Convincing empirical evidence of the impacts of land reforms is extremely rare, both in South Africa as well as elsewhere in the developing world. Two key obstacles to scientific study of this question are that historically, most land reforms occur during epochs of social strife and political upheaval, and in countries where land reforms are undertaken during periods of relative normality, participation by the beneficiaries is often selective. These factors pose a serious intellectual challenge: how does one go about isolating the impact of land transfers using non-experimental data? The primary objective of this study was to directly confront this analytical challenge.
Activities
1. Fieldwork, Data Capture and Validation
The study was initially conceptualized by a team of researchers at the World Bank and University of KwaZulu Natal. A key feature of the work that was undertaken is that it employed a quasi-experimental survey design with a control-treatment type of data structure. A nationally representative sampling frame of all households who had either received land or had applied for land under one of the three main programmes (redistribution, restitution and tenure reform) was constituted and a treatment sample (households that were already beneficiaries) and control sample (households that are prospective beneficiaries), was drawn. Malcolm Keswell was initially invited to become involved in the analysis of the data that would result from this survey, funded under a sub-contracting agreement with the University of KwaZulu Natal. SALDRU was then subsequently awarded part of the fieldwork and data capture responsibility and this component of the study was funded by the National Department of Land Affairs. In total, the sample comprises 1963 households in the treatment group and 1703 households in the control group.
2. Archival Searches and Administrative Data Collection and Capture
An idealised experimental design would have the following two features: (i) the sampling frame would be limited to households already in the system (either as applicants or beneficiaries); (ii) treatment status is randomly assigned within this group. The initial sampling described above embodies the first of these two features but not the second. To approximate the second feature, detailed information on the most important pre-existing differences between the treatment and control groups was required. Much of this information is qualitative in nature, scattered across numerous paper records which are kept at satellite offices of the land affairs department. Unearthing this information entailed conducting extensive archival work.
3. Analysis and Dissemination
The iteration between the on-going fieldwork and archival searches described above permitted a far richer account of the approval process than that which could be inferred solely from the household survey alone. By combining insights gleaned from this combination of data sources, as well as the quasi-experimental design of the QOL survey, we then matched treatment households to control households on the basis of their observable individual and group characteristics.
Papers
Please contact Malcolm Keswell if you wish to access any of the papers below.
- Keswell, M., Halliday, S., Brophy, T., Warren, H. and Ogle, R. (2007) “Quality of Life Survey 2007” Technical Report prepared for the National Department of Land Affairs, SALDRU, University of Cape Town.
- Keswell, M., Carter, M. and Deininger, K. (2009) “Poverty and Landownership: quasi experimental evidence from South Africa” SALDRU, University of Cape Town.
- May, J., Keswell, M., Bjastad, E. and van den Brink, R. (2009) “Monitoring and Evaluating the Quality of Life of Land Reform Beneficiaries and the Impact of Land Transfers: 2005/2006” Report prepared for the Department of Land Affairs.