An Economic Interpretation of Colombia's SISBEN: A Composite Welfare Index Derived from the Optimal Scaling Algorithm
By Eduardo Vélez, Elkin Castaño, Ruthanne Deutsch (11/98, En) See also Poverty and Inequality
SISBEN is a proxy means test index widely used as a targeting system for social programs in Colombia. It serves as an indicator of households' economic well-being and is based on a cross-section sample of Colombian households. The SISBEN index is a function of a set of variables related to the consumption of durable goods, human capital endowment and current income. SISBEN was created with the purpose of simplifying, expediting and reducing the cost of targeting individual beneficiaries of social programs at the various government levels. The system has been very successful and has generated high demand both in municipalities and departments, presumably due to its low operation costs and also to the social benefits being brought to participating communities. SISBEN has expanded quite rapidly. It is currently used in a wide range of subsidized social programs, most heavily by those that benefit from health subsidies established by Colombia's Social Security Law.
This article examines the conformity between the SISBEN index and the degree of economic well-being of beneficiary households, and evaluates -ex post- the adequacy of the specific statistical methods used in terms of that objective. Of particular relevance is the treatment given to an important set of categorical information about household consumption of durable goods.
This analysis allows the authors to conclude that SISBEN represents a composite index of alternative forms of the household utility function or welfare level. The use of "redundant" information on consumption, income and human capital endowments -which is endogenously determined- is necessary and useful, given that the utility of households is unobservable and traditional statistical methods of model fitting do not apply when categorical information is heavily used. The method used, "Optimal Scaling and Alternate Least Squares" solves the statistical challenges as it imposes a metric to the categorical variables that are used to compose the index in order to maximize their contribution to a model of principal components. In addition, in the case of SISBEN the ALS-OS algorithm was forced to concentrate the optimal scaling adjustments according to their contribution to the first principal component, which precisely reprpesents the utility index of every household. For the data set -from CASEN household survey, the ALS-OS algorithm converges ans is stable to different starting configurations. The procedure selects a relevant set of variables to identify the poor and the signs, values and the order that this algorithm assigns to the different categories of all variables are consistent with economic intuition on consumption and poverty.
Last updated: 05/08/07
