Towards an Asset-based Indicator of Poverty

Consumption and Poverty

The limited focus of the studies measuring poverty in India is quite evident when we distinguish between the two realms in which poverty manifests itself: the tangible and the intangible. Tangible poverty deals entirely with the inadequacy of elements that have a physical form, like the products consumed by the household. Intangible poverty deals with acute inadequacies that are not entirely discernible in physical terms. The exclusion of individuals and groups from social consultations, like keeping women away from discussions that result in actions that affect their safety, would be an example of intangible poverty. The discussion on poverty in India, and, indeed, elsewhere, has largely confined itself to tangible poverty.

This may be treated as no more than an issue of the study of poverty being concentrated within the discipline of economics. But the task of eradicating poverty is not a matter of a single discipline and is deeply influenced by the distinction between tangible and intangible poverty. Poverty in the tangible domain can be removed by the redistribution of the materials involved. Inadequate income can be addressed by increasing the income of the poor; limited consumption can be addressed by increasing access of the poor to goods and services. Poverty in the domain of the intangible, in contrast, cannot be completely removed by the provision of physical elements. It is sometimes convenient to view intangible poverty in terms of its effects on the material domain. And it is not unusual for intangible poverty to be reflected in the material domain as well. Women could be paid less than men for doing the same work; labour controlled through extra-economic coercion can be expected to work without pay. But to reduce intangible poverty to its tangible manifestation alone loses several elements of the multifarious nature, and hence the intensity, of poverty. Guaranteeing women equal pay for equal work would undoubtedly be an important step forward, but it need not, and usually does not, remove all aspects of gender discrimination in the workplace. Reservation for Dalits is another significant step forward, but it too does not guarantee the removal of intangible forms of exclusion.

Within the realm of the tangible, the study of poverty through the lens of consumption is further constrained by the role of the market. The actual levels of consumption of a household consist of products and services that are accessed through the market, as well as those that are accessed through other sources. The non-market support a household gets would include support provided by the extended family to a household in distress. It would extend to support provided by social groups. This support could take a variety of forms, ranging from the support of caste organisations, to free food being made available by religious institutions. The state too plays a prominent role in providing non-market support. Examples of such support include free healthcare or the supply of free or subsidised food. Any measure of poverty based on consumption expenditure looks only at goods and services a household buys and is hence confined to the realm of what is marketed.

A view of poverty even in the realm of the tangible that is based only on marketed products is quite severely limited. The effect of marketed tangible poverty can be offset by tangible benefits provided by the family, society and the state. Individuals gain a variety of material benefits from the household as a whole. A woman who does not earn an income, and is even deprived of money she can call her own, certainly faces a form of tangible poverty. Yet to the extent that she has access to the material assets of the household, say a television, some part of that poverty is offset to howsoever small an extent it may be. Households can also get the benefit of support from members of a family who do not belong to their household, in the sense that they do not share their kitchen. This could take the form of a member of the extended family playing the role of a caregiver when someone in the household is ill. It can also extend to support in the city for a member of the family. A family member coming from the village can be accommodated in the household for as long as it takes him to find a job in the city. At the same time, workers who have come to the city have been known to leave their small children with grandparents in the village (Pani & Singh, 2012). The support system could extend to caste networks, as when a person going to the city stays in the household of a member of his caste. There are also social institutions that can provide the material support of providing free food. There are a variety of examples of this phenomenon, from the langars in Sikh Gurudwaras to more family-related traditions in Karnataka (Karanth, 2014). There are thus potentially substantial socially determined offsets to levels of material poverty determined by the market.

Alongside these socially determined offsets, there are also efforts of the state to reduce the effects of poverty. A welfare state provides a number of measures designed to offset poverty. Some of these measures are directly reflected in the expenditure of a household. One of the larger welfare measures in India is the Mahatma Gandhi National Rural Employment Guarantee Scheme. This scheme guarantees 100 days’ work per family at fixed wages. This work and the payments received are expected to offset the poverty of the households, which have individuals willing to do physical labour. Since this benefit is through wages, it will result in increased expenditure by the households and hence be reflected in consumption-based measures of poverty. This would also be true of direct fund transfers to the poor. But there are other state benefits that are provided in kind. These benefits accrue to the household without being a part of their expenditure.

Arguably, the two most important of these non-market material benefits are the subsidies received by the household when procuring food from the public distribution system (PDS) and the benefit of free healthcare in government hospitals. The food grain procured from the PDS is paid for and hence forms a part of a household’s expenditure. But to the extent that there is a subsidy involved, the expenditure of the household is less than what it would otherwise have paid. That is, the household has received more grain than it would have otherwise received for the same amount of expenditure. This additional grain is a material benefit that the household does not pay for, and it is hence not reflected in its consumption expenditure. The material support of public health systems also takes multiple forms. In its rudimentary form, it may be the useful but limited access to an Accredited Social Health Activist (ASHA). At the next level, it would be access to a doctor at a Public Health Centre (PHC). At a higher level, the person would have access to a community or government hospital. In each case, the household has gained a benefit which is not reflected in its consumption expenditure.

It is also important to remember that expenditure is not an end in itself. And as a means to an end, the same level of consumption expenditure can mean different things to different households. As Amartya Sen has suggested when making his case for the Capability approach, a household that has to look after people with disabilities, or with older dependents, would need to have higher levels of consumption expenditure without necessarily enjoying a better quality of life (Sen, 2000). Comparisons across households would be further complicated if they live in very different ecological zones. Individuals and households in the northern parts of India would need to have warm clothes to deal with severe winters, something that their counterparts in the southern parts of the country would not need. Social customs in different parts of the country would also place different demands on household consumption expenditure. In addition to differences across regions, there are also differences in situations households face over time. There could be years when households are able to afford relatively higher consumption expenditure and years when circumstances force a drop in consumption expenditure. While this difference is usually ignored, it can be a major limitation in situations where incomes vary very substantially from year to year, as usually happens under rainfed agriculture. In a good year, a prudent farmer household may save for non-rainy days, while a less cautious farmer would increase his consumption expenditure. A consumption expenditure–based estimate of poverty would see the prudent farmers as being poorer than the less cautious one, though, in terms of the resilience of the household to the longer-term vagaries of rainfall, the former would be better equipped than the latter.

The use of consumption expenditure as the basis to measure poverty tends to be quite insensitive to those facing the greatest deprivation. Decades ago, in the 1970s, it was pointed out, based on studies of rural Bihar, that the bulk of the rural poor households were “deficit ones in the sense that their bare minimum consumption expenditures exceed their incomes. They are thus forced to take consumption loans both in cash and in kind from the big land owning class who are the main constituent of the rural rich” (Prasad, 1974, p. 1305). Agrarian relations have undoubtedly seen substantial transformation in the succeeding decades. Yet the demand by the rural poor for small consumption loans has continued (Khandelwal, 2007). The extreme deprivation faced by a section of rural households forces them to borrow for consumption. Since consumer expenditure data will only capture what they have spent, the aspect of dissaving is completely ignored. These households are put on par with those sections of the poor who still retain the ability to carry out the same levels of minimal consumption without having to borrow. The difference between those facing such extreme deprivation that they are forced to borrow for survival and those among the poor who are able to pay for the same levels of minimal expenditure is thus lost in data on consumer expenditure.

The difference between those who need to borrow for survival and those who can afford to meet basic survival expenditure may be small in terms of absolute numbers, but when seen in terms of the experience of poverty, the difference is far from trivial. Those who need to borrow for survival are only a step away from hunger and even starvation. Their need for loans for the very survival of their families can place them at the mercy of the more exploitative sections of local society. The need to break out of this dependence on informal credit systems is reflected in the extremely high charges the poor are willing to pay for loans from formal microfinance institutions. Some estimates of the average of these charges in India put it as high as 33 per cent (Pauli, 2019). This level of deprivation is substantially greater than that of those who can avoid this forced indebtedness, even if their household consumption expenditure levels are not very different.

The difference between those who are dependent on consumption loans and those who are not is further complicated by the temporariness that can sometimes enter the picture. There could be individuals who break out of the extreme dependence in one year only to fall back into it in another year. A marginal cultivator may be able to break out of indebtedness in a good rainfall year only to fall back into it in a drought year. This variability, and the resultant uncertainty, at the bottom of the pyramid adds to the difficulties of using consumption expenditure as an indicator of poverty. Consumption expenditure presents conditions in a particular year, while the experience of poverty is a longer, drawn-out process. The debts the poorest have to take for survival in a bad year have to be repaid in the relatively better years. And even in the good years, the relationship between the poorest and those higher up in the economic hierarchy would be dependent on the possibility of those at the bottom of the pyramid having to borrow for survival in bad years. Consumption expenditure does not necessarily reflect the experience of poverty by the poorest.

The need to address differences among the poor—to sufficiently distinguish between the poorest and those just below the poverty line—has contributed to the emergence of a number of measures from Sen’s early work (1976) to later indices, including the Poverty Gap Index of Deaton and Dreze (2002). To the extent that these indices are designed to be used on consumption data, they are not immune to the inherent difficulties of a consumption-based approach. These measures also cannot distinguish between those at the lowest level of consumption who need to borrow to avoid starvation and those who can afford similar consumption patterns without having to borrow.

Assets as an Indicator of Sustained Poverty

The critique of consumption-based measures of poverty would not be of much practical value unless we can provide an alternative measure that is not affected by most, if not all, of the difficulties of relying on consumption to develop a picture of poverty. As we have seen, the list of difficulties in using consumption is not small. A comprehensive approach would seek a measure that captures both the tangible and intangible aspects of poverty. Even if we were to be less ambitious and focus on the realm of the tangible alone, there are serious shortcomings of the use of consumption expenditure as an indicator of poverty that need to be overcome. The need to recognise the role of non-marketed support is far from trivial in India. Given the prominent role played by the PDS in most states, and by the provision of public healthcare in some states like Kerala, the condition of the poor can be deeply affected by state support. In some parts of the country, non-state social support such as the provision of free food, education and healthcare for the poor by social institutions is also not small. To treat a household that has a low-consumption expenditure because it enjoys these benefits on par with a household that has the same level of consumption expenditure without these benefits would clearly distort the measure of poverty. The long list of difficulties with using consumption expenditure as an indicator of poverty places a set of stringent requirements on any alternative measure. The rapid growth of the microfinance sector with its high interest rates also points to the magnitude of the difference between households that survive on dissaving and those with a similar consumption expenditure without the burden of debt. And with uncertainty in price and yield variations far from minimal in rural India, the difference between the pictures of poverty provided by annual consumption figures and that by a longer-term measure would also not be trivial.

One alternative to consumption expenditure, which is not prone to these difficulties, is to use assets as the basis for the measurement of poverty. The ownership of an asset is likely to better capture the longer-term economic condition of the household. This would be particularly true when the asset reflects the long-term savings of the household. An asset may have been bought in a good year, but if it remains in the household, it has withstood the pressure in a bad economic year to sell or mortgage it. The availability of an asset to sell or mortgage would also help the household stay out of particularly exploitative debt. The ability of a household to save enough to buy an asset would also reflect the overall economic status of the household, irrespective of whether that status has been achieved through the procurement of marketed products or through support measures provided by the extended family, social institutions or the state.

In addition to these universal considerations, there are at least two features of the Indian economy that make assets a more useful indicator of the economic condition of a household. First, it is now quite well established that there is a high-level informality in the Indian economy (Agarwala, 2013). Apart from relatively better-studied aspects like informal rural credit, there are also informal elements within what is otherwise a formal process. A family moving from one city to another may, in line with a formal economy, hand over the entire task of packing and shifting their household items to a company that specialises in this activity. If in the process they find an asset they do not believe is worth taking along, say a gas stove, it is not unknown for them to hand it over to one of the workers helping them move. The assets in the worker household would then include those that have not been bought through savings from the income the worker formally earns. Second, we also need to take into account the levels of rent-seeking at multiple levels within the Indian economy. When bribes are paid in kind, they could take the form of household assets. The focus on assets would thus provide a more realistic picture of the economic condition of an Indian household.

The value of focusing on the stock of assets, rather than the flow of household income and expenditure, has been recognised in studies of race and poverty, beginning with Michael Sherraden’s pioneering work in 1991 (Sherraden, 1991). Later work in the same tradition has explored wealth in different race groups (Oliver & Shapiro, 2006). A focus on assets helps explore the ability of the poor to save and even, in a few cases, invest their way out of poverty. We could build on this tradition to develop an alternative to consumption expenditure as the lens through which we can estimate poverty in India.

Taking on board all assets is, however, not entirely suited to capturing the experience of poverty. The same monetary value of assets could have very different effects on the quality of life at different levels of the economic hierarchy. The monetary value of differences that attract attention at the very top of the range of incomes would tend to be very much larger than the differences at the bottom of the asset range. A difference of ₹10,000 in the value of the assets of the two richest households in a population in India may not be considered very significant. In contrast, for two households at the very bottom of the rural asset range, the same absolute amount could mark the difference between households that face starvation and those that do not.

The need to be more sensitive to conditions at the bottom of the economic hierarchy in India is also accentuated by the localised experience of poverty. A local community may be poor as a whole, but the experience of poverty of the poorest in that community could be quite different from that of those on top of the local economic hierarchy. It is quite possible for the household at the top of the economic hierarchy in a remote village to have so few assets that it would be considered poor in any national reckoning. Yet the head of this household, and usually its other members, could wield considerable power within the village, leading to their being able to tap the services of the worst-off in the village through extra-economic coercion. A measure of poverty in India would then need to be sensitive to both the extent of poverty and the variation in levels of deprivation.

Measuring Deprivation

An effective measure of poverty, which is sensitive to the extent and variation in deprivation, would be easier to detail once we make explicit the very different requirements of a measure of poverty from one that seeks to capture inequality. An index of inequality must necessarily cover the entire range of the indicator that is being used to measure the phenomenon, whether it is income, consumption or assets. This would be essential to make any meaningful comparison between the rich and the poor. When poverty is seen entirely in relative terms, it can be defined as the lower end of an overall index of inequality. But such an estimate would not capture the different experiences of the same assets at different levels in the economic hierarchy. The experience of having a cycle in a poor household is very different from that of a person in a rich household owning a cycle. Again, the variations in the levels of deprivation that are critical to the experience of poverty in a poor village would fade into insignificance when compared to the tangible inequalities that exist at the national or global levels. An effective measure of poverty would then not see it as just the lower end of a range of inequalities, but as an experience in itself. Such a measure would then need to focus on the experience of the poor, including the different experiences between those facing starvation and those who do not; between those poor who are at the top of the economic hierarchy of a poor village and those who are at the bottom of that hierarchy.

Such a focus on the lower end of the economic hierarchy would provide several academic freedoms. It would allow us to ignore the very rich, and their assets, thereby allowing for a closer examination of the assets of the poor and others who are in the lower half of the economic hierarchy. It would track more closely the ownership of assets of the poor and the middle class. It would necessarily include assets that the poor would consider to be realistically within their capabilities at least within a reasonable future, such as a cycle. It would go on to include assets that would be owned by the non-poor, while ignoring those that would only be owned by the rich. A set of assets, chosen with some sensitivity to the conditions of the poor and the middle class, would provide an indicator of the distance of a household from absolute deprivation.

In such an asset-based index of poverty, a household that owns no assets at all can be seen to be facing absolute deprivation. Such a household would not have been able to save enough to buy even the cheapest asset that poor households tend to own. We can then calculate the distance of individual households from absolute deprivation in terms of the value of the assets they own. In practice, much would depend on the assets that are chosen. The cheaper among the chosen assets should reflect what a poor household would reasonably be expected to own. This would be, in twenty-first century India, a bicycle, and in the past may have been no more than a transistor radio. The higher end of the range of assets would extend to what a middle-class household would expect to own, such as a car. The price of each asset could then be normalised, treating the price of the most expensive asset as being equal to ₹100. A household that does not have any of these items would be classified as facing absolute deprivation. An index of the distance from absolute deprivation (IDFAD) can then be calculated from the quantity of each of the assets in the household and their prices. The sum of the normalised values for each of the assets the household owns can then be taken to be the IDFAD of the household. That is:

I D F A D j = ∑ ​ n i s i

Where

IDFAD_j = distance from deprivation of the household j

n = number of asset i in the household

s = normalized value of asset i

The focus of the IDFAD on the lower ends of the asset hierarchy has implications for our exploration of poverty. By restricting itself to the assets of the relatively poorer sections of the population, the index is much more sensitive to differences among the relatively less well off. Village studies going back to that of MN Srinivas (1976), or even earlier, point to the very substantial inequalities in everyday village life. These inequalities extend to the poorer sections of village society, as in the differences in tangible assets between Dalits and the poorer non-Dalit caste groups. The IDFAD is designed to be sensitive to these differences between the households facing varying degrees of vulnerability, as reflected in their distance from absolute deprivation.

State Support as an Offset to Poverty

The IDFAD provides us an overall indicator of poverty, including what has been accessed in the market as well as what has been provided by the family, the society and the state. It defines absolute deprivation as the condition of a household not having even the most basic assets. There can be little doubt that such a household should be considered poor, and that it faces absolute deprivation. The difficulty would arise when we seek to use the indicator to address policy concerns. It should be quite clear that this point of absolute deprivation cannot be the poverty line. It would be much too harsh to insist that a house must come right down to a point where it cannot afford any asset at all before it can be considered to be deserving state support, such as a Below the Poverty Line ration card. The poverty line would necessarily have to be above the point of absolute deprivation. The exact point where the poverty line should be has been subject to considerable, and sometimes acrimonious, academic debate. In practice, however, the two lines have evolved through a political process. This can be seen in the distribution of ration cards. The Antyodaya ration card is expected to be given to households that would be quite close to facing absolute deprivation. The Below Poverty Line (BPL) cards tend to cover a much larger section of the population.

The fact that BPL cards in the states typically cover a much larger section than those who are seen to be below the various poverty lines, can be responded to in quite different ways. In a purely technical sense, it can be argued that some of those above the official poverty line are getting the benefit of BPL cards, thereby generating a leakage in government subsidies. In contrast, a wider, human development oriented, view could argue that economic efficiency and growth are not an end in itself. It is only a means to ensuring that basic human requirements are met. Something as basic as food grain should then necessarily be available at affordable prices to anyone who wants it. The liberal cut off for the BPL cards, in practice, is quite consistent with the latter approach.

This is not to suggest that the liberal upper limit for the receipt of government subsidies is not prone to misuse. The cornering of state support by the better off could result in the poor being crowded out of the limited facilities that are available. The issue then is not where the line for the eligibility of state support is drawn, but on whether the extent of state support increases as we move to households that are closer to absolute deprivation. That is, there exists a negative relationship between the IDFAD and government support. In order to see whether this is the case, we need a measure of the extent of state support in India through the PDS and the provision of health services.

The extent of the benefit of the PDS is determined by the type of ration card the household holds. While the specific details may vary across India, there are three types of ration cards usually in existence in a region: ration cards for households above the poverty line (APL), ration cards for households below the poverty line (BPL) and cards for the extremely poor (Antyodaya). This can be taken to reflect four levels of support from the PDS: households that do not have a ration card and hence do not receive any support, households that have an APL card and hence receive some support, households that have a BPL card and hence receive greater support and households that have an Antyodaya card and hence receive maximum support. In an index ranging from 0 to 1, this study has taken households without a card to have a score of 0, those with an APL card to have a score of 0.5, those with a BPL card to have a score of 0.75 and those with an Antyodaya card to have a score of 1. The index thus provides a score of the extent of state support received through the PDS.

When measuring the access to health services, it is sometimes convenient to evaluate this access in geographical terms, that is, whether the public health facility is available in the vicinity of the village. But the constraints faced by the poor in accessing public health facilities are not from distance alone. It is not entirely unknown for doctors in some public health facilities in some parts of the country to charge an illegal fee for their services. There are also cases of doctors not being regular in attending to their duties in state-owned health facilities. It is thus more prudent to focus, instead, on the actual practice rather than the mere existence of a health facility. This can be done by tracking where the head of the household went when he or she had last needed medical help. Those who did not use a public health facility could be given a score of 0. This would include those who did not receive any medical help at all as well as those who used non-state medical services. Those who used a public health facility could be given a score, depending on the highest level of the governmental health facility they used. Those who had used the services of an ASHA can be given a score of 0.5. Those who had used the PHC (whether or not they had also used the services of an ASHA) could be given a score of 0.75. And those who had used the services of a community or government hospital (whether or not they had also used the services of an ASHA or the PHC) could be given a score of 1.

An overall index of tangible state support for a household can then be taken to be the average of the scores of the household in the index of state support through the PDS and the index of state support in health services. The overall index of tangible state support for the household would then range between 0 and 1.

The relationship between the overall index of tangible state support and the IDFAD would provide an indicator of the effectiveness of the support. When the relationship between IDFAD and the index of tangible state support is negative, it would mean the closer the household is to absolute deprivation, the greater the state support. This is as it should be. When the relationship between IDFAD and the index of tangible state support is positive, it would mean there is elite capture of state interventions against poverty.

Conclusion

The estimates of poverty in India that are based entirely on consumption expenditure have severe limitations, that go beyond the accuracy of the data. They focus entirely on the money spent in the market, though the actual experience of poverty by the household would also be influenced by the extent of non-market state support it receives. It ignores the ecological and social pressures on the expenditure of households, as well as the effects of variations across time. It also does not distinguish between poor households that are borrowing to consume and those that are not. An asset-based index avoids these limitations. By focusing on whether a household can save enough to buy an asset, it provides a better indicator of the distance of a household from absolute deprivation. The IDFAD and its relationship to other measures of state support can also be used as a pointer to the effectiveness of anti-poverty strategies.