Zambezia (2002), XXIX (ii)Determinants of Youth Earnings: The Caseof HarareHONEST ZHOUDepartment of Economics, University of Zimbabwe1AbstractThis article investigates the factors that are important in determining youthearnings in the formal sector in Harare. The theoretical approach adopted is thehuman capital theory. Youth earnings are regressed against a number of humancapital variables, personal characteristics variables as well as socio-economicvariables. The model is applied to a particular age gioup. The results suggest thathuman capital variables are important determinants of youth earnings in theformal sector. These include the number of years spent in education, the highestlevel of education achieved and the choice of subjects at GCE 'O' level.IntroductionEarly classical models of wage determination argued that firms simply takethe market wage rate as given when making employment decisions.According to these models wages are determined through the interaction ofmarket forces. The argument is that if workers were underpaid in oneindustry they simply withdraw their labour services and go to thoseindustries offering higher wage rates. This would continue until equilibriumis reached.The above classical models of wage determination are based on twocrucial assumptions, that workers are identical and that mobility betweenjobs is costless. Recent theories of wage determination recognise that workershave different characteristics. For example, theories of wage discriminationargue that workers are judged by factors which have nothing to do withtheir productivity. These factors include gender, race and religion amongothers. On the other hand, human capital theory argues that individualwages differ because workers possess some acquired characteristics.According to human capital theory, individuals invest in themselves throughDr Honest Zhou is a lecturer in the Department of Economics, University ofZimbabwe. I would like to thank the Professional Development Training in Economicsand the University of Hull for sponsoring the survey. Thanks also to Dr M. Ncubefor the useful comments provided on earlier drafts of this article.213214 Determinants of Youth Earnings: The Case of Harareschooling and training. Through such investment individuals attain certaincharacteristics which will have an influence on their productivity and henceon their remuneration as well.This article sets out to investigate the determinants of earnings in Harare'sformal sector.2 The article uses a sample cohort of 21-year olds who wereobserved in 1996. The survey collected information on the individual'searnings, employment history, education, occupation and training and socio-economic characteristics. Hence, our model incorporates human capitalvariables as well as personal and socio-economic variables. The resultsshould not be interpreted more broadly for the following two reasons. First,because the sample used is confined to Harare. Second, because the study isconfined to a particular age-group.The analysis of the determinants of wages is important because of theassertion by human capital theory that individual characteristics acquiredthrough education and training have a significant impact on earnings.Hence, the analysis has important implications on private investment ineducation. The investigation of the determinants of earnings is undertakenby estimating an earnings equation. The results obtained show that humancapital variables are important in explaining youth earnings.Most of the research on earnings determination has been confined todeveloped countries. Studies from developed countries include those byDolton and Makepeace (1986, 1987, 1990), and Dolton, Makepeace andKlaauw (1989) among others. The limited number of similar studies indeveloping countries is because data on human capita! variables is sketchy.In Zimbabwe, not much research has been done on earnings determination.Velenchik (1994) estimated an earnings function for Zimbabwe'smanufacturing sector with hourly earnings as the dependent variable. Theresults suggest that human capital variables, race and sex have a significantinfluence on earnings.This article is different from the Velenchik study in that it concentrateson a particular age group. The article assesses earnings determinationduring the individual's youthful career.The article is organised into five sections. Following this introductorysection, Section 2 describes the Methodology. Sections 3 looks at the dataset. Section 4 presents and discusses the research findings. Finally, Section 5concludes the article.MethodologyAs noted above, human capital theory argues that an individual acquireshuman capital in order to increase his/her future expected earnings. Human2. Harare is defined here to include Chitungwiza.H.ZHOU 215capital is defined as the stock of personal attributes which increase tneindividual's productivity and hence the individual's earnings. Importantdeterminants of earnings according to human capital theory would includesuch factors as the time spent in education and the individual's workexperience.A framework for specifying the relationship between earnings andschooling and work experience was given by Mincer (1974). Thus, followingMincer's framework it has now become standard in labour economics toexpress the earnings equation as,ln(w)i = a + p,Xiwhere /' indexes the individual, X is a vector of individual characteristicswhich include the number of years spent in education and work experienceand fij is a vector of coefficients to be estimated. We expect the coefficientfor the number of years spent in education to be positive. Human capitaltheory also argues that an individual's earnings initially rises and then falls.This implies that the coefficients of the variables work experience and workexperience1 should be positive and negative respectively.The quality of the individual human capital can be measured by theindividual's highest level of education. Three variables indicating therespondent's highest level of education are included in the model namely,Primary, Zimbabive Junior Certificate (ZJC), Ordinary Level ('O' Level), AdvancedLevel ('A' Level) and Higher Education. It is important to point out thathuman capital is not homogeneous. For example, two individuals with thesame level of education could earn different wages because they studieddifferent subjects. Thus, the labour market could value some subjects morethan others. The survey collected information on some subjects studied atGCE 'O' level. Hence, we include in our model the subjects obtained at 'O'level. Other human capital variables used are the number of jobs and on-the-job training. The number of jobs is used as a proxy for the individual'swork experience.The demographic variables included in our model, gender, marital statusand dependents have been found in other studies to have a significantinfluence on earnings (Miller 1987, De Beyer and Knight 1989). In particular,the variable gender is meant to test the hypothesis involving genderdiscrimination in the labour market. The variables SClassl and SClass2 canhave two interpretations. First, they can be interpreted as human capitalvariables as they indicate the type of human capital that the respondentpossesses. Second, they can be interpreted as socio-demographic variablessince they provide an indicator of certain advantages that the individualcan enjoy. For example, an individual whose parents are well educated canhave certain social networks while boarding secondary education is generally216 Determinants of Youth Earnings: The Case of Harareconsidered superior to day secondary education. The full list of the variablesis given in the next section.DataIn 1996 researchers from the School of Economics, University of Hull andthe Department of Economics, University of Zimbabwe, conducted a surveyof 21-year olds in Harare. The survey was undertaken under the project'School-to-Work Transition and Youth Unemployment in Zimbabwe.'Households were identified with the assistance of the Central StatisticalOffice. A sample of 100 enumeration areas was randomly selected andenumerators were sent to conduct the survey. At each household theenumerators asked if there was a 21-year old. Once a 21-year old wasidentified the individual was asked a series of questions about his/herpersonal details, current earnings, labour force experience, expectationsafter leaving school and social background.A total of 660 individuals were identified and interviewed. The surveycovered the employed, the unemployed and those who were still ineducation. In this article we have decided to focus on those who wereformally employed. Other groups will be considered in subsequent articles.Most of the information used in this article such as earnings, marital statusand academic qualifications was obtained directly from the answers of therespondents. The full list of the variables used in the estimations is asfollows:Gender: A dummy variable taking value 1 if the respondent is male and 0if female.Marital Status: A dummy variable taking value 1 if the respondent is singleand 0 otherwise.Dependents: A variable indicating the number of dependents that therespondent has.Educate: A variable indicating the number of years spent in education.Subjectl: A dummy variable taking value 1 if respondent passed English atGCE 'O' Level and 0 otherwise.Subjects A dummy variable taking value 1 if respondent passed eitherShona and/or Ndebele at GCE 'O' Level and 0 otherwise.Subject3: A dummy variable taking value 1 if respondent passedMathematics at GCE 'O' Level and 0 otherwise.Subject-*: A dummy variable taking value 1 if respondent passed Science atGCE 'O' Level and 0 otherwise.Primary: A dummy variable taking value 1 if respondent's highest level ofeducation is primary education and 0 otherwise.ZJC: A dummy variable taking value 1 if respondent's highest level ofeducation is the Zimbabwe Junior Certificate and 0 otherwise.'O' Level: A dummy variable taking value 1 if respondent has 5 'O' levelsubjects and 0 otherwise.H.ZHOU 217'A' Level: A dummy variable taking value 1 if respondent has 3 'A' Levelsubjects and 0 otherwise.Higher Educate: A dummy variable taking value 1 if respondent has eithercollege or university education and 0 otherwise.Work Experience: A variable indicating the number of jobs the respondenthas had.Training: A dummy variable taking value 1 if the respondent had receivedor is receiving on-the-job training and 0 otherwise.SClassl: A dummy variable taking value 1 if respondent attended a boardingsecondary school and 0 if the respondent attended a day secondary school.SClass2: A variable indicating the social status of the respondents' parents.Professional parents have a score of 4, skilled parents 3, semi-skilled parents2, and unskilled parents 1.Unfortunately, not all respondents answered all questions. In some casesalthough respondents indicated that they were formally employed, theydid not give all the details. Thus, after confining ourselves to the formallyemployed respondents and ignoring those who gave inconsistent answersand those who did not answer all questions, we ended up with a sample of194 observations. The 194 respondents consisted of 38 women and 156 men.All the 194 individuals included in the analysis had received some formaleducation. Since all the individuals included in our sample were blackZimbabweans, it meant that the racial discrimination hypothesis could notbe tested. Table 1 below gives the summary statistics of the resultant sample.Table 1: Summary of StatisticsEarnings (Primary)Earnings (ZJC)Earnings ('0' level)Earnings ('A' level)Earnings (Higher Education)Total EarningsYears of SchoolingNumber of DependentsAverage747.80878.051 884.392 036.363 329.161 553.6611.251.22Observations577891112194194194Empirical FindingsUnfortunately, due to degrees of freedom constraints it was not possible toinclude all the variables in a single estimation. Our first estimations includedall the variables defined above minus the subjects obtained at 'O' level. Thehuman capital variables; Educate, 'O' Level and Higher Educate had a positiveand significant influence on earnings. The variables Primary, Z/C and 'A'Level were not statistically significant. The variable ZjC even had a wrongnegative sign.218 Determinants of Youth Earnings: The Case of HarareContrary to our expectations the variables Work Experience, WorkExperience1 and Training were not statistically significant. The demographicvariables as well as the socio-economic variables were also not statisticallysignificant. It is important to note that the insignificance of these variablesmight be due to some omitted variables or some mispecification errors.In the second estimation we introduced the subjects studied by theindividual at 'O' level. As noted above we had information for the followingsubjects: English, Shona or Ndebele, Mathematics and Science. We had todrop some variables in order to accommodate the new variables. The obviouscandidates for removal were the insignificant variables that is, Primary, ZJCand 'A' Level. The results obtained are presented in Table 2 below. Thevariables Educate, 'O' Level, Higher Educate and Subjectl have the rightpositive signs and are statistically significant. The other three subjects,though having the right signs, are statistically insignificant. Again thedemographic and socio-economic variables were statistically insignificant.The fact that Primary, ZJC and 'A' Level are not significant determinantsof earnings is consistent with the experiences in the labour market. Few ifany firms in Zimbabwe use either primary education, the Zimbabwe JuniorCertificate or 'A' level for recruitment processes. These have becomeirrelevant to employers. They are simply used to screen those who wish toproceed with their education. Employers use 'O' level as a screening devisefor the recruitment process. After 'O' level the relevant information used byemployers is contained in diplomas and degrees offered by institutions ofhigher learning.Table 2: Regression Results and their InterpretationsVariablesConstantGenderMarital StatusDependentsEducate'0' LevelHigher EducateSubjectlSubject2Subject3Subject4Work ExperienceWork Experience2TrainingSCIasslSClass2Adjusted R2Number of observationsCoefficient5.660.080.020.030.070.250.460.360.110.080.12-0.050.0030.100.020.120.47194t-ratio17.140.820.171.162.152.482.633.470.980.761.17-0.970.671.230.701.24H.ZHOU 219From the results in Table 2 we can consider two individuals with identicalvalues for all the variables except for earnings and the highest level ofeducation achieved. We note that the earnings for an individual who has 5'O' level subjects will be earning 25% higher compared to a similar individualwithout the 5 'O' level subjects. On the other hand, earnings for an individualwho has higher education will be 46% higher than an individual withouthigher education. We can also draw some conclusions between the earningsdifferentials for two individuals, one with 5 'O' level subjects and one withhigher education. The individual with higher education will earnapproximately 21% higher than the one with 5 'O' level subjects. If wecompare two individuals, one with English at 'O' level and one without, wenote that the individual with English at 'O' level will earn approximately36% more.Our results compare well with other studies from both the developedand developing countries. For example, Jones and Teal (1993) and Velenchik(1994) estimated earnings equations for Ghana and Zimbabwe respectivelyusing the Regional Programme on Enteiyiise Development data and found thathuman capital variables have a significant influence on earnings. Ourfindings are also consistent with Velenchik's findings in that the returns toeducation are larger at higher levels of education. Velenchik also foundsupporting evidence of gender and racial discrimination. We however, findno evidence to support the gender discrimination hypothesis. The Jonesand Teal study found that returns to education are highest at low levels ofeducation. De Beyer and Knight (1989) found evidence in support of humancapital theory in Kenya and Tanzania. In Tanzania, they also found evidenceof discrimination in favour of men and non-Africans.ConclusionsThe main objective of this article was to determine the factors that influenceyouth earnings in Harare's formal sector. This was achieved by estimatinga youth earnings equation. The results indicate that the number of yearsspent in education, 'O' Level and higher education are importantdeterminants of earnings. English at 'O' level is also found to have a majorinfluence on earnings. After 'O' level the results suggest that higher educationlead to higher wages. Thus, the results presented in this article confirm thepredictions of the human capital theory. The results are also generallysimilar to those found in other countries. However, contrary to evidencefound in other developing countries, we found no evidence to support thegender discrimination hypothesis. This could be a result of Governmentpolicy which since independence has been aimed at correcting both genderand racial discrimination in the labour market.The results presented in this article have important implications forprivate investors in education. The results imply that there are benefits to220 Determinants of Youth Earnings: The Case of Harareprivate investment in education. Confirming our expectations is the findingthat private returns to education are larger for those with higher education.As noted above these results have to be treated with great care because;firstly, the study concentrates on a particular age group and secondly, thesample is confined to Harare. Nevertheless, the results make an importantstarting point in further research in this area. For example, with a largersample it will be important to estimate two separate earnings equations,one for males and the other for females. In addition, post-secondary schooltraining could be separated into different categories.ReferencesDE BEYER, J. AND KNIGHT, J. B. 1989, 'The role of occupation in the determinationof wages,' Oxford Economic Papers, 41 (iii): 595-618.DOI.TON, P. J. AND MAKEPEACE, G. H. 1986, 'Sample selection and the male-female earnings differential in the graduate labour market', Oxford EconomicPapers, 38:317-341. '1987, 'Marital status, child rearing and earnings differentials in the graduatelabour market', Economic Journal, 97: 897-922.1990, 'Graduate earnings after six years: Who are the winners?', Studies inHigher Education, 15 (i): 31-55.Don ON P. j., MAKEPEACE, G. H. AND KLAAUVV, W. VANDER 1989, 'Occupationalchoice and earnings determination: The role of'sample selection and non-pecuniary factors', Oxford Economic Papers, 14: 573-594.ElIRENBERC,, R. G. AND SMITH R. S. 1994, Modern Labor Economics: Theory andPublic Policy, 5th Edition, New York: Harper Collins.FALLON, P. AND VERRY, D. 1988, The Economics of the Labour Markets, London:Philip Allan.JONES T. AND TEAL F. 1993, 'The apprenticeship system, training and earnings',in World Bank, Regional Program on Enterprise Development, Case Studies ofEnterprise Finance in Ghana. 'MAIN, B.C. M. AND Sl'TELLY, M. A. 1990,'The effectiveness of the youth trainingscheme as a manpower policy', Economicd, 57: 495-514.MILLER, P. W. 1987, 'Segregation of women iri Britain', Economic Journal, 97: 885:896. ' ŁŁ-.-ŁŁŁŁŁŁ ŁŁŁŁŁŁŁŁŁ ŁŁŁ Ł Ł ŁŁ'MINCER, J. 1974,' Schooling, Experience and Earnings, New York: NationalBureauof Economic Research. ' ......VELENCH1K, A. 1994, 'The labour market: Wages and earnings,' in Gunning, J.W., The Manufacturing Sector in Zimbabwe: Dynmtiics and Constraints, freeUniversity and University of Zimbabwe, Mimeo.