Treating schools to a new administration. The impact of South Africa’s 2005 provincial boundary changes on school performance

The impact that the systems and practices of the education authorities, as opposed to the management at the school, have on school performance is usually difficult to quantify. Provincial boundary changes occurring in South Africa after 2005 appear to create a quasi-experiment that lends itself to impact evaluation techniques. A total of 158 secondary schools experienced a switch in provincial administration and at least two types of switches, one from Limpopo to Mpumalanga and another from North West to Gauteng, were sufficiently common to make statistically significant trends a possibility. Various indicators of Grade 12 mathematics performance are explored which take into account passes at a low threshold, achievement at an excellent level and selection into mathematics. Models used and critically discussed include a simple value-added school production function, a difference-in-difference model and a fixed effects panel data analysis. The data include annual Grade 12 examination results for the period 2005 to 2012, which allow for lags in the impact to be explored. Spatial analysis is used to identify schools located close to switching schools to establish whether student commuting effects could have confounded the results. A key finding is that schools moving from North West to Gauteng appear to enjoy benefits associated with the treatment especially as far as the production of students excelling in mathematics is concerned. However, a strong caveat is that the finding depends heavily on just 2012 values and that 2013 examination data will have to be included in the analysis before the study can inform policy recommendations. A brief comparison of institutional aspects of the education authorities in the two provinces North West and Gauteng, drawing from publicly available plans and reports, is provided to help interpret the differences seen in the data. The paper ends with some tentative conclusions in relation to how governance responsibilities in education can be optimally spread across the national, provincial and local levels in South Africa.

C21, H11, I21

South Africa, school improvement, mathematics education, impact evaluation