How poor is poor?

The exclusion criteria adopted to determine the poor are too expansive and exclude many households that are likely to be poor, while the inclusion criteria are too limited and may not have captured many such households and individuals, especially vagrants and homeless people. Here, a slum clearance board tenement in Salem.

The results of the Socio Economic and Caste Census (SECC) conducted in 2011 were finally released to the public in early July 2015. And even this was only a partial release, with data covering only rural India and that too based on a combination of draft (351 districts) and final (277 districts) lists, with lists for 12 districts not yet available. The urban data (for which a different methodology has been used) are not yet available. Also, while caste data were collected for each household, these have not yet been released despite the demand for their release from several quarters.

The data emerging from this census have been eagerly awaited for several reasons. Unlike the National Sample Surveys, which explicitly do not cover the entire population, this was a census that supposedly included every household (although it is likely that the SECC—as in the case of the census on whose lists it was based —did not actually fully cover groups such as the homeless and migrants). The data will form the basis of the lists of “poor” and “non-poor” that are used by both Central and State governments to determine the beneficiaries of various government programmes as well as access to a wide range of other publicly provided goods and services. Since the strategy of targeting “the poor” appears set to continue despite the many concerns with it and the greater advantages of universal provision, this is obviously a matter of great economic and political significance.

And then, of course, there is the political hot potato of caste-wise distribution of material conditions, which should also be revealed by this census. It is rather surprising that despite the massive upsurge in caste-based identity politics over several decades in India and the use of caste-based reservation in public education and employment, we do not have any recent reliable indicators of the actual conditions of people according to caste. The SECC was supposed to fill this gap.

The current difficulty with releasing caste data appears to be that (unlike all the other data collected in this census) these were not codified, but rather filled in on the basis of information provided by individual respondents, so that several thousand caste names have been recorded. The State governments have been asked to match these names with those in their own lists of Scheduled Castes/Scheduled Tribes/Other Backward Classes/Others, and many States have not yet done so. At least, that is the official story. But it does nevertheless strain credibility that such lists (on the basis of which caste certificates are routinely provided by all State governments) could not be accessed by a central agency that could simply categorise all households according to the classifications already used by other statistical agencies in India like the National Sample Survey Office.

However, even without the caste data, it is clear that the SECC provides a wealth of information on living conditions in rural India and enables a multidimensional understanding that is sorely required in public policy formation. But then the important question is, how will these data be used for policy purposes?

The best way to deal with these more comprehensive household listings would be to avoid creating binaries of “poor” and “non-poor” that become the basis for differential access to public programmes. This strategy is now well known to be beset with problems of unfair exclusion (Type I error) and unjustified inclusion (Type II error). It also leads to a mismatch between requirement and delivery, as when a household defined as non-poor is deprived of access to something that it is clearly deficient in, such as, say, pucca housing or access to subsidised food.

Instead of the blanket approach, it makes more sense to look at the evident gaps in different dimensions, such as housing, access to utilities and basic infrastructure, nutrition, health and sanitation, education, and secure employment, and to develop policies designed to address these gaps separately with respect to all the households found to be lacking in that particular dimension.

Sadly, all indications are that this nuanced strategy will not be adopted. Rather, the government will stick to its plan of simply determining a category of “poor” that will then be used across the board for different public programmes. This is likely to be a huge mistake, and one that in all likelihood will prevent government policies and programmes from reaching many of the people who deserve them and who they are intended to serve.

The method of determining the poor that has been adopted in this census is as follows: There is to be automatic exclusion of households on the basis of 14 parameters, automatic inclusion on the basis of five parameters and grading of deprivation on the basis of seven criteria, which will determine the final list of those defined officially as “poor”. Unfortunately, the exclusion criteria are too expansive and exclude many households that are likely to be poor, while the inclusion criteria are too limited and the SECC may not have captured many such households and individuals (especially vagrants and homeless people, inter alia ).

What this means is that the multidimensional poverty estimates derived accordingly are likely to avoid Type I errors (that is, they will not overcount the poor) but will still be prone to Type I errors (that is, they will exclude many households and people who should be considered as poor). This underestimation will then get reflected in reduced public allocations, which will further affect public spending on social programmes that are already being substantially cut. This also explains why nearly 40 per cent of rural households have been classified as “automatically excluded” even when a significant proportion of them show incomes of less than Rs.10,000 a month.

It also explains why less than 1 per cent of rural households have been automatically included in the list of the poor, when there is likely to be a much higher proportion of people who should meet even these limited criteria (which are: those without shelter; those living on alms; manual scavenger families; primitive tribal groups; legally released bonded labour). It is worth noting that the pilot survey conducted by the Ministry of Rural Development actually found much higher proportions of landless and destitute people as well as legally released bonded labour.

Exclusion list The exclusion list may be the most problematic. Consider the criteria that are being used to automatically exclude households from being defined as poor. Exclusion from the list will occur if a household fulfils any one of the following criteria:

i) Motorised two/three/four wheeler/fishing boat; ii) Mechanised three-four wheeler agricultural equipment; iii) Kisan credit card with credit limit of over Rs.50,000; iv) Household member a government employee; v) Households with non-agricultural enterprises registered with government; vi) Any member of household earning more than Rs.10,000 a month; vii) Paying income tax; viii) Paying professional tax; ix) Three or more rooms with pucca walls and roof; x) Owns a refrigerator; xi) Owns landline phone; xii) Owns more than 2.5 acres of irrigated land, with one irrigation equipment; xiii) Five acres or more of irrigated land for two or more crop seasons; xiv) Owning at least 7.5 acres of land or more, with at least one irrigation equipment.

Exclusion on the basis of any one of these is far too stringent and ignores the current realities of rural life. For example, many households engaged in petty self-employment, for example retail trade, may have no choice but to own a motorised two-wheeler (often purchased second- or third-hand at very low prices) to engage in this work, and yet earn much less than Rs.10,000 a month. This is especially the case in hilly areas and other places where such motorised transport is not a consumer durable but a means of livelihood, even if it does not necessarily enable an escape from material poverty.

Similarly, many poor fisher families have motorised boats—indeed it is now impossible in many coasts to fish at all without one—but still have extremely low and precarious incomes. Many farmers have been given Kisan credit cards of relatively high value, which do not reflect their monthly incomes, and they could even be in substantial debt that further erodes their disposable income and makes them poor.

The housing criterion is similarly problematic. It is quite common in rural India for extended families to occupy a single dwelling, which is therefore relatively large, even though they would de facto be separate households. There are also cases of very large households (found often, for example, among Muslims and in some castes in north India) who would therefore have pucca dwellings of three or more rooms, but still actually be poor.

Deprivation points Clearly, therefore, both the automatic exclusion and automatic inclusion criteria would be erring so as to significantly undercount the poor. The seven deprivations that are considered for grading would also have precisely the same effect. Most of them should result in automatic inclusion in the category of poor. Instead, each deprivation is to be allocated points and only those achieving the requisite number of points will get counted among the poor. The seven criteria are as follows:

i) Households with only one room, kuccha walls and kuccha roof (which constitute an astounding 13.25 per cent of rural households, or nearly 24 million households); ii) Households with no adult member between the ages 18 and 59 years (3.64 per cent); iii) Female-headed households with no adult member between the ages of 18 and 59 years (3.85 per cent); iv) Households with differently abled member and with no other able-bodied adult member (0.4 per cent); v) S.C./S.T. households (21.53 per cent); vi) Households with no literate adult above the age of 25 years (23.52 per cent); vii) Landless households deriving a major part of their income from manual labour (29.97 per cent).

Indeed, if only the first and last criteria, which are clearly indicative of being poor, had been included automatically in the list of poor, the proportion of households in that list would have gone up by 30 times! Instead, by putting such households in a graded category, there is a real chance that many of them will end up being excluded when their material conditions are clearly adverse and calling for special consideration.

There are many other shocking or depressing results of the SECC, which deserve to be discussed in greater detail than can be done here. But the main point to note is that the methodology for determining the poor is deeply flawed, even with the available data emanating from the SECC. Indeed, the SECC data themselves suggest that the methodology for determination of the poor must be rectified urgently if the resulting determination is not to be a travesty of the original intention. And, in any case, governments at all levels should use this information to realise the extent of deprivation in terms of different indicators and take active steps to deal with those.