The analysis of causality in rural development is difficult because some variables of the underlying developmental process are only indirectly observable. In theories of rural development, socio‐economic variables are important in linking agricultural performance variables with policy measures. Applying canonical correlation, latent variable models permit research on rural development to measure simultaneously the unobservable socio-economic variables and the observable manifest variables. In this article the stage of development of rural households and agricultural performance in South African homelands are measured indirectly. Empirical results show that the current stage of development has little impact on agricultural performance.
Notes
Respectively Department of Agricultural Economics, Extension and Rural Development, University of Pretoria, and Department of Agricultural Economics, University of Kiel, Germany.
The Alexander von Humboldt Foundation is thanked for support.