THE DESIRABLE PROPERTIES OF A
POVERTY MONITORING AND ANALYSIS SYSTEM
(PMAS)

A poverty monitoring and analysis system may be conceived as an input-output device. One may distinguish three basic elements in this setting. Human and non-human resources are mixed in an institutional framework to produce information on living standards. Thus the desirable properties of such a system may be discussed in terms of : (1) the information which is put out, (2) the structure and functioning of the relevant institutions, and (3) the resource base.

With respect to the output, a good poverty monitoring system should produce information on the distribution of living standards that is policy-relevant, reliable and timely. Policy relevance has implications for the structure of the needed information. Information may be qualitative or quantitative. To take effective action to reduce poverty, it is not enough to know where and when poverty is greatest, it is also important to know why. In this respect, it may well be more desirable to combine both the quantitative and qualitative approaches to poverty measurement and analysis. Carvalho and White (1997) argue that such an integration is done at three levels: (i) integrating methodologies, (ii) conducting an integrated analysis whereby the findings of one approach are confronted with those of another to confirm, refute, enrich or explain them, (iii) merging the findings of the two approaches into an integrated set of policy recommendations.

Another structural aspect of data produced by a monitoring system relates to policy targeting. When the policy objective is focused on poverty alleviation, a picture of the distribution of the standard of living at one point in time provides not only a benchmark against which to assess future progress, but also a means for the identification of target groups, commodities or activities that deserve special attention in the design and implementation of poverty alleviation policies. What then are the socioeconomic groups that deserve special attention in such a context?

The answer depends on the costs and benefits of targeting. In principle the argument for targeting is based on cost-effectiveness of policy implementation. In this context, effectiveness is to be understood as a matter of maximizing the poverty alleviation benefits associated with an accurately targeted policy from a given cost (Sen 1995:12). However there are both direct and indirect costs involved. Such costs relate to the need for identification and policing, and from the fact that a targeted policy may create incentives for beneficiaries and others to change their behavior. The more accurate the targeting the costlier. Therefore, in the search for characteristics by which to classify individuals or households, it is desirable to consider in priority those that may be difficult to adjust (such as gender and age) and those that lead to the identification of broader groups (such as the region of residence).

There are other considerations that make age, gender and geographical regions important characteristics for the disaggregation of information on living standards along the profile of a population. Children and women deserve special attention in the fight against poverty because they tend to be among the most vulnerable groups and also because investment in these groups tend to have higher social returns than for other groups in society (as in the case of health and education). Furthermore, age and gender concerns have implications with respect to the choice of the unit of observation. In the case of data collected by survey methods, the household is the standard unit of observation. If the well-being of women and children is an important social concern, then a good poverty monitoring system should provide information on the distribution of the living standard within the household.

The geographical dimension of the data gathered is also important in connection with decentralized planning. Thus regional disaggregation should be consistent with the governance structure in place. Such a structure is understood as the distribution public decision power among various levels of government.

What dimensions of the living standard should be observed? This issue is important given the fact that the living standard is a multidimensional concept with income and non-income dimensions. Furthermore, poverty in any country may be defined as a low level of achievement in terms of the living standard and the explanation of such an observation may be sought in constraints on both individual and collective ability to achieve. These considerations suggest that a good monitoring system should provide information on demographics, education, health, anthropometrics, social infrastructure at the community level, assets and household durables, employment and productive activities, and prices.

The above discussion of information structure leads us to the conclusion that it is desirable to have a monitoring system that provides a comprehensive and integrated information set on the living standard of the population. The data set is integrated to the extent that it combines both qualitative and quantitative information on both the individuals and the community in which they live. It is comprehensive both with respect to the population and the dimensions of well-being that it covers. Such comprehensiveness guarantees unbiasedness of the well-being indicator.

The quality of information goes beyond the unbiasedness of the measurement method. It involves events at different stages of the process: collection, storage and retrieval, processing (cleaning and such), analysis and dissemination. It is desirable to capture information as it is born. This concern has implications for both surveys and the collection of administrative data. For surveys it may be desirable to input data in the field and conduct initial checks at the source of information. For administrative data, it is important for data quality to integrate the statistical function to most administrative processes so that policy-relevant information is collected and processed in a timely fashion.

As stated earlier, an information set with certain attributes is an output of an institutional setting. Therefore the production of reliable and policy-relevant information on a timely basis has institutional implications. There ought to be a legal framework (or social contract) to facilitate the interaction between the population and the "information workers" or statistical authorities. In particular, the rules of the game should be such that socioeconomic agents would feel free to provide accurate information about their status and activities. Furthermore, it is desirable to have the statistical agency working in collaboration with planning agencies and policymakers. In any case it is useful to have the monitoring system integrated to the policy process. Such an integration would help determine the frequency with which each type of information is collected and analyzed.

Finally, it is important to note that the desirable properties of the informational input to the policy process and those of the associated institutional framework cannot be achieved on a sustainable basis unless there is also a sustainable flow of resources in terms of skilled manpower, material and funds. Any deficiency at this level or at that of the institutional setting is bound to have debilitating effects on the ability to conduct effective poverty alleviation.