On the Choice of a Policy-oriented Poverty Measure: The Case of Israel 1997-2002

Poverty is easily recognized when encountered "face to face," but when measured for a society as a whole it becomes evasive and ambiguous, raising fundamental and practical issues. In this paper we examine several poverty measures from the aspect of how well they can serve as policy targets and to what extent they can be used to monitor the effect of social and economic policy on poverty. A rational poverty-reduction policy requires a quantifiable poverty target which has a sufficiently long time horizon and which enables shorter term policy monitoring. The measure should also reflect poverty intensity, as does the distribution-sensitive Sen poverty measure. Definitions of poverty often reflect a lack of basic needs. The vital food component can be objectively determined by physicians, but other components of the essential basket of goods and services, such as clothing, shelter, health services and education require a public consensus. We suggest a policy-oriented choice based on various poverty definitions. We argue that the measure should be based on basic needs rather than the relative approach. The poverty threshold should be based on an absolute gender- and age-determined food component, derived from an objective, medically-determined dietary reference intake (DRI) that fosters sustained health, thus resembling the Canadian Market Basket Measure (MBM). In the other components our proposal follows the methodology of a relative-needs-based approach recommended by the American National Research Council (NRC). We follow NRC methodology concerning the treatment of income resources by taking into account the relatively widespread incidence of owner occupancy in Israel, even among the poor. Unlike the NRC we deduct only interest payments on mortgages and housing loans. However, we deviate from that approach with respect to health and education: We reflect health services twice in our poverty measure: (1) we include average out-of-pocket health expenditure in the threshold, and (2) we deduct health expenditure in excess of this average in our calculation of income resources. While the suggested measure is a significant improvement on the official half-median poverty measure, there remain important shortcomings that need to be addressed in the future: these include the need for improved knowledge about geographic differences in the quality and quantity of essential public services such as health, education, social services, housing, infrastructure. It is important to include such information in a policy-oriented poverty measure in order to get a full account of the poverty situation and of the effect of the government's and other institutions' policies on poverty. Nevertheless, we believe that the proposed measure provides a basis for immediate implementation of anti-poverty policy.