© 2003 by default! slide 1 poverty lines celia m. reyes introduction to poverty analysis nai,...

© 2003 By Default!© 2003 By Default!Slide Slide 11

Poverty Lines Poverty Lines

Celia M. ReyesCelia M. Reyes

Introduction to Poverty AnalysisIntroduction to Poverty AnalysisNAI, Beijing, ChinaNAI, Beijing, China

Nov. 1-8, 2005Nov. 1-8, 2005


Step 2: How to define a poverty lineStep 2: How to define a poverty line

After choosing a measure of household well-After choosing a measure of household well-being, say consumption expenditure, the being, say consumption expenditure, the next step is to choose a poverty line. next step is to choose a poverty line. Households whose consumption Households whose consumption expenditure fall below the poverty line are expenditure fall below the poverty line are considered poor.considered poor.

The poverty line is obtained by specifying a The poverty line is obtained by specifying a consumption bundle considered adequate consumption bundle considered adequate for basic consumption needs and then for basic consumption needs and then estimating the costs of these basic needs.estimating the costs of these basic needs.


How to define a poverty lineHow to define a poverty line

The poverty line is conceptualised as a The poverty line is conceptualised as a minimum standard required by an individual minimum standard required by an individual to fulfil his or her basic food and non-food to fulfil his or her basic food and non-food needs.needs.

The poverty line defines the level of The poverty line defines the level of consumption (or income) needed for a consumption (or income) needed for a household to escape poverty.household to escape poverty.



Well-being follows a continuum, and given Well-being follows a continuum, and given how arbitrary the choice of the poverty line how arbitrary the choice of the poverty line is, the notion of a turning point is not very is, the notion of a turning point is not very compelling.compelling.

It may make sense to define more than one It may make sense to define more than one poverty line. One poverty line may mark poverty line. One poverty line may mark households that are poor and another level households that are poor and another level could indicate those that are extremely poor. could indicate those that are extremely poor.



One could also construct a “food poverty One could also construct a “food poverty line” which is based on some notion of line” which is based on some notion of minimum amount of money that a household minimum amount of money that a household needs to purchase some basic-needs food needs to purchase some basic-needs food bundle.bundle.

If the cost of the basic needs is estimated, If the cost of the basic needs is estimated, then the food poverty line added to the non-then the food poverty line added to the non-food needs will equal the overall poverty food needs will equal the overall poverty line.line.



The poverty line for a household, zThe poverty line for a household, z ii, may be , may be defined as the minimum defined as the minimum spending/consumption (or other measure) spending/consumption (or other measure) needed to achieve at least the minimum needed to achieve at least the minimum utility level, uutility level, uzz, so:, so:

zzii = e(p,x,u = e(p,x,uzz))


How to define a poverty lineHow to define a poverty lineTwo approaches in constructing poverty Two approaches in constructing poverty

lines:lines:

1.1. Compute a poverty line for each Compute a poverty line for each household, adjusting it from household household, adjusting it from household to household to take into account to household to take into account differences in the prices they face and differences in the prices they face and their demographic composition. their demographic composition. Example - ThailandExample - Thailand

2.2. Construct one per capita poverty line for Construct one per capita poverty line for all individuals, but to adjust per capita yall individuals, but to adjust per capita y ii for differences in prices and household for differences in prices and household composition. composition.

Example - PhilippinesExample - Philippines



The approach taken for Cambodia in The approach taken for Cambodia in 1999 somewhere between these two 1999 somewhere between these two extremes. extremes.

Separate poverty lines were constructed Separate poverty lines were constructed for each of the three major regions for each of the three major regions based on the prices prevailing in those based on the prices prevailing in those areas. areas.

These poverty lines are shown in Table These poverty lines are shown in Table 3.1.3.1.


Table 3.1Table 3.1Summary of Cambodia poverty linesSummary of Cambodia poverty lines

Food Poverty Food PovertyPhnom Penh 1,185 1,578 1,737 2,470 Other Urban 995 1,264 1,583 2,093 Rural 881 1,117 1,379 1,777 Note: All values are in Riels per person per daySource: Prescott and Pradhan (1997); Gibson (1999)

1999 CSES 1993/ 94 SESC



Why do poverty lines change over time:Why do poverty lines change over time:

– As prices change, the poverty line will As prices change, the poverty line will changechange

–We could change the way we compute the We could change the way we compute the poverty line (i.e., real poverty threshold is poverty line (i.e., real poverty threshold is revised).revised).


Table 3.2 Table 3.2 Average poverty line of ThailandAverage poverty line of Thailand

Year Poverty Line1988 4731990 5221992 6001994 6361996 7371998 8781999 886


Annual Per Capita Food ThresholdsAnnual Per Capita Food Thresholdsby Province, 2000by Province, 2000

All Areas Urban Rural

PHILIPPINES 7,810 8,323 7,688

NCR 9,561 9,5611st District 9,561 9,5612nd District 9,561 9,5613rd District 9,561 9,5614th District 9,561 9,561

Region V 8,185 9,202 7,979Albay 8,057 9,508 7,741Camarines Norte 7,906 9,006 7,626Camarines Sur 8,123 9,168 7,862Catanduanes 8,132 9,211 7,977Masbate 8,491 9,139 8,414Sorsogon 8,189 9,245 7,955

2000Province


Annual Per Capita Poverty Thresholds Annual Per Capita Poverty Thresholds by Province, 2000by Province, 2000

All Areas Urban Rural

PHILIPPINES 11,620 12,933 11,130

NCR 15,678 15,6781st District 16,190 16,1902nd District 15,710 15,7103rd District 14,972 14,9724th District 16,345 16,345

Region V 11,517 13,742 10,955Albay 11,858 14,970 11,115Camarines Norte 11,481 13,785 10,817Camarines Sur 11,436 13,529 10,792Catanduanes 11,700 13,854 11,226Masbate 11,488 13,217 11,214Sorsogon 11,138 12,879 10,653

2000Province


Relative PovertyRelative Poverty

Sometimes we are interested in focusing on Sometimes we are interested in focusing on the poorest segment (e.g. a fifth, or two-the poorest segment (e.g. a fifth, or two-fifths) of the population.fifths) of the population.

In comparing poverty lines across countries, In comparing poverty lines across countries, rich countries have higher poverty lines than rich countries have higher poverty lines than do poor countries as shown in Figure 3.1. do poor countries as shown in Figure 3.1.

3.1 How to define poverty lines3.1 How to define poverty lines



This explains why, for instance, the official This explains why, for instance, the official poverty rate in the early 1990s was close to poverty rate in the early 1990s was close to 15% in the United States and also close to 15% in the United States and also close to 15% in (much poorer) Indonesia. Many of 15% in (much poorer) Indonesia. Many of those counted as poor in the U.S. would be those counted as poor in the U.S. would be considered to be comfortably off by considered to be comfortably off by Indonesian standards. The poverty lines Indonesian standards. The poverty lines across countries are, to some extent, across countries are, to some extent, relative.relative.



Figure 3.5Figure 3.5Poverty lines across countriesPoverty lines across countries

Note: Each point is one country. Both poverty line and mean consumption are measured at purchasing power parity.Source: Ravallion, Datt and van de Walle (1991)



The most common practice in setting The most common practice in setting relative poverty lines is to use some relative poverty lines is to use some proportion of the arithmetic mean or median proportion of the arithmetic mean or median of the distribution of consumption or income of the distribution of consumption or income as the poverty line; for example, many as the poverty line; for example, many studies have used a poverty line which is set studies have used a poverty line which is set at about 50% of the national median:at about 50% of the national median:

SESC 1993/94 Mean: 1833 --> poverty line = 916SESC 1993/94 Mean: 1833 --> poverty line = 916

too low?too low?




In comparing poverty lines across countries, rich In comparing poverty lines across countries, rich countries have higher poverty lines than do poor countries have higher poverty lines than do poor countries. countries.

This explains why, for instance, the official poverty This explains why, for instance, the official poverty rate in the early 1990s was close to 15% in the rate in the early 1990s was close to 15% in the United States and also close to 15% in (much United States and also close to 15% in (much poorer) Indonesia. Many of those counted as poor poorer) Indonesia. Many of those counted as poor in the U.S. would be considered to be comfortably in the U.S. would be considered to be comfortably off by Indonesian standards. The poverty lines off by Indonesian standards. The poverty lines across countries are, to some extent, relative.across countries are, to some extent, relative.




As countries become better off, they have a As countries become better off, they have a tendency to revise the poverty line upwards tendency to revise the poverty line upwards – with the notable exception of the United – with the notable exception of the United States, where the line has (in principle) States, where the line has (in principle) remained unchanged for almost four remained unchanged for almost four decades.decades.




To the extent that it is to identify and target To the extent that it is to identify and target today’s poor, then a relative poverty line is today’s poor, then a relative poverty line is appropriate, and needs to be tailored to the appropriate, and needs to be tailored to the overall level of development of the country. overall level of development of the country. For instance, the European Union typically For instance, the European Union typically defines the poor as those whose per capita defines the poor as those whose per capita incomes fall below 50% of the median. As incomes fall below 50% of the median. As median incomes rise, so does the poverty median incomes rise, so does the poverty line.line.




Based on a sample of 36 countries, Ravallion, Datt Based on a sample of 36 countries, Ravallion, Datt and van de Walle estimated the following and van de Walle estimated the following relationship:relationship:

t=5.1t=5.1 t=-3.6 t=5.1 R2=0.89 t=-3.6 t=5.1 R2=0.89

They found that at the mean value of per capita They found that at the mean value of per capita consumption (which they measured in purchasing consumption (which they measured in purchasing power parity terms), the elasticity of the poverty power parity terms), the elasticity of the poverty line with respect to consumption per capita was line with respect to consumption per capita was 0.71. 0.71.

ii vcapCcapCz 2)]/[ln(228.0)/ln(773.1704.6ln




This means that if per capita consumptions This means that if per capita consumptions were to rise 10%, then the poverty line would were to rise 10%, then the poverty line would rise 7.1% on average. But the non-linear rise 7.1% on average. But the non-linear relationship implies that the elasticity of the relationship implies that the elasticity of the poverty line with respect to consumption per poverty line with respect to consumption per capita was close to 0 in low-income capita was close to 0 in low-income countries, and was almost 1 in high-income countries, and was almost 1 in high-income countries.countries.




To the extent that it is to identify and target To the extent that it is to identify and target today’s poor, then a relative poverty line is today’s poor, then a relative poverty line is appropriate, and needs to be tailored to the appropriate, and needs to be tailored to the overall level of development of the country. overall level of development of the country. For instance, a $1 per day poverty line might For instance, a $1 per day poverty line might be useful in Vietnam, where 27% of the be useful in Vietnam, where 27% of the population would be considered poor by this population would be considered poor by this standard in 1998 (Haughton 2000), but would standard in 1998 (Haughton 2000), but would be of little relevance in the United States be of little relevance in the United States where almost nobody would be poor by this where almost nobody would be poor by this standard.standard.



Absolute PovertyAbsolute Poverty

An absolute poverty line is one that is fixed An absolute poverty line is one that is fixed in terms of standard of living. A relative in terms of standard of living. A relative poverty line, by contrast, varies depending poverty line, by contrast, varies depending on the standard of living.on the standard of living.

An absolute poverty line is essential if one is An absolute poverty line is essential if one is trying to judge the effect of anti-poverty trying to judge the effect of anti-poverty policies over time, or to estimate the impact policies over time, or to estimate the impact of a project (e.g. microcredit) on poverty.of a project (e.g. microcredit) on poverty.




Legitimate comparisons of poverty rates Legitimate comparisons of poverty rates between one country and another can only between one country and another can only be made if the same absolute poverty line is be made if the same absolute poverty line is used in both countries.used in both countries.

Examples:Examples:– an estimated 1,200 million people an estimated 1,200 million people

worldwide live on less than “$1 per day”worldwide live on less than “$1 per day”– over 2 billion people worldwide live on over 2 billion people worldwide live on

less than $2 per day.less than $2 per day.




The $1 per day standard

Cross-country comparisons of poverty rates are notoriously difficult, but the World Bank has tried to get around this problem by computing the proportion of the population in different countries living on less than $1 per capita per day (in purchasing power parity terms, and in 1985 US dollars). The numbers shown below suggest that the poverty rate in Vietnam (computed by Haughton 2000) compares favorably with that of India, but lags behind (more affluent) China and Indonesia.

The $1 per day standard

Cross-country comparisons of poverty rates are notoriously difficult, but the World Bank has tried to get around this problem by computing the proportion of the population in different countries living on less than $1 per capita per day (in purchasing power parity terms, and in 1985 US dollars). The numbers shown below suggest that the poverty rate in Vietnam (computed by Haughton 2000) compares favorably with that of India, but lags behind (more affluent) China and Indonesia.




Vietnam 45 1993 Indonesia 8 1996Vietnam 27 1998 Nigeria 31 1992- 93China 22 1995 Philippines 27 1994India 47 1994

Sources: World Bank. 1999c. Entering the 21 st Century: WorldDevelopment Report 1999/2000 . Washington DC. Haughton 2000.

% in poverty % in poverty



Decide the standard of livingDecide the standard of living

There is an important conceptual problem There is an important conceptual problem that comes up when working with absolute that comes up when working with absolute poverty lines, which arises from the issue of poverty lines, which arises from the issue of what is meant by “the standard of living”.what is meant by “the standard of living”.

In practice almost all absolute poverty lines In practice almost all absolute poverty lines are set in terms of the cost of buying a are set in terms of the cost of buying a basket of goods (the “commodity-based basket of goods (the “commodity-based poverty line”, which we denote by z). If we poverty line”, which we denote by z). If we assume that u=f(y)assume that u=f(y)

3.2 Issues in choosing an absolute poverty line3.2 Issues in choosing an absolute poverty line



If we assume that u=f(y), which says that utility or If we assume that u=f(y), which says that utility or “standard of living” (u) depends on income or “standard of living” (u) depends on income or expenditure (y), thenexpenditure (y), then

y=fy=f-1-1(u)(u)This says that for any given level of utility, there is This says that for any given level of utility, there is some income (or expenditure) level that is needed some income (or expenditure) level that is needed to achieve it. If to achieve it. If uuzz is the utility that just suffices to is the utility that just suffices to avoid being pooravoid being poor, then, then

z=fz=f-1-1(u(uzz))In other words, given a poverty line that is absolute In other words, given a poverty line that is absolute in the space of welfare (i.e. gives uin the space of welfare (i.e. gives uzz) there is a ) there is a corresponding absolute commodity-based poverty corresponding absolute commodity-based poverty line.line.3.2 Issues in choosing an absolute poverty line3.2 Issues in choosing an absolute poverty line



But suppose we make another equally But suppose we make another equally plausible assumption, which is that utilities plausible assumption, which is that utilities are interdependent. are interdependent.

Then, one’s well-being may depend not just Then, one’s well-being may depend not just on what one consume, but on how one’s on what one consume, but on how one’s consumption stacks up against that of the consumption stacks up against that of the rest of society. A household of four with an rest of society. A household of four with an income of $12,000 per year would not be income of $12,000 per year would not be considered poor in Indonesia, but when this considered poor in Indonesia, but when this household compares its position with household compares its position with average incomes in the US, it may feel very average incomes in the US, it may feel very poor. poor.




We may capture this idea by assumingWe may capture this idea by assuming

where is the mean income in the society. where is the mean income in the society. In this caseIn this case

and so, making the standard assumption of and so, making the standard assumption of invertibility,invertibility,


y

),,(y

yygu

),(y

zzguz

).,(1zuygz



This means that for a poverty line to be This means that for a poverty line to be absolute in the space of welfare (i.e. to yield absolute in the space of welfare (i.e. to yield uz), the commodity-based poverty line (i.e. z) uz), the commodity-based poverty line (i.e. z) may have to rise as rises. The commodity-may have to rise as rises. The commodity-based poverty line would then look more like based poverty line would then look more like a relative poverty line.a relative poverty line.



Decide utility of the poverty line uDecide utility of the poverty line uz z and and commodity value of the poverty line Z)commodity value of the poverty line Z)

Even if we assume that the commodity-Even if we assume that the commodity-based poverty line remains constant, we are based poverty line remains constant, we are still left with two problems:still left with two problems:

– The Referencing problem. What is the The Referencing problem. What is the appropriate value of uappropriate value of uzz – i.e. the utility of – i.e. the utility of the poverty line? The choice is arbitrary, the poverty line? The choice is arbitrary, but “a degree of consensus about the but “a degree of consensus about the choice of the reference utility level in a choice of the reference utility level in a specific society may well be crucial to specific society may well be crucial to mobilizing resources for fighting poverty” mobilizing resources for fighting poverty” (Ravallion)(Ravallion)




– The Identification problem. Given uThe Identification problem. Given uzz, what , what is the correct value of z – i.e. of the is the correct value of z – i.e. of the commodity value of the poverty line. This commodity value of the poverty line. This problem arises both because the size and problem arises both because the size and demographic composition of households demographic composition of households vary and also because “the view that we vary and also because “the view that we can measure welfare by looking solely at can measure welfare by looking solely at demand behavior is untenable” (Ravallion)demand behavior is untenable” (Ravallion)




The implication is that external information The implication is that external information and judgments will be required to answer and judgments will be required to answer the referencing and identification problem, the referencing and identification problem, and hence to determine the absolute poverty and hence to determine the absolute poverty line in practice.line in practice.

Table 3.3 presents absolute and relative Table 3.3 presents absolute and relative poverty rates for different regions in the poverty rates for different regions in the world. world.



Table 3.3Table 3.3Absolute and Relative Poverty RatesAbsolute and Relative Poverty Rates

Share of population living on less than

$1 per day ( in 1998)

Share of the population living on less than one-third of average national

consumption for 1993 ( in 1998)

East Asia and Pacific 15.3 19.6 East Asia and Pacific excluding China 11.3 24.6 Europe and Central Asia 5.1 25.6 Latin America and the Caribbean 15.6 51.4 Middle East and North Africa 1.9 10.8 South Asia 40.0 40.2 Sub- Saharan Africa 46.3 50.5 Total 24.0 32.1 Total excluding China 26.2 37.0 Source: World Bank (2000)



We might note that an absolute poverty line is best We might note that an absolute poverty line is best thought of as one which is fixed in terms of living thought of as one which is fixed in terms of living standards, and fixed over the entire domain of the standards, and fixed over the entire domain of the poverty comparison (Ravallion).poverty comparison (Ravallion).

Absolute poverty will deem two persons at the Absolute poverty will deem two persons at the same standard of living to both be either “poor” or same standard of living to both be either “poor” or “not poor” irrespective of the time or place being “not poor” irrespective of the time or place being considered, or with or without some policy change, considered, or with or without some policy change, within the relevant domain.within the relevant domain.




The poverty comparison is then “consistent” in the The poverty comparison is then “consistent” in the specific sense that individuals who are identical in specific sense that individuals who are identical in all relevant respects are treated identically. all relevant respects are treated identically. However, depending on the purpose of the However, depending on the purpose of the comparison, the relevant domain may vary.comparison, the relevant domain may vary.

For example, a global comparison of absolute For example, a global comparison of absolute consumption poverty may entail using a poverty consumption poverty may entail using a poverty line (e.g. $1 consumption per capita per day) that is line (e.g. $1 consumption per capita per day) that is low by the standards of rich countries. low by the standards of rich countries.




If, however, one is trying to form a poverty If, however, one is trying to form a poverty profile for one country only, the choice of an profile for one country only, the choice of an absolute poverty line should be appropriate absolute poverty line should be appropriate to that country (e.g. a poverty line of $1 per to that country (e.g. a poverty line of $1 per day might be appropriate in Vietnam, and day might be appropriate in Vietnam, and $10 per day might be suitable in the United $10 per day might be suitable in the United States). Judgments of what constitutes a States). Judgments of what constitutes a reasonable absolute poverty line must first reasonable absolute poverty line must first specify the domain of comparisons, and specify the domain of comparisons, and recognize that the answer may change if the recognize that the answer may change if the domain changes.domain changes.



Solution A: objective poverty linesSolution A: objective poverty lines

The poverty line should be set at a level that The poverty line should be set at a level that enables individuals to achieve certain capabilities, enables individuals to achieve certain capabilities, including a healthy and active lif and full including a healthy and active lif and full participation in society. participation in society.

In practice, this almost certainly would imply that In practice, this almost certainly would imply that the commodity-based poverty line would rise as a the commodity-based poverty line would rise as a country becomes more affluent, because the country becomes more affluent, because the minimum resources needed to participate fully in minimum resources needed to participate fully in society probably rises over time.society probably rises over time.

Thus, “an absolute approach in the space of Thus, “an absolute approach in the space of capabilities translates into a relative approach in capabilities translates into a relative approach in the space of commodities.”the space of commodities.”


Solution A: objective poverty linesSolution A: objective poverty lines

A common way of approaching capabilities A common way of approaching capabilities is to begin with nutritional requirements. is to begin with nutritional requirements. Two common ways of making this Two common ways of making this operational:operational:– Cost-of-basic Needs approachCost-of-basic Needs approach– Food energy intake methodFood energy intake method


Food energy Intake methodFood energy Intake method

The goal here is to find the level of The goal here is to find the level of consumption expenditure (or income) that consumption expenditure (or income) that allows the household to obtain enough food allows the household to obtain enough food to meet its energy requirements.to meet its energy requirements.

Consumption will include non-food as well Consumption will include non-food as well as food as items; even underfed household as food as items; even underfed household typically consume some clothing and typically consume some clothing and shelter, which means that at the margin shelter, which means that at the margin these “basic needs” must be as valuable as these “basic needs” must be as valuable as additional food.additional food.

3.3 Solution A: objective poverty lines3.3 Solution A: objective poverty lines


Food energy Intake methodFood energy Intake method

The basic idea is captured in figure 3.2, The basic idea is captured in figure 3.2, which shows a calorie income function; as which shows a calorie income function; as income (or expenditure) rises, food energy income (or expenditure) rises, food energy intake also rises, although typically more intake also rises, although typically more slowly. Given some level of just-adequate slowly. Given some level of just-adequate food energy intake k, one may use this curve food energy intake k, one may use this curve to determine the poverty-line level of to determine the poverty-line level of expenditure, z. expenditure, z.



Figure 3.2 Calorie income functionFigure 3.2 Calorie income function

Calorie income function

Food energy intake (Cals/day)

Income (or expenditure) , y

2100

Calorie income function

Food energy intake (Cals/day)

Income (or expenditure) , y

2100


Food Energy Intake Method Food Energy Intake Method

Formally, the function showsFormally, the function showsk = f(y)k = f(y)

So given monotonicity, So given monotonicity, y = fy = f-1-1(k)(k)

Or given minimum adequate level of calorie Or given minimum adequate level of calorie (k(kminmin), we have), we have

z = fz = f-1-1(k(kminmin))where z is the poverty line.where z is the poverty line.



Food Energy Intake MethodFood Energy Intake Method

First one needs to determine the amount of First one needs to determine the amount of food that is adequate. Vietnam pegs this food that is adequate. Vietnam pegs this level at 2,100 Calories per person per day, in level at 2,100 Calories per person per day, in line with FAO recommendations, but it is line with FAO recommendations, but it is recognized that individuals may need more recognized that individuals may need more or less food that this-clearly the needs of or less food that this-clearly the needs of young children, growing children, growing young children, growing children, growing teenagers, manual workers, pregnant teenagers, manual workers, pregnant women, or sedentary office workers may women, or sedentary office workers may differ quite markedly; physical stature also differ quite markedly; physical stature also plays a role.plays a role.


© 2003 By Default!© 2003 By Default!Slide Slide 4747Table 3.3Table 3.3Per Capita Daily Calorie Intake Used in Per Capita Daily Calorie Intake Used in Poverty Line ConstructionPoverty Line Construction


Vietnam 2,100Indonesia 2,100Philippines 2,000Thailand 1,978China 2,150


Food Energy IntakeFood Energy Intake

For Vietnam using data from the VLSS 93, separate For Vietnam using data from the VLSS 93, separate food expenditure were estimated for urban and food expenditure were estimated for urban and rural areas in each of seven provinces; the cost of rural areas in each of seven provinces; the cost of obtaining 2,100 Calories of food per person per day obtaining 2,100 Calories of food per person per day was then computed, and the associated poverty was then computed, and the associated poverty lines- one for each rural and urban area in each lines- one for each rural and urban area in each province. This give a headcount index of 55% province. This give a headcount index of 55% (Dollar et al. 1995).(Dollar et al. 1995).

However, the food energy intake method has However, the food energy intake method has several problems, and should not be used unless several problems, and should not be used unless the alternatives are infeasible. the alternatives are infeasible.



The urban-rural problemThe urban-rural problem

The problem begins when one recognizes that food The problem begins when one recognizes that food energy – typically shown on the calorie income energy – typically shown on the calorie income function – depends on other factors as well as function – depends on other factors as well as income.income.

The other influences include The other influences include – the tastes of the household (e.g. urban states in the tastes of the household (e.g. urban states in

food may differ from rural tastes)food may differ from rural tastes)– the level of activity of household members; the level of activity of household members; – the relative prices of different foods, and of food the relative prices of different foods, and of food

to non- food items;to non- food items;– and the presence of publicly-provided goods.and the presence of publicly-provided goods.

3.3.1 Food Energy Intake3.3.1 Food Energy Intake



Figure 3.3 shows hypothetical (but plausible) Figure 3.3 shows hypothetical (but plausible) calorie income functions for urban and rural calorie income functions for urban and rural households.households.

Rural households can obtain food more Rural households can obtain food more cheaply, both because food is typically less cheaply, both because food is typically less expensive in the rural areas and also expensive in the rural areas and also because they are more willing to consume because they are more willing to consume foodstuffs that are cheaper per calorie (ex. foodstuffs that are cheaper per calorie (ex. Cassava rather than rice).Cassava rather than rice).




The key finding of Ravallion and Bidani (1994), The key finding of Ravallion and Bidani (1994), based on 1990 data from the SUSENAS household based on 1990 data from the SUSENAS household survey in Indonesia, was that the urban poverty line survey in Indonesia, was that the urban poverty line (20,614 rupiah per person per month) was much (20,614 rupiah per person per month) was much higher than the rural one (13,295 Rp higher than the rural one (13,295 Rp /person/month). This gap far exceeded the /person/month). This gap far exceeded the difference in the cost of living between urban and difference in the cost of living between urban and rural areas.rural areas.

Using these poverty lines, Ravallion and Bidani Using these poverty lines, Ravallion and Bidani (1994) found that poverty in Indonesia appeared to (1994) found that poverty in Indonesia appeared to be higher in the urban than in the rural areas (Table be higher in the urban than in the rural areas (Table 3.4), a completely implausible result.3.4), a completely implausible result.


© 2003 By Default!© 2003 By Default!Slide Slide 5353Table 3.4Table 3.4Poverty Lines in Indonesia using Food Poverty Lines in Indonesia using Food Energy Intake Method, 1990 Energy Intake Method, 1990

Indonesia overall

Urban areas

Rural areas

P0 15.1 16.8 14.3

P1 2.42 3.23 1.06

P2 0.66 0.94 0.53

Source: Ravallion and Bidani 1994.



The point is also illustrated in Figure 3.4, The point is also illustrated in Figure 3.4, which shows the cumulative distribution of which shows the cumulative distribution of consumption per capita, for rural and urban consumption per capita, for rural and urban areas, and marks the poverty lines and areas, and marks the poverty lines and headcount poverty rates.headcount poverty rates.


© 2003 By Default!© 2003 By Default!Slide Slide 5555Figure 3.4Figure 3.4Cumulative Distribution functions for Cumulative Distribution functions for consumption,Indonesia, 1990consumption,Indonesia, 1990


The relative price problemThe relative price problem

When researchers tried to apply the Food When researchers tried to apply the Food Energy Intake approach to data from the Energy Intake approach to data from the Vietnam Living Standards Survey of 1998, Vietnam Living Standards Survey of 1998, the method failed.the method failed.

As with the 1993 data, the idea was to As with the 1993 data, the idea was to compute food expenditure per capita, which compute food expenditure per capita, which would then serve as a poverty line.would then serve as a poverty line.




After undertaking this exercise, researchers After undertaking this exercise, researchers found a higher level of poverty in 1998 than found a higher level of poverty in 1998 than in 1993, an implausible result in an economy in 1993, an implausible result in an economy whose real GDP grew by 9% annually whose real GDP grew by 9% annually between 1993 and 1998, and where there between 1993 and 1998, and where there was a widespread sense that the benefits of was a widespread sense that the benefits of this growth had spread sense that the this growth had spread sense that the benefits of this growth had spread widely.benefits of this growth had spread widely.



What went wrong? Figure 3.5 shows the What went wrong? Figure 3.5 shows the situation.situation.



Relative price problemRelative price problem

The food expenditure function shifted down The food expenditure function shifted down between 1993 and 1998. For a given real income, between 1993 and 1998. For a given real income, households in 1998 would buy less food than in households in 1998 would buy less food than in 1993.1993.

The main reason was that the price of food rose by The main reason was that the price of food rose by 70% between 1993 and 1998, while the price of non-70% between 1993 and 1998, while the price of non-food items rose by only 25%.food items rose by only 25%.

As a result, consumers shifted away from food to As a result, consumers shifted away from food to non-food consumption. non-food consumption.

Poverty line rose drastically from zPoverty line rose drastically from z9393 to z to z98 98

(implausibly large).(implausibly large).


The Cost-of-Basic Needs MethodThe Cost-of-Basic Needs Method

A more satisfactory approach to building up A more satisfactory approach to building up a poverty line, while remaining in the spirit of a poverty line, while remaining in the spirit of trying to ensure that the line covers basic trying to ensure that the line covers basic needs, proceeds as follows:needs, proceeds as follows:

– Stipulate a consumption bundle that is Stipulate a consumption bundle that is deemed to be adequate, with both food deemed to be adequate, with both food and non food components; and and non food components; and

– Estimate the cost of the bundle for each Estimate the cost of the bundle for each subgroup (urban/rural, each region, etc.).subgroup (urban/rural, each region, etc.).




Note that the poverty line is measured in Note that the poverty line is measured in money. We are therefore not insisting that money. We are therefore not insisting that each basic need be met, only that it could be each basic need be met, only that it could be met.met.

The steps to follow are these: The steps to follow are these: – Pick a nutritional requirement for good Pick a nutritional requirement for good

health, such as 2,100 Calories per person health, such as 2,100 Calories per person per day.per day.




Estimate the cost of meeting these food Estimate the cost of meeting these food energy requirement, using a diet that reflects energy requirement, using a diet that reflects the habit of households near the poverty line the habit of households near the poverty line (e.g. those in the lowest, or second-lowest, (e.g. those in the lowest, or second-lowest, quintile of the income distribution; or those quintile of the income distribution; or those consuming between 2000 and 2200 consuming between 2000 and 2200 Calories). This may not be easy if diets vary Calories). This may not be easy if diets vary widely across the country. Call this food widely across the country. Call this food component zcomponent zFF..




Add a non-food component (zAdd a non-food component (zNFNF).). Then the basic poverty line is given by Then the basic poverty line is given by

zzBNBN = z = zFF + z + zNFNF



Box. The US poverty line Box. The US poverty line

In 1963 and 1964, Mollie Orshansky of the In 1963 and 1964, Mollie Orshansky of the Social Security Administration computed the Social Security Administration computed the cost of an ‘adequate’ amount of food intake, cost of an ‘adequate’ amount of food intake, to get to get zzFF. She then multiplied this number by . She then multiplied this number by 3 to get 3 to get zzBNBN. Why? Because at the time, the . Why? Because at the time, the average food share for all consumers in the average food share for all consumers in the United States was 1/3. This line is still used, United States was 1/3. This line is still used, updated regularly for price changes. updated regularly for price changes.

Box. The US poverty line Box. The US poverty line

In 1963 and 1964, Mollie Orshansky of the In 1963 and 1964, Mollie Orshansky of the Social Security Administration computed the Social Security Administration computed the cost of an ‘adequate’ amount of food intake, cost of an ‘adequate’ amount of food intake, to get to get zzFF. She then multiplied this number by . She then multiplied this number by 3 to get 3 to get zzBNBN. Why? Because at the time, the . Why? Because at the time, the average food share for all consumers in the average food share for all consumers in the United States was 1/3. This line is still used, United States was 1/3. This line is still used, updated regularly for price changes. updated regularly for price changes.




To illustrate how this might work, suppose To illustrate how this might work, suppose that we use a food energy threshold of 2100 that we use a food energy threshold of 2100 Calories per day. Let there be three Calories per day. Let there be three foodstuffs: rice, corn and eggs. Table 3.5 foodstuffs: rice, corn and eggs. Table 3.5 shows the expenditure on each item, and the shows the expenditure on each item, and the amount consumed by a household in the amount consumed by a household in the second (from bottom) quintile.second (from bottom) quintile.


© 2003 By Default!© 2003 By Default!Slide Slide 6666Table 3.5Table 3.5Illustration of Construction of Cost of Food Illustration of Construction of Cost of Food Components of PovertyComponents of Poverty

Calories Expenditure per day (pesos)

Calories, Adjusted to give 2,100

Calories

Expenditure, adjusted to cover 2,100

Calories

Rice 1,400 60 1,470 63Corn 400 20 420 21Eggs 200 20 210 21Total 2,000 100 2,100 105



The choice of which diet to use when The choice of which diet to use when estimating the cost of obtaining 2100 estimating the cost of obtaining 2100 Calories is not a trivial one, a point Calories is not a trivial one, a point emphasized in the context of Indonesia by emphasized in the context of Indonesia by Pradhan et al (2000).Pradhan et al (2000).

Table 3.6 illustrates this. Data is drawn from Table 3.6 illustrates this. Data is drawn from

the VLSS of 1992-93.the VLSS of 1992-93.


© 2003 By Default!© 2003 By Default!Slide Slide 6868Table 3.6Table 3.6Food consumption by expenditure quintile, Food consumption by expenditure quintile, Vietnam, 1992-93Vietnam, 1992-93

Quintile

Expenditure per capita,

‘000 dong/ year

% of expenditure devoted to

food

Calories per capita per day

Dong per Calorie

Lowest 562 70 1,591 0.68Low- mid 821 65 1,855 0.79Middle 1,075 60 2,020 0.87Mid- upper 1,467 54 2,160 1Upper 2,939 47 2,751 1.38Source: Vietnamn Living Standards Survey 1992- 93



To compare poverty over time, this poverty To compare poverty over time, this poverty line was updated to 1998. The cost of each line was updated to 1998. The cost of each item in the poverty-line diet of 1993 was item in the poverty-line diet of 1993 was recomputed using the prices of 1998; non-recomputed using the prices of 1998; non-food expenditure was inflated using data food expenditure was inflated using data from the General Statistics Office’s price from the General Statistics Office’s price index.index.

This yielded a poverty line of 1,793,903 and This yielded a poverty line of 1,793,903 and an associated poverty rate of 37% of which an associated poverty rate of 37% of which details are in Table 3.7.details are in Table 3.7.


© 2003 By Default!© 2003 By Default!Slide Slide 7070Table 3.7Table 3.7Poverty Lines and Headcount Measures of Poverty Lines and Headcount Measures of Poverty, VietnamPoverty, Vietnam


Poverty line (m dong/ capita/ yr)

Headcount poverty rate

Poverty overall1993 1,160 ($92) 581998 1,790 ($128) 37

Food poverty1993 750 251998 1,287 15

Note: The food poverty rate excludes any provision for non-food items; it sets the poverty line at zF.Sources: Vietnam Living Standards Surveys of 1992- 93 and 1997- 98.



There is no wholly satisfactory way to There is no wholly satisfactory way to measure the non-food component of the measure the non-food component of the poverty line and the procedures followed poverty line and the procedures followed tend to be somewhat ad hoc.tend to be somewhat ad hoc.

We saw that for Vietnam, researchers We saw that for Vietnam, researchers essentially used the level of non-food essentially used the level of non-food spending by household that were in the spending by household that were in the middle expenditure quintile in 1993.middle expenditure quintile in 1993.




The poverty line developed for South Korea The poverty line developed for South Korea (KIHASA 2000) measure the cost of food (KIHASA 2000) measure the cost of food plus the cost of housing that meets the plus the cost of housing that meets the official minimum apartment size plus the official minimum apartment size plus the cost of non-food items as measured by cost of non-food items as measured by average spending by household in the average spending by household in the poorest two-fifths of the income distribution.poorest two-fifths of the income distribution.




Following Ravallion (1998), let bFollowing Ravallion (1998), let bFF be the cost be the cost of buying 2100 Calories. Then an upper of buying 2100 Calories. Then an upper poverty line may be given bypoverty line may be given by

ff-1-1(b(bFF) = z) = zFF

which measures the income level at which which measures the income level at which the household would buy 2100 calories of the household would buy 2100 calories of food; this is essentially the poverty line used food; this is essentially the poverty line used in Vietnam.in Vietnam.

The non-food component is given by A.The non-food component is given by A.3.3 Solution A: objective poverty lines3.3 Solution A: objective poverty lines


Figure 3.6 Food Expenditure FunctionFigure 3.6 Food Expenditure Function

Consider the expenditure function shown in Consider the expenditure function shown in Figure 3.6.Figure 3.6.



A lower poverty line might be given by A lower poverty line might be given by zzFF

L L = b= bFF

which measures the income level at which which measures the income level at which the household could just buy enough food; the household could just buy enough food; but would not have any money left over to but would not have any money left over to buy anything else. This is essentially the buy anything else. This is essentially the poverty line used in Vietnam.poverty line used in Vietnam.

But even in this case, households will But even in this case, households will typically buy non-food items, as shown by C.typically buy non-food items, as shown by C.3.3 Solution A: objective poverty lines3.3 Solution A: objective poverty lines



Ravallion suggested that one might want to Ravallion suggested that one might want to compromise, and measure non-food at the mid-compromise, and measure non-food at the mid-point between two extremes, giving B.point between two extremes, giving B.

In each case, the poverty line would be given byIn each case, the poverty line would be given byz = bz = bFF + 0 (or A or B’) + 0 (or A or B’)

When Ravallion and Bidani apply the Cost of Basic When Ravallion and Bidani apply the Cost of Basic Needs approach to Indonesia, using the SUSENAS Needs approach to Indonesia, using the SUSENAS data of 1990, they find the poverty rates shown in data of 1990, they find the poverty rates shown in Table 3.8.Table 3.8.


© 2003 By Default!© 2003 By Default!Slide Slide 7777Table 3.8Table 3.8Headcount Measures of Poverty in Headcount Measures of Poverty in Indonesia, 1990Indonesia, 1990

Food Energy Intake Method

Food Food + non-food

Indonesia overall 7.9 19.6 15.1Urban 2.8 10.7 16.8Rural 10.2 23.6 14.3

Source: Ravallion and Bidani 1994.

Cost of Basic Needs Method



They computed poverty rates using these two They computed poverty rates using these two measures, for each of the main regions of measures, for each of the main regions of Indonesia, and found almost no correlation Indonesia, and found almost no correlation between the two measures.between the two measures.

This is a serious indictment of the Food Energy This is a serious indictment of the Food Energy Intake method. But is should also be clear that Intake method. But is should also be clear that every measure of poverty can be faulted, because it every measure of poverty can be faulted, because it rests in part on arbitrary assumptionsrests in part on arbitrary assumptions

In measuring poverty, there is no single truth.In measuring poverty, there is no single truth.



Philippine MethodologyPhilippine Methodology

Uses cost of basic needsUses cost of basic needs Estimates food threshold directly Estimates food threshold directly Estimates non-food costs indirectlyEstimates non-food costs indirectly Separate thresholds per province, urban and Separate thresholds per province, urban and

ruralrural



Philippine Methodology: food componentPhilippine Methodology: food component

Determines menu that meets the following Determines menu that meets the following requirements:requirements:– 2000 calories2000 calories– 100% adequate for energy and protein100% adequate for energy and protein– 80% adequate for vitamins and minerals80% adequate for vitamins and minerals

Different menu per regionDifferent menu per region Prices per province are used to determine Prices per province are used to determine

provincial food thresholdsprovincial food thresholds



Philippine Methodology: Poverty thresholdPhilippine Methodology: Poverty threshold

FE = actual food expenditure of families within +/- ten FE = actual food expenditure of families within +/- ten percentile of the food thresholdpercentile of the food threshold

TBE = total basic expenditure of families within the +/- TBE = total basic expenditure of families within the +/- ten percentile of the food threshold. TBE is an ten percentile of the food threshold. TBE is an aggregate of expenditures on food; clothing and aggregate of expenditures on food; clothing and footwear; fuel; light and water; housing footwear; fuel; light and water; housing maintenance and other minor repairs; rental or maintenance and other minor repairs; rental or occupied dwelling units; medical care; education; occupied dwelling units; medical care; education; transportation and communication; non-durable transportation and communication; non-durable furnishing; household operations and personal furnishing; household operations and personal care and effects.care and effects.

PT = FT/ (FE/TBE)PT = FT/ (FE/TBE)



Solution B: Subjective poverty linesSolution B: Subjective poverty lines

We could measure poverty by asking people to We could measure poverty by asking people to define a poverty line, and using this to measure the define a poverty line, and using this to measure the extent of poverty.extent of poverty.

For instance, in a survey one might ask “What For instance, in a survey one might ask “What income level do you personally consider to be income level do you personally consider to be absolutely minimal? This is to say that with less absolutely minimal? This is to say that with less you could not make ends meet.”you could not make ends meet.”

Answers will vary from person to person, and by Answers will vary from person to person, and by size of household, but they could be plotted, and a size of household, but they could be plotted, and a line fitted through them, to get a subjective poverty line fitted through them, to get a subjective poverty line such as z* in Figure 3.7.line such as z* in Figure 3.7.



It may also be possible to get adequate results by It may also be possible to get adequate results by asking “Do you consider your current consumption asking “Do you consider your current consumption to be adequate to make ends meet?”to be adequate to make ends meet?”

Mahar Mangahas has amassed extensive Mahar Mangahas has amassed extensive information on subjective poverty in the Philippines information on subjective poverty in the Philippines as part of the Social Weather Stations project.as part of the Social Weather Stations project.

Collected biannually since 1985, and quarterly Collected biannually since 1985, and quarterly

since 1992, the surveys poll about 1200 since 1992, the surveys poll about 1200 households.households.



Each household is shown a card with a line running Each household is shown a card with a line running across it; below the line is marked “poor” ad above across it; below the line is marked “poor” ad above the line “non-poor”, and each household is asked the line “non-poor”, and each household is asked to mark on the card where it fits.to mark on the card where it fits.

Each household is shown a card with a line running Each household is shown a card with a line running across it; below the line is marked “poor” ad above across it; below the line is marked “poor” ad above the line “non-poor”, and each household is asked the line “non-poor”, and each household is asked to mark on the card where it fits.to mark on the card where it fits.

Separately, households are also asked to define a Separately, households are also asked to define a poverty line.poverty line.


Findings of Datt regarding SWS dataFindings of Datt regarding SWS data

1.1. Self-rated poverty lines are high. In 1997, Self-rated poverty lines are high. In 1997, the median poverty line was about P10,000 the median poverty line was about P10,000 per month for a “typical” household; at that per month for a “typical” household; at that time, the government’s poverty line stood time, the government’s poverty line stood at P4495 per month. at P4495 per month.

The implication is that self-rated poverty The implication is that self-rated poverty lines are high - 60% of all households in lines are high - 60% of all households in 1997, compared to 25% using the basic 1997, compared to 25% using the basic needs line.needs line.



2.2. The self-rated poverty line has risen rapidly The self-rated poverty line has risen rapidly over time, by about 60-70% between 1985 over time, by about 60-70% between 1985 and 1997. One consequence is that there is and 1997. One consequence is that there is no trend in self-rated poverty. no trend in self-rated poverty.

Another implication is that even when there Another implication is that even when there is an economic slowdown, as occurred in is an economic slowdown, as occurred in 1997-1998, the self-rated poverty rate 1997-1998, the self-rated poverty rate hardly changes: it rose from 59% in 1996-hardly changes: it rose from 59% in 1996-1997 to 61% in 1998. 1997 to 61% in 1998.



3.3. The self-rated poverty line given by poor The self-rated poverty line given by poor households is only slightly lower than that households is only slightly lower than that for non-poor households, and in fact is not for non-poor households, and in fact is not statistically significant. statistically significant.

One might have expected poor households One might have expected poor households to have a less generous measure of the to have a less generous measure of the poverty line.poverty line.



4.4. There is a clear urban/rural difference in There is a clear urban/rural difference in proportions of the poverty line, with urban proportions of the poverty line, with urban households setting a poverty line at about twice households setting a poverty line at about twice the level of rural households. The cost of living is the level of rural households. The cost of living is certainly higher in urban areas, but by a factor of certainly higher in urban areas, but by a factor of 1.2-1.5 rather than by a factor of 2. Thus, the 1.2-1.5 rather than by a factor of 2. Thus, the urban poverty line is, in real terms, higher than its urban poverty line is, in real terms, higher than its rural counterpart. rural counterpart. Possible reasons:Possible reasons:-there is more inequality in the urban areas, and -there is more inequality in the urban areas, and this raises expectationsthis raises expectations



Possible reasons:Possible reasons:- households in urban areas may have - households in urban areas may have more exposure to the media , and may have more exposure to the media , and may have ben more affected by consumerismben more affected by consumerism

- urban households may be more attuned - urban households may be more attuned to political processes and their estimate of to political processes and their estimate of the poverty line may include and element of the poverty line may include and element of strategic behavior – trying to influence strategic behavior – trying to influence policy makerspolicy makers



Conclusion: Self-rated measures of poverty Conclusion: Self-rated measures of poverty cannot supplant the more “objective” cannot supplant the more “objective” measures of poverty.measures of poverty.

© 2003 by default! slide 1 poverty lines celia m. reyes introduction to poverty analysis nai,...

Documents

food poverty line

overall poverty line

poverty line wellbeing

poverty analysis nai

poverty lines celia

basic needs

household needs

basic consumption