Returns to education and skills in the non-OECDs: evidence from Urban Kenya

Oledibe, Nnanna Osita (2024) Returns to education and skills in the non-OECDs: evidence from Urban Kenya. PhD thesis, University of Glasgow.

Abstract

This thesis examines the returns to education and skills of the labour force (aged, 15-64) in urban Kenya, a non-OECD¹ country. Comprised of four self-contained analytical chapters with interconnected themes, this thesis uses the World Bank’s STEP² Household Survey for Kenya. Beyond reduced-form equation modelling, this study deploys a structural equation modelling approach. Hence, raising the internal and external validity of return estimates. Using the 1985 curriculum structural reforms in Kenya, this study exploits exogenous variations in schooling and skill from which causal inference is drawn, providing empirical evidence that informs policymaking. In this study, the consideration for skill as opposed to mere schooling³ presents a novel approach to examining human capital, particularly, this unravels useful insights that improve existing understanding of the mechanisms through which schooling raises skill, and in turn, other labour market outcomes such as earnings.

For the first analytical chapter — Chapter 2: Educational Attainment and Skill Proliferation — In examining the effects of schooling on skills in Kenya⁴, the evidence suggests substantial ‘inefficiency in schooling’ in urban Kenya. The term ‘inefficiency in schooling’ describes a state where workers with relatively low educational attainment have relatively high reading proficiency. To give more understanding to this phenomenon, I stress the importance of the effects of background characteristics⁵ on access to schooling (and skill). Findings provide a basis for an argument for more effort to incentivise equity in access to schooling for all, over recent arguments for increased quality in schooling. I argue that whilst the latter is of ‘noble aim’, resource constraints in education provision mean efforts towards increased quality over equity-in-access-to-schooling inhibit skill proliferation, particularly in developing contexts. Exploiting the 1985 curriculum reform in a Difference-in-Differences (DiD) analysis, the evidence suggests an upward mobility in schooling for the ‘disadvantaged’ (respondents that have fathers without post-secondary schooling). However, no useful evidence (from which causal inference can be drawn) on skill is attributable to parental post-secondary education or socioeconomic status at age 15. Further evidence suggests this inconsistency in the effects of background characteristics on schooling and skill is partly due to the DiD estimator that gives the Average Treatment Effects on the Treated (ATET); and the inefficiency in education, in urban Kenya. Interestingly, turning to the Two-Stage Least Squares Instrumental Variables (2SLS-IV approach which gives the Local Average Treatment Effects (LATE), the evidence suggests that relative to the ‘advantaged’ (those that have fathers with post-secondary education), for the ‘disadvantaged’, the effect of an additional year of schooling on skill is positive and statistically significant. Although quite different, in that the LATE in this case captures the effect of schooling on skill strictly for those impacted by the reform; on the other hand, the ATET, captures the effects of parental education on schooling and skill, regardless of the reform. I find outcomes of the ATET and LATE complementary and strongly responsive to the 1985 curriculum reform and the prevailing inefficiency in schooling in urban Kenya. Particularly, having the coefficients (ATET and LATE) indicating a substantial rise in the schooling and skills of the ‘disadvantaged’ is strongly attributable to the reform and the inefficiency in schooling. This is not to suggest the ‘disadvantaged’ have higher schooling and skills relative to the ‘advantaged’ as the reverse is the case, with evidence of substantial skill differential attributable to differences in schooling endowments between the ‘advantaged’ and the ‘disadvantaged’. However, as earlier highlighted, the evidence suggests, relative to the ‘advantaged’ that the reform (and the inefficiency in schooling) drives the schooling and skill of the ‘disadvantaged’. Ultimately, whilst the evidence suggests the effects of inefficiency in schooling looms large and should be addressed speedily possibly by addressing the quality needs, particularly, at higher levels of schooling (above ISCED2⁶), further evidence from the 2SLS-IV approach suggests, a more positive and substantial effect of an additional year of schooling on the skill, for all impacted by the reform (regardless of background characteristics). This suggests efforts aimed at raising equity in access to schooling should not be discouraged out of quality concerns. Hence, I argue that reforms that incentivise access to schooling for increased educational attainment are more crucial for skill proliferation than efforts to raise the quality of school inputs. Besides, raising the quality of school inputs can inhibit access to schooling, due to resource constraints.

In the second analytical chapter — Chapter 3: Private Returns to Education and Skills — I estimate the private returns to education and skills and examine the wage differential across gender and employment categories. The findings suggest, controlling for schooling, the OLS (Ordinary Least Squares) return estimates of non-cognitive skills are robust, with Openness to Experience and Conscientiousness yielding positive and statistically significant wage effects from which causal inference is drawn. Openness to Experience has the strongest effect with a standard rise in Openness explaining a 35.9% rise in hourly earnings, statistically significant at the 0.1% level. However, a standard rise in Conscientiousness explains a 12.6% rise in hourly earnings, statistically significant at the 5% level. The 2SLS-IV estimates show consistent estimates of returns to schooling and cognitive skills. Findings suggest, no evidence of statistically significant wage effects of schooling and skill from which causal inferences are drawn. Further evidence from subsampling (heterogeneity analysis) shows that relative to the female gender, the male gender has positive returns to their schooling and cognitive skills, with an additional year of schooling explaining a 25.6% rise in hourly wage, statistically significant at the 1% level. For the measure of cognitive skills (reading proficiency, unstandardised), the evidence suggests that a unit rise in reading proficiency in PV (Plausible Values) explains a 0.77% rise in hourly wage, statistically significant at the 5% level. Using the first stage of the Oaxaca-Blinder decomposition as the baseline estimates, the evidence suggests a 23% hourly wage differential across genders in urban Kenya. Differences in schooling and skills characteristics/endowments explain about 37% of the wage difference across genders. Further evidence suggests that a substantial proportion of the wage differential between genders is due to (potential) discrimination. Particularly, females are not discriminated against based on their cognitive skills or schooling but rather, the evidence suggests the potential discrimination in wage between genders comes through differences in their non-cognitive skills (or personality traits), specifically, whilst the males are better rewarded for their Openness to Experience; the females are better rewarded for their hard work (Conscientiousness). Substantial policy insights abound in these outcomes.

In the third analytical chapter — Chapter 4: Human Capital Externalities and Social Returns — I examine pecuniary and non-pecuniary human capital externalities⁷. Evidence from OLS output suggests substantial negative externalities of schooling in urban Kenya. Specifically, the findings show negative pecuniary and non-pecuniary externalities (of schooling) that become less negative with rising aggregate (district-level) schooling. Hence, the negative pecuniary and non-pecuniary externalities of schooling become non-negative (positive) at a certain level of aggregate schooling. This is consistent with the argument for more schooling (over quality inputs), as in Chapter 2—the first analytical Chapter. Interestingly, findings show the pecuniary externalities of skills are positive and statistically significant. The differences in the externalities of schooling and skill unravel interesting insights that question some ‘stylised’ facts in the literature. Particularly, findings strongly suggest aggregate schooling is meaningful (or makes economic sense) only at a certain threshold, and any level below this threshold inhibits earnings and skill proliferation. Interestingly, on the other hand, increasing aggregate skill (regardless of the skill level) has favourable effects on earnings.

Finally, in the fourth analytical chapter — Chapter 5: Returns to Education and Skill in a Dynamic Framework. Amidst limitations that accrue from using a cross-section of data, the main objective of this chapter is to test the robustness of estimates from previous chapters (chapters 2, 3, and 4) that use a single cross-section of data. Hence, a dynamic framework that accounts for data limitations and supports causal identification helps to raise the external and internal validity of estimates addressing possible biases in return estimates. Inspired by the study of Krishnakumar and Nogales (2020) in Bolivia, I deploy the Technology of Skills formation—a dynamic framework, pioneered by Cunha and Heckman (2007). Overall, findings from the Structural Equation Models (SEMs) are consistent with estimates in reduced form, as in the previous chapters. The SEMs further unravel some useful insights that improve understanding of the return estimates. Particularly, the findings from the SEMs improve understanding of the Difference-in-Differences analysis of Chapter 2 that show father’s post-secondary education impacts the schooling but not the skills of the offspring. Deploying the dynamic framework (SEMs) affirms having a father with post-secondary education not only explains the schooling, but the skills of their wards. These findings present substantial evidence of at least persistence (and upward mobility) in skill and education between parents and their offspring. This strongly accentuates an intergenerational transmission mechanism for educational attainment and skill proliferation that should not be overlooked in education, skills, and employment policymaking. This finding in sub-Saharan Africa is consistent with related findings from studies in the OECDs.

¹ The OECDs (Organisation for Economic Co-operation and Development) is a group of thirty-eight (38) high-income countries including the United States of America and the United Kingdom. Please, see the full list here https://www.oecd.org/about/. On the other hand, the non-OECDs here mean low- and mid-income countries of which the sub-Sahara Africa is part of.

² STEP is an abbreviation for Skills Toward Employment and Productivity.

³ Schooling is taken to mean ‘time’ spent in formal education.

⁴ Kenya is part of the sub-Sahara Africa known to have the least schooling and skills, relative to the other regions of the world.

⁵ Background characteristics are proxied with parental education and socioeconomic status at age 15. The ‘advantaged’ as used in this study, are those (respondents) that have a father with post-secondary schooling, in most cases, these have high socioeconomic status at age 15. On the other hand, the ‘disadvantaged’ are respondents who have a father without post-secondary schooling, in most cases these have low socioeconomic status at age 15.

⁶ Please, see the data subsection of the first analytical chapter for credential categories. The ISCED2 credential category represents the credential category of the employed who attained lower-secondary education. This is equivalent to an average of eight (8) years of schooling, in urban Kenya

⁷ The former, pecuniary externalities, entail an examination of aggregate schooling and skill (across districts) on individual wage. The latter, non-pecuniary externalities involve an examination of aggregate schooling (across districts) on individual skill.

• Item Type: Thesis (PhD)
• Qualification Level: Doctoral
• Subjects: H Social Sciences > HB Economic Theory
• Colleges/Schools: College of Social Sciences > Adam Smith Business School
• Supervisor's Name: Shi, Dr. Yukun
• Date of Award: 2024
• Depositing User: Theses Team
• Unique ID: glathesis:2024-84630
• Copyright: Copyright of this thesis is held by the author.
• Date Deposited: 23 Oct 2024 15:27
• Last Modified: 30 Oct 2024 09:55
• Thesis DOI: 10.5525/gla.thesis.84630
• URI: https://theses.gla.ac.uk/id/eprint/84630