Subjective Versus Objective Welfare: Poverty in Palanpur

Poverty LinesMichael LokshinDECRG-POThe World Bank1Poverty LinesThe welfare ratioThe theoretical idealPractice: Objective poverty linesPractice: Subjective poverty linesRecommendations for practice

2Poverty lines1. The welfare ratio3Poverty linesAdd up expenditure on all commodities consumed (with imputed values at local market prices) and

Deflate by a poverty line that might depend on household size, composition, and location/date)

The real expenditure or welfare ratio is:

where Pij is a price paid for good j by household i Qij is a quantity of good j consumed by household i

2. The theoretical ideal 4Poverty linesPoverty line should be absolute in the space of welfare:Two individuals with the same level of welfare are treated similarlyPoverty line should satisfy the weak Pareto principle that states that a welfare gain cannot increase povertyThe ideal poverty line is the minimum cost of achieving a reference level of welfare by a given household:

Pi are the prices paid by a household i for goods Xi are household characteristics that could affect households welfare Zw is is the reference level of utility

4= cost to a person with characteristics (Xi) of the level of welfare needed to escape poverty (Zw) at prevailing prices Pij , j=1,..,m. The theoretical ideal5Poverty linesLinear approximation of the expenditure function:

where Qij is the consumption of good j that brings household i to the reference welfare level Zw

Common issues in PracticeIdentification of consumers cost function from demand behavior

Are there goods that should be included in the consumption aggregate but not the poverty line (qat in Yemen?)

Identification of external effects (interdependence). Utility depends on own consumption and relative position.

Absolute versus relative poverty? Is there really a difference? Absolute in the space of welfare can mean relative in the commodity space (Ravallion and Chen, 2009)

6Poverty linesCalibration of poverty lines to supplementary welfare data7Poverty linesImperfect welfare indicator Wi for household I

Examples:Food shareNutritional/Health statusSelf-rated welfare (Cantrils scale)Perceived consumption adequacyIncome Yi and other welfare-relevant characteristics

Poverty line could be derived from estimating:

Absolute vs. relative poverty?8Poverty linesWelfare depends on relative income:

M is mean income in a country. The reference level of utility:

Thus, the poverty line is absolute in welfare space, but is relative in the consumption space the poverty line is a function of mean income:

For a poverty line to be absolute in the space of welfare (that is yield Zw) the commodity-based poverty line Z have to rise as M is rising.

Poverty lines for different countries 9Poverty lines

Poverty lines10

Poverty rates for different countries3. Practice: Objective Poverty Lines11Poverty linesCost of Basic Needs (CBN) method: Poverty line is a cost of a bundle of goods deemed sufficient for basic needs.Food-share version of CBN : Poverty line is a cost of the food-energy requirement

Food-Energy Intake (FEI) method:Find expenditure or income at which food-energy requirements are met on average for each region/socio-economic group

Problems to be aware of12Poverty linesDefining basic consumption needsSetting food energy requirements: problems with variability of activity levels; multiple equilibrium.Setting basic non-food consumption needs

Consistency in terms of welfare:Is the same standards of living being treated the same way in different subgroups of the poverty profile? Is the definition of welfare consistent with the definition of poverty? Should some goods purchased by the poor be included into the poverty bundle?

How sensitive are the rankings in a poverty profile to these choices?Consistency of Poverty LinesPoverty lines13Two bundles yield same food-energy intake, but the urban bundle is almost certainly preferable to the rural bundleThe standard of living at the urban poverty line is higher than at the rural lineThese two poverty lines are inconsistent providing different welfare levels% of calories from foodUrbanRuralRice5040Corn1040Vegetables2010Meat2010Difference in the cost per capita Poverty lines14Food Consumption and Cost of a calorie by quintileExpenditure QuintilePercentage of expenditure on foodCalories per capita, per dayCost per calorieLowest701,5910.682651,8550.793602,0200.874542,1601.00Highest472,7511.38Example 2: Food-energy intake Method15Poverty lines

Different sub-groups attain food energy requirements at different standards of living, in terms of real consumption expenditures. 16IncomeFood intake,kcal/dayZuZr2100UrbanRuralZAllowing for differences in relative prices17Poverty linesIdeally we only want to adjust the poverty bundle for differences in relative prices

The problem is how to implement this ideal in practice

The identification problem remains

Key thing: To control for welfare differences

Allowing for differences in relative prices18Poverty linesParametric demand models: If we know the parametric utility function then or we can figure it out from demand behavior then use this to determine the cost of the reference welfare level in each region

Numerical methods:Look at consumption behavior of poorest x% nationally in each region of the countryCost the consumption bundle of that group in each regionCalculate the poverty rate nationally and iterate if the answer differs too far from xMethods of Setting Poverty Lines Do Matter!19Poverty linesPoverty Headcount %UrbanRuralIndonesiaFEI method16.814.3CBN method10.723.6TunisiaFEI method7.35.7CBN method3.513.1World Bank Method: Cost per calorie 20HHIDFood itemsCaloric content per 1 kgQuantity purchased (kg)Price paid ($)1001Milk1000.51.41001Bread2001.03.01001Meat6000.35.01002Bread2001.01.51002Butter12000.24.01002Milk1000.51.51002Sugar7001.04.21002Potatoes3005.07.31003Meat6000.77.01003Beans5000.52.21003Veg. Oil4500.33.4Caloric value5020018020024050700150042025013540.53925$0.010TotalCost per calorieWorld Bank Method: Cost per calorie 21HHIDFood itemsCaloric content per 1 kgQuantity purchased (kg)Price paid ($)1001Milk1000.51.41001Bread2001.03.01001Meat6000.35.01002Bread2001.01.51002Butter12000.24.01002Milk1000.51.51002Sugar7001.04.21002Potatoes3005.07.31003Meat6000.77.01003Beans5000.52.21003Veg. Oil4500.33.4Caloric value50200180200240507001500420250135$0.015Mean cost per calorie0.022Cost per calorie0.0070.016World Bank Method: Cost of a calorie Poverty lines22Pick a nutrition requirement: 2100 Kcal per day (NR)Select a group of households around the poverty lineUsually take 2nd to 5th deciles of expenditure distributionCalculate the total spending of these households on food (FS).Calculate the total caloric content of these purchases (TC)Calculate the cost of a calorie for this group: CC=FS/TCCalculate the cost of food poverty line as: FPL = NRCC=NRFS/TCEstimating Cost per CaloriePoverty lines23Expected location of povery lineFourth decileSecond decile24IncomeZ125Price(mean)2.453.252.052.01MarketSupermarketPrice of a calorie/ goodsIs the width of the band important? Case of dual prices25IncomeZ125Price(mean)3.503.453.653.75MarketSupermarketPrice of a calorie/ goodsStreet vendorLocal storeIs the width of the band important? Case of multiple pricesPoverty lines26Adjust Food Poverty Line for non-food expenditures

LPL: Select a group of households whose total expenditure is equal (close) to Food Poverty Line

Estimate average share of non-food consumption SL in their total consumption expenditure

Calculate: LPL = FPL/(1- SL)

Example: FPL = $50, SL=0.2 LPL = $50/(1-0.2)=62.5

World Bank Method: Lower poverty line World Bank Method: Upper poverty linePoverty lines27UPL: Select a group of households whose food expenditure is equal to Food Poverty Line

Estimate average share of non-food consumption in their total consumption expenditure SU

Calculate: UPL = FPL/(1- SU)

Example: FPL = $50, Su=0.35 UPL = $50/(1-0.35)=76.9

Non-food adjustment diagram28Poverty lines

Updating poverty lines over timePoverty lines29Once poverty line is established, it is important to update it correctly for the new time period

It is incorrect to recalculate poverty lines every year: by doing that we would use relative poverty lines (similar to urban/rural example)

Two ways to go: update old poverty line using new prices. That would answer the question: How many people can afford the old basket now?update new poverty line using old prices. That would answer the question: How many people could afford the new basket in the past?

4. Practice: Social Subjective Poverty Line30Poverty linesMinimum Income Question (MIQ): What income do you consider to be absolutely minimal, in that you could make ends meet with less?

Practice: Social Subjective Poverty LinePoverty lines31Latent individual welfare Wi :

Derive the Social Subjective poverty line as:

In practice Wi can be approximated as:MIQ (OLS)Consumption adequacy questions (Ordered Probit)Economic ladder questions (Ordered Probit)

Examples of subjective welfare questions for Jamaica and Nepal32Poverty linesRespondents asked whether their food, housing and clothing expenditures were adequate for their family needs.

The implied subjective poverty lines are robust to alternative methods of dealing with other components of expenditure

The aggregate poverty rates accord closely with the poverty rates based on poverty lines derived with CBN method.

However, the geographic and demographic poverty profiles differ substantially from those based on objective poverty linesPoverty rates based on subjective and objective poverty lines:Poverty lines33self rated poverty lines are higher than official poverty linessubjective poverty rates are also higher than the official (income) poverty rates

5. RecommendationsPoverty lines34The WB recommends using objective, absolute poverty lines in developing countries

Usually, several poverty lines are calculated. For example, Lower and Upper poverty lines in the WB method

Always conduct sensitivity analysis. Test the degree to which the results are sensitive to the choice of poverty lines. This can be done by repeating the calculations for different lines and comparing results