Poverty: First principles

8.2.1. Conceptual issues At the heart of all discourses on poverty is the notion of a poverty line: a critical threshold of income,

consumption, or, more generally, access to goods and services below which individuals are declared to be poor. The poverty line, then, represents a minimum level of “acceptable” economic participation in a given society at a given point in time. For instance, we could collect data on minimum nutrient levels that make up an adequate diet, on the prices of foodstuffs that contain such nutrients, and on the costs of shelter and clothing, and then add up the consumption expenditures needed to obtain these basic requirements to obtain an estimate of the poverty line for a particular society. We could use the prevailing legally decreed minimum wage in a country as an estimate for the poverty line of that country. Alternatively, we could fix some other norm, say, 60% of the mean income of a country, to arrive at an estimate of its poverty line.

168

Nutrition-based poverty lines are not uncommon. The poverty line used in the United States is based on Orshansky’s [1963, 1965] estimates, which scale by three a minimum-budget estimate for food requirements (the scaling proxies for other requirements such as rent and clothing). Indian poverty lines have traditionally been drawn by using estimates of expenditure necessary to guarantee a minimum consumption of calories. Of course, such poverty lines (and probably all poverty lines) should be approached with some caution and scepticism: the poorer the country, the better the nutrition- based approximation. Issues of scaling become more problematic as the average standard of living rises.

The following subsections explain some of the fundamental concerns that surround poverty measurement.

Overall expenditure or item-by-item consumption? Should we declare a person to be poor when her actual, observed consumption basket falls below certain prespecified thresholds or when her expenditure (or overall income) falls below the minimum required to obtain these consumption standards? Certainly, we could conjure up examples where the two approaches yield different results; for instance, what are we to make of the wealthy ascetic who starves himself on an ongoing basis? At a more serious level, nutrition levels may not unambiguously rise with income.2 For instance, canned foods may become quite popular at certain levels of income, even though their nutritive value is questionable. Thus, even through elasticities may be high with respect to changes in income, nutrient elasticities may not be correspondingly high. Income represents the capacity to consume, not consumption itself. Nevertheless, income- or (aggregate) expenditure-based poverty lines are far easier to use, given the scarcity of available data.

Absolute or relative? Clearly, there is something absolute about the notion of poverty. Regardless of the society we live in, people need adequate levels of food, clothing, and shelter. Whereas it is certainly the case that there are variations in what might be considered “adequate” (shelter, in particular, might be subject to varying society-specific interpretations), nobody would deny the biological imperative of nutrition, for instance, or the near-universal norms of adequate clothing. At the same time, it is unclear that the phrase “acceptable levels of participation in society” can be given absolute meaning, independent of the contours of the society under consideration. In some societies, the ownership of a television may be deemed socially necessary for living a “full” life; in others it is not. Likewise, minimal standards of leisure, access to scientific education, ownership of private means of transportation, and so on, are all concerns that must be evaluated relative to the prevailing socioeconomic standards. These considerations quite naturally give rise to the need for poverty lines that share certain common components, but vary (perhaps widely) from country to country.

Note carefully that although poverty lines should (and do) incorporate relative notions of what constitutes “necessity” or “basic needs,” we must still think of them as fulfilling some absolute notion of the ability to function in a society. The previous paragraph chooses our examples carefully to make this point.3 For instance, it would be foolish to define poverty by, say, the percentage of the population earning less than half the average income of society. Such a measure confuses poverty with inequality. For instance, the measure would remain completely unchanged if all incomes were scaled down by the same proportion, plunging half the population into famine!

Temporary or chronic? As we will see, people who live in (or close to) a state of poverty, however that state is measured, often experience significant fluctuations in their income and consumption. This is especially true for the poor or near-poor in developing countries, where a large fraction of the population may depend on a quirky, weather-dependent agriculture. Expressed as fractions of their average earned income, these fluctuations are large. As Morduch [1994] pointed out, notions of “structural” or chronic poverty must therefore be complemented by a study of “temporary poverty.” The latter occurs when, because of bad economic shocks (such as poor rainfall or low prices for one’s production), individuals temporarily enter a poverty sample. The distinction is not just for the sake of a distinction: the policies required to combat temporary as opposed to chronic poverty may be very different.

The temporary versus chronic distinction is closely related to Friedman’s [1957] famous distinction between temporary and permanent income. Income in a given year may be far from capturing the smoothed or “permanent” stream of consumption that an individual or household enjoys over time. For this reason, household or individual

169

expenditures are often thought of as a more reliable way to assess chronic poverty.

Households or individuals? Often household-level data on expenditure and income are all that is available. It is tempting, then, to simply express household consumption as individual averages (so that household size can be accounted for), and then apply one’s favorite measure of poverty. However, this neglects an exceedingly important issue: that the allocation of expenditures within the house-hold are often significantly skewed. Among the potential victims are females and the elderly. There is some evidence that such discrimination grows sharper with the overall level of destitution of the household. Macroestimates of poverty should therefore be complemented by “microstudies” that study intrahousehold allocation. We will study some examples in the subsequent text.

Neglecting altogether the problems of distribution, a second set of concerns arises from the fact that larger households typically have more children. Some correction for the presence of children is desirable, because they consume somewhat less than adults. The construction of adult equivalence scales—conversion factors that express the consumption of children as a fraction of a representative adult—would get around this problem.4

Finally, there are fixed costs in setting up and running a household. Smaller households cannot spread these fixed costs over several household members. They are therefore at a disadvantage. We return to this and related issues later.

Why a poverty line, anyway? It is possible to argue that a fixed notion of the poverty line is untenable. In part this is because of some issues raised earlier; for example, the relativity of poverty or its fluctuating nature. Even if we stick to chronic, nutrition- based measures of poverty, we still are unable to find some magic level of nutrition below which people abruptly go up in little puffs of smoke (in which case there would probably be no poverty to speak of, anyway). As we shall see later in this chapter, undernutrition is not the same as immediate and obvious disaster, and therefore it is more insidious. The world can indefinitely carry a stock of undernourished people, living and breeding under impaired circumstances. Although more will be said presently on such issues, it is important to realize that poverty lines are always approximations to a threshold that is truly fuzzy, more because the effects of sustained deprivation are often felt at a later point in time. There is really little to be done about this criticism except to realize that quantitative estimates of poverty lines are not to be memorized all the way down to the third decimal place and that they are basically (important) pointers to a deeper and less quantifiable concept.

8.2.2. Poverty measures With the preceding qualifications in mind, then, we will consider a poverty line to be an expenditure threshold

that is regarded as minimally necessary for “adequate” participation in economic life. People below this threshold will be said to be poor.

A little notation will be useful. As in Chapter 6, y denotes income (or expenditure) and subscripts i, j, . . . , refer to individuals. Let’s denote by p the poverty line5 and by m the mean income of the economy.

One natural measure that comes to mind is simply to count the number of people below the poverty line. We might be interested in the numbers per se or in the relative incidence of the poor. In the latter case, divide by the total population of the country or region under consideration. The first measure is known as the head count, and the latter as the head-count ratio, which is just head count as a fraction of population. In part because they don’t place great strains on available data, these measures are widely used. In our notation, the head count (HC) is given by the number of individuals i such that yi < p, whereas the head-count ratio (HCR) is just

where n is the total population. An obvious problem with the head-count ratio is that it fails to capture the extent to which individual income (or

expenditure) falls below the poverty line. This is related, of course, to observation 5 (Why a poverty line, anyway?) 170

in the previous section that poverty is not a “zero–one” concept. People further below the poverty line are “poorer” than people closer to it, and the head count is insensitive to this observation. However, matters are worse than plain insensitivity: use of the head count can lead to problematic policy decisions, as the following example suggests.

Example 1: You are a planner in Ping, a poor land, where the poverty line is set at 1000 pah a year. It turns out that in Ping there are two equal-sized groups below the poverty line. One group consists of 100 individuals: they have equal earnings of 500 pah a year each. The second group also has 100 people: they earn 900 pah a year each. Of course, there are also people who are above the poverty line. You have been allocated a budget of 20,000 pah a year. You must allocate this budget among the 200 poor people.

(i) Suppose you were to forget about the poverty line. Who would you give the money to?

(ii) Now suppose that you are firmly told by the President of Ping to use this money to minimize, as far as possible, the head count. Who would you give the money to?

The point of the example is very simple. The use of the head count as a measure of poverty systematically biases policy in favor of individuals who are very close to the poverty line. Statistically, these people offer the biggest bang for the buck, because they are most easily taken above the poverty line. Yet of all the poor, they are relatively in the least need of help. A benevolent government that is perfectly secure and without fear of losing the next elections may ignore the problem and act in the best interests of the people, but most governments, like most people, are more interested in maximizing the observable and seemingly objective measures of their success.

One way to partially offset this bias, and more fundamentally take account of the extent of poverty, is to use a measure of the average income shortfall from the poverty line. An example is the poverty gap ratio, defined as the ratio of the average of income (or extra consumption) needed to get all poor people to the poverty line, divided by the mean income (or consumption) of the society. The reason for dividing by the average for society as a whole is that this gives us an idea of how large the gap is relative to resources that potentially may be used to close the gap. In this sense, the poverty gap ratio is not really a measure of poverty itself, but a measure of resources required to eradicate it.

In terms of our notation, the poverty gap ratio (PGR) is given by

where m, you will recall, is mean income. Dividing by average economywide income might give a misleading impression of poverty in highly unequal (but

overall wealthy) societies with a large number of poor people. The poverty gap ratio in such societies may look pretty small, even though the plight of the poor is made no less acute by this maneuver. Therefore, a close relative of this measure, called the income gap ratio, is often used. This is exactly the same measure of total shortfall of the poor from the poverty line, except that we divide the shortfall by the total income required to bring all the poor people to the poverty line. This places a slightly different perspective on things. It captures more directly the acuteness of poverty, because it measures it relative to the total income needed to make that poverty go away.6 Thus the income gap ratio (IGR) is given by the formula

where we recall that HC is just the number (head count) of the poor. The PGR or the IGR is not susceptible to the same kind of policy distortion as the head count, as the following

example shows.

Example 2: Return to the problem of Example 1. Now imagine that you are told to minimize (as far as possible) the PGR or the IGR. Does the way you now spend your money necessarily contrast with the intuitive reactions you noted in part (i) of Example 1?

It should be clear from the discussion that the PGR or IGR avoids the “bang for the buck” problem by deliberately neglecting numbers or fractions of people that are below the poverty line. In a sense, PGR and IGR only capture the “per capita intensity” of poverty. The head count (or HCR), whatever its other failings, does not

171

suffer from this problem. For this reason, it is a good idea to use measures of each type jointly, where possible, to evaluate the extent of poverty.

Finally, we note that both the head count and the poverty gap class of measures share an additional drawback relating to the fact that both these measures ignore the important issue of relative deprivation among the poor.7 Relative deprivation is just another phrase for inequality among the poor. The new phrase is used to capture the fact that we are concerned only with the inequality among the deprived, or poor. The main concern is captured by the following example.

Example 3: Return to Example 1, where, as you will recall, there are 200 people below the poverty line; half of them have an income of 500 pah and the rest have an income of 900 pah.

(i) Suppose that each person who earns 500 pah gave 50 pah to each person who earns 900 pah. The new income levels are then 450 and 950 pah. What do you think would happen to the intensity of poverty in this new situation relative to the old? Now compute the HCR and PGR (or IGR) in both situations. Compare what the measures say with what you feel intuitively.

(ii) To make the point even more starkly, transfer 110 pah each (instead of 50 pah) between the same groups and redo the exercise.

Even if we were to take the head count and the gap-ratio measures together, there are other aspects of poverty that may be left out. This observation leads to more sophisticated measures of poverty that have been proposed by economists such as Sen [1976] and Foster, Greer, and Thorbecke [1984]. With better data, these more demanding measures can be easily applied. The Appendix to this chapter contains a discussion of the Foster–Greer–Thorbecke index.