The measure of poverty

Estimates based on SECC and NSS data have different purposes.

By: C. Rangarajan and S. Mahendra Dev 

Recently, the government released data from the Socio-Economic Caste Census (SECC) 2011. There has been comment that hereafter, we need not have consumption-based poverty estimates using NSS (National Sample Surveys) data. It is thought that SECC data will alone be enough to estimate poverty and deprivation. Here, we briefly examine the differences between the two and clarify that NSS consumption-based poverty estimates are still relevant. SECC-based estimates are important, but no substitutes for NSS-based poverty ratios. In India, we have a long history of studies on the measurement of poverty. The methodology for the estimation of poverty used by the erstwhile Planning Commission was based on recommendations made by various expert groups. In June 2012, the government of India appointed an expert group (with C. Rangarajan as chairman) to take a fresh look at the methodology for the measurement of poverty. 

The Rangarajan expert group has gone back to the idea of separate poverty line baskets for rural and urban areas, unlike the Tendulkar Committee, which took urban poverty as a given and used it as the common basket for rural and urban households. In defining the consumption basket separating the poor from the rest, the new expert group took the view that it should contain a food component that satisfied certain minimum nutrition requirements, as well as some normative level of consumption expenditure for essential non-food item groups (education, clothing, conveyance and house rent) besides a residual set of behaviourally determined non-food expenditure. The introduction of norms for certain kinds of non-food expenditures is an innovation. In the absence of any other normative criteria, the median fractile class expenditures were treated as the norm. Based on the analysis presented in the expert group report, monthly per capita consumption expenditure of Rs 972 in rural areas and Rs 1,407 in urban areas is treated as the poverty line at the all-India level. Assuming five members for a family, this will imply a monthly per household expenditure of Rs 4,860 in rural areas and Rs 7,035 in urban areas.

 The expert group estimates that 30.9 per cent of the rural population and 26.4 per cent of the urban population were below the poverty line in 2011-12. The all-India ratio was 29.5 per cent. Poverty estimates provide the proportion and size of the poor population and their spread across states and broad regions. But they - cannot be used for identification of the individual poor, which is necessary to ensure that the benefits of programmes and schemes reach only the deserving and targeted group. After the release of the SECC estimates, some commented that earlier targeted programmes were designed based on sample surveys and the SECC was an innovation. This is not true. Previously, too, for identification of the poor, BPL (below poverty line) censuses were conducted. The first two BPL censuses (conducted in 1992 and 1997) yielded an estimate of the percentage and number of poor households at the village, block, district and state levels, and the beneficiaries in these programmes were chosen by state governments depending on their position on the BPL list. The third BPL census was conducted in 2002. It did not identify the number of poor households straightaway or estimate their numbers, as in the previous two censuses. Instead, it ranked households within the village in terms of their socio-economic status, based on 13 indicators reflecting the levels of living and quality of life. SECC 2011 is thus part of this continuing process. 

According to the SECC data, 8.69 crore out of all rural households have one of the seven deprivations. In other words, 48.5 per cent of all rural households suffer from at least one of seven deprivations. Thirty per cent of households suffer from two deprivations while 13 per cent have three. Only 0.01 per cent households suffer from all deprivations. The automatically included contributed 0.92 per cent of the total rural households. Information on urban households is not yet publicly available. The rural poverty ratio estimated by the expert group based on NSS data was around 30.9 per cent in 2011-12. This is almost equal to households with two deprivations, plus the automatically included. However, NSS-based estimates are per capita, while the SECC data refer to households. A look at seven deprivations shows that they are not deprivations in the conventional sense of deprived of income, health and education, etc. Therefore, the question is whether the SECC data is appropriate for estimating poverty ratios. For example, it is true that landless households deriving a major part of their income from manual labour constitute the largest number of under-deprivation households. It is not clear whether landlessness (or manual labour) can be sufficient to conclude they are suffering from poverty. Over time, landlessness will increase and people will diversify their income with a rise in non-agricultural activities and migration. In the same way, some of the other criteria are not clear indicators of poverty. What is the rationale for having poverty estimates based on consumption estimates? First, in the minds of most people, being rich or poor is associated with levels of income. The various non-income indicators of poverty are in fact reflections of inadequate income. 

Defining poverty in terms of income or in the absence of such data in terms of expenditure seems more appropriate, and it is this method that is followed in most countries. Second, historically, the number of identified poor based on successive BPL -censuses in rural areas has differed widely from the measured poverty. For example, the percentage of households identified as poor in the first BPL census in 1992 was nearly twice the poverty ratio estimated by the Planning Commission. Usually, the poor households identified through these censuses contain a mix of poor and non-poor, for which there could be several reasons. One of the main reasons behind such a mix-up is the fact that people know beforehand that the census is going to decide the status of the household as poor or non-poor, and therefore its entitlement. Third, the deprivation criteria by themselves do not indicate the level of poverty. A judgement has to be made as to whether the number of deprivations taken together constitute a measure of poverty. This can be highly subjective. What is the purpose of NSS consumption-based estimates if they are not used for the allocation of funds and capping beneficiaries under government programmes? These poverty ratios would be used basically to assess the changes in poverty at national and state levels, or at the district level. This will be useful for creating appropriate policies. For example, one can examine changes in poverty in different phases of the post-reform period to understand what impact anti-poverty programmes, in conjunction with growth, have had on poverty. To conclude, SECC and NSS data-based estimates have different purposes. The SECC would be important for the identification of beneficiaries of programmes while NSS-based estimates would be useful for assessing changes in levels of living at the macro level over time. -