , I • I." I , I I I I Itt I • Iii , ' I ,; 1 ,,' I ~ 11 I I II , 1- i If; "I "I ; f \. I;' ;, I I ! i i ~ !! I r- 'l-r: 'T- I 1 ' • • • ; 000 '2 "7';' 'I';"!' . -: .' . ~I"'" '. : : .' . . /1 "t·~ : I I .... ' ! .... ! II! I I ., :' ! I' ,: • 4...1;. ::;, i" 1f l ' ~: t I I l ~Q . I' v·· ",,"" t ., ," I ,., l' I , ·1 .. Y.'r~ . ..• . '~' 1, q 1to. ... 1'1': ! I I ' • I t •• I i . • f • I I 'I J , i • t f i .,. t f It, f t , I I f I I I ,t! ..., ~ ·I ·. l -:";/'; ;,; 1 : : :: :;:: ('1'1-..; 7 . :: .! I " .. %, _'~r< !: :;!:::(: ,)\1 ' ::~: ~::I ___ : - ' I ! " ,ii, 1 :: i ! , •• :" I J- ' I' · I • I • I J I I i . 'i ' .: ." : r'l'L' I, t"j1 .f , I I I , • 500 I 1 1 : II I jl- ., '~ ~ '; T"!!"!'''i,''t trr'! t! i !! .. I: · . " !(F r ' ;:- r' ;·t·; ! i j'l f f ! !1 1 i' t , It I • .f ;"r :/ ; / tL ;: 1 l ' .;! . .' ~, ~ I . ; ,I'! t ~ ; j'1 ~ .; t '; '1-! rr~ -ttl"'!,' r-l~'ft 1'"·ff.·,t :'. I i 1 I .: 1 ,I 400 .1 I . ', • • , ,:: /: .... I 1 t. ~. , : ~ _---.....,.__ I . ! II. . i ''I 1' i·l.; ~!! .; :."! ·~l t I .. l'~-· j • ' -I-'~-T"'I'I • I! . -+ . I . .. tT'·-~l r f I ; :--,; I;:' !;J;,: :~~'-:: ,~:;' ~·:LLt0-.i:·~fJ·~rl.. 14}'jJHi~Ji.l-:"'lILLpq~ l .~/·;I ~".,' ~' :"':'i:;~: ;l.i.t, 300. : · /1; :'.', ",,' t': "t . .:. ,.! I' .' t, t·!.: ;;1 :!. I, Iffii 1 i I .. I ' .1. : ,:,1··,l1'il , ; I l;' + • ,.. ,' J r I:! ! I. ",. • - j .. i • , , •.. ~ • 'h .' ! ; : . ' J: ',: :lL ; t i j' j ~ 1 ~! ;;.j Ii ; ! -:-"t:' -:-. :'. t 1 ' t ': ;. ., t i l ~.: ~ ~. t • • . ~ - r '~'... " •••• - -'f : . 1· . I ,. .; i. ,! 'T4 ..;··r-H. -1·~t.. · ~ :r' f; · ,. -~. ~I 1 ; I ! if: . -.~. '1'''7''~ , !~~~I~l;t.t f , I I .. :, . .. Itt : I ~!~~ ' , t J rr If 'r: i I t· .,. 7 , ~ . ! .1 f ' , : : f j 1 ~ 1 ,"'f ,~ ,~ ..:: :.1 t ~ ~ I "1 ! ; 200 I'!' • • • : 1 .•. , • .. , ~ • ',t I , I ., I 1, I t · I' . r . . , I. 4, • f I f· I ' t' I I , I) If •• + ~. , I ! r I:" I:;: .:; ! ; '1': . t I j . t • I • ""i '~' r :' f i t .!. ~~. : ,.~. '~I'; ,t • -; •• t I f ,~ ~ ~ 1 •• I • , , t • t" : , I • t • • t • • ' • I i i I , . : I ! : , t t .! • I • , r I. I .; I • ~ : ' I • I' I l' I • 'I' ' j . I ,, ". l .. f , ! ' , , 100 i I ! I ! i . ! i ! I I , I ; : I' : , ' : I , I I ! ! t j I I 1 , I I t I. , ~.. I , ., , I . •• r f ' I , i f I I I f I I I , I i I I • f • i , I 500Oto 4 ':' 41 .'1 I. i • t .. ~ t i ~ t i.~ 1 ~ Ii. , I t I + t. ;.; t • I j 000 2. .'-:- 7" :-,. r• :.~, ...!. ~-'---I' I t· •• I : ! lit :/' t . ,, . , it •. '. !: /: V; ·t· .. ! t , I • I I : ~ , I , i , I 500 .1 r Itt'~ : ! ! I i i i III . ;-'1 t ·1, lw.·l·.-, .1 : 1 i!·t I 400 I •• ~ ~ • f"lI j 1 ! i I I ;"1-;' ~ 1 I I I I JO . , 1 I • i I t ". ~ ... f , I t : ..·-t-~f! "1" I ~ j : , ; I I I . I I I ,.' I · I I I I I I I , 200 ' I i I • , I I : I I I I I , • I " •• ,. ~ I ~ , t ~'I" -I I I I , I I I I I lit t : ./1 , • I • t I j I·: ~ I • I !. : ! I • t I • I • I , , I I \ ! J t I I , i I I .. I·· I I ! . I ! I • I 100 1 ., :.: . j I \ , f ! , ; ':. ~" I I I ! I • I , , I I I • soUrce! I • t I I .... · I t ' .. • Y:·"1 I 1. 1 I • I ~ I I ( I I I I" .I !, I i I ,! I I I, : , ~ I i ! I I I ! I I I ! , . I I· I 1 I I. I I I I I • I • i. jl Ii, i I I I I I; I I I I • : I 50 _ .......__.......--;~__......___'I"""'!!;~.___ i I _·.....~~I... ! ! _ _ _. . . . . ._ _. . . . , ........_ _ _ I I I ....-.I____ ! ! i , I! I i 20 30 40 50 60 Age (Years) • 500 , I I t t , . t I I . 200- .. .. I I , • I ., I I I t • j I I I j ! t I lOc}' .. t • t ~ I '! . t I ' ,"I l • I ! I I • sa-t-i--+- 5 4 3 ! ~ .· I I ! Z . : : W t'l I t /'. t·· I I k -<-.j.. •• 2 1:~ 100 ~ t. Ii' • t I i t -: t • l • I I I : ; i ·i t • -~~·~~¥~tt I ':.l..J ·t + ~ •• +. I , , f I I ~ I • ,.. , I .. ; .;. · r , f I i ! I ' I i ., • i ; I I ., I . t I ~ I I I • I I ; • I I , I , I I I ~ I , , • I • I f i I ' • I I • ! , I j I ' • I i , I ' 500 ... ·l "\' · , ,. -f ·.1 I ., t·; I . I '7; (alJJ)t i~' ;·1 • 1 • • ~ ~ I I, t ·1· .... t • ! " " j ! ! I t t f r J I f· -, of" ·i· -:-+'1'+' l' I' r,"l 400" • ! ~; : t t 'r ~.~ ...; i • • : i I · .. \": ~ ~ t ·u, t I: ! : ! i i • t ., .... ~ l' 1+~f'+ l·+- i - + '1' ,~ I , - • I, : I ; ; . ; 1. • ,t.' ; 300 1 i .. • , : I. t I" :.. '!,_ i · • + • 1 ! - ; r : ~t: 1f :'1 i : :i I I i i . ,I 1.. •' i • • I ill 4 ";-t! ~·TT- ,.-::+~. ·~-t~ · ., r • •.• '''1· .. ' • I;· ~! ' : !. [ ~ !., • .; "". • . ::: I 'I 'I ~ t , 200 .1 I . I . I '. , . .:" ..~ " I :. : : }oI! t 'I' •t, . .i I • '... • ; : , T':.;"]:, :;j: :: I • I • I I I I I J .. I • I • I I ; , I • I 1 I, I. I ; f I I I • I I I I , I i I i I I I ' I • I ' I I I I r i 100 :i' ! Source,; Table I I I i I • I I ; I ,I .I , , • . I I , I I • t , • , I I ! I I , , I • 50 30 40 50 A~0 (Years) lOOO 2 J I ! I f -1-.+ ~ • I • • + • , I , ! , I I • t -i •, I I I . f I I I 500 400 ~1 300 .., . , . . , . , 200 ~ .. I , I - 1 I I I I f • . i 100 '2 .....'4; I ....' •• I , I t t • , • I ! , ! 1 \ 5000 JOOO 500 400 300 , ! 200 \ .. I t • I • I I I • ·, I ! I • i I I I i , • .1 , I J I ·; I Source: I , I ' · . I I . 100 ! I , , 1''''''' I t , "I. I I I r , • 1. ,,, .I i . I I I : : j : ! I I l • J • , ! ! ! I I, ,I .. I •• t 1 t I ,I I i • j I tit 1 • , I ! i I~ :~ ! . I ,I I I I' , t 1 t , j , • I , I I , 1 50 I t ! 20 )0 40 50 60 Age (Ye8,X'S) l' ., 'J : h ) , 20 30 40 50 60 Age (Years) • o o z l> r ::0 1"'1 "'0 c: CCIIA101/ [[) r n KENYA PROVINCES a COUNTIES TRIBAL GROUPS PROVINCIAL BOUNDARY COUNTY BOUNDARY EMIU TRIBAL GROUP \ INTERNATIONAL BOUNDARY \ E===~20E===~40~~'~O~=3'§O==:3'QO /.IlLES NOV~MBER 1969 IBRO ~ 2616R