This paper explores the spatial determinants of poverty among livestock keepers by taking an
econometric approach that combine poverty indices for livestock keeping areas in Kenya as the
dependent variable and relating this to a variety of spatial variables likely to contribute to poverty
at a local scale. We use both global and local regression models. In carrying out this analysis,
elimination of spatial autocorrelation was done by use of the Moran’s I and Lagrange multiplier.
The results show different spatial variables to influence poverty at the different scales and to be
geographically related at the local scale. Soil quality, agro-climatic conditions, slope, land use
and demographic variables are important factors in determining poverty. These variables offer a
challenge to policymakers in deciding on the measures to take to enable the reduction of poverty
in the rangelands of Kenya. With the establishment of constituency development funds, tackling
the problem at the local scale could be the most feasible option for the national government when
such information is made available.