This study uses a variety of econometric techniques lo estimate the shape of the South African schooling-earnings profile. The two main approaches to measuring the effect of schooling on worker productivity are to either estimate a production function that compares how output responds [0 the hiring of better educated workers, or to estimate an earnings regression that reveals the effect of an additional year of schooling on an individual's expected earnings. Both techniques are vulnerable to econometric issues such as unobserved productivity differences, sample selection, non-linearities in the schooling-productivity profile, heterogeneous parameters and measurement error. Within the class of earnings functions, two very different methods have been developed to deal with endogenous schooling: instrumental variable techniques and discrete choice dynamic programming models. This study estimates the South African schooling-earnings profile using a wide range of identification strategies and estimators, and investigate. .. whether accounting for certain econometric issues can produce consistent estimates of this profile across different approaches. Chapter 2 uses instrumental variable techniques to estimate the causal effect of schooling on the earnings of black South African men in the period directly following political transition. It explores different ways to identify the shape of the schooling-earnings profile in the presence of endogenous schooling, heterogeneity in the schooling returns parameter and sample-selection bias due to the highly selective employment process. Our analysis shows that the presence of sample selection changes both the conditions under which the parameters of interest can he estimated, as well as the interpretation of these estimates. The empirical results show that the schooling returns in South Africa art! high, on average, and an increasing function of the level of schooling. However, the degree of convexity estimated by the control function approach is considerably less than that suggested by either ordinary least squares or two-stage least squares, especially when also correcting for sample selection bias or allowing for a non-linear relationship between schooling and market unobservables. Many recent descriptive studies find convex schooling-earnings profiles in African countries. Apart from the potential endogeneity issues that may confound such estimates, it also is not clear whether a convex earnings profile can be reconciled with the fact that most individuals choose to obtain intermediate schooling outcomes. Chapter 3 uses dynamic programming techniques to investigate the schooling decisions of rational, dynamically optimising agents in a model that seeks to incorporate key aspects of the South African labour market. Our results produce a schooling-earnings profile which is convex, although the degree of convexity is less than suggested by previous studies that use OLS to estimate this relationship. Furthermore, the marginal cost of schooling is required to be a steeply increasing function of schooling years if schooling decisions ate to be reconciled with convex earnings profile without resorting to assumptions of irrational expectations, imperfect information or schooling restrictions. Our analysis indicates that the main benefit of completing early schooling years is that it allows individuals access to more advanced schooling years for which the benefits are much higher. Chapter 4 estimates production functions for South African industries using a novel panel dataset that combines education and employment data from a series of household surveys with output and physical capital data from the South African Reserve Bank Quarterly Bulletins. The analysis starts by exploring the estimates produced by the pooled ordinary least squares (POLS) estimator, and then proceeds to alternative estimators that exploit less restrictive identifying assumptions that allow for industry fixed effects, endogeneity in the factors of production, measurement error, cross-sectional dependence and parameter heterogeneity. The POLS results indicate that the returns to education are substantial and concave. Although the effect of education disappears when industry fixed effects are included in the regression, once an allowance is made for measurement error, parameter heterogeneity and cross-sectional dependence, the results are not significantly different from POLS.
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:617096 |
| Date | January 2012 |
| Creators | Burger, Rulof |
| Publisher | University of Oxford |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
Page generated in 0.0021 seconds