Generating Small Area Statistics for Household Income in Southern Province of Sri Lanka

Abstract

Human development is a central objective of economic activity and forms the main purpose of social investment in any country. The Human Development Index (HDI) defined by the UNDP 1 is based on three dimensions: namely, opportunity to lead a long and healthy life, ability to acquire knowledge and learning, and power to access adequate resources. Sri Lanka’s HDI is by far the highest in South Asia 2 and exceeds that of some developed countries. This high level of achievement is due to provisions made for accessing health and education, and continued investment in the social sectors. Country’s economic development on the other hand is lagging consistently behind, and appears to have bypassed the rural sector where nearly 80% of the population live. Human poverty 1 is a result of inability to meet the three social dimensions mentioned above. Sri Lanka’s level of poverty is moderately high, with a high proportion (56%) having no electricity, a moderate proportion (28%) without access to safe drinking water, and a considerable proportion (24%) without adequate sanitation facilities.3 The country has gone through several poverty-alleviation programs in the past. However, the success of any such program would largely depend on the availability of up-to-date and reliable poverty related information at small geographic levels. According to latest information 4 2 88% of the poor resides in the rural areas compared to 8% in urban, and 4% in the estate. Another study 2 concluded that, this 4% is among the poorest in Sri Lanka. These figures tell us that more work is involved in getting reliable information from the rural sector, if these programs are to be effective. “Samurdhi” is the largest single welfare program, which accounts for nearly 1% of the GDP, aimed at poverty alleviation in Sri Lanka.2 Southern Province is reported to have a high percentage of Samurdhi recipients and therefore, this study is focused on the Southern Province. The effectiveness of any such program would largely depend on the reliability of systems that provide information at the smallest geographic levels that need food assistance. Censuses conducted every ten years and surveys conducted more often than that, are the only sources of information on poverty variables. Even though censuses provide reliable information on the smallest of geographic areas such as Divisional Secretariat (DS) or Grama Niladari (GN) levels, the information at these levels is updated only once in a decade. Surveys do not bridge this gap either, as information is available only at larger administrative units such as provincial or at most district, due to budgetary constraints. The present work is motivated by the fact that in Sri Lanka no attempt has been made in the past to come up with reliable and up-to-date estimates of poverty related variables at small geographic levels. Therefore, in this paper we demonstrate how small area statistics could be generated by a method called small area estimation 5, 6, 7, .. which has been successfully used elsewhere. The method combines information from censuses and surveys through a statistical approach. We have focused on an important variable, household income.