Presentation on theme: "NON-INCOME POVERTY IN HOUSEHOLD SURVEYS Poverty & Inequality Group Development Research Group The World Bank Poverty and Inequality Course Module 1: Multi-Topic."— Presentation transcript:
NON-INCOME POVERTY IN HOUSEHOLD SURVEYS Poverty & Inequality Group Development Research Group The World Bank Poverty and Inequality Course Module 1: Multi-Topic Household Surveys 03/08/2013
Consumption/income Traditional measure for poverty – Consumption Detailed list of food and non-food consumption items Diary or recall method – Income Detailed information on all source of income including wage and non wage work, farm and non-farm work Also includes remittances, rental income, and pensions Consumption is more commonly used to measure poverty headcount than income Full/Complete consumption used to measure – Poverty headcount – MDG #1 ($1/day) – Private consumption for National Accounts
When you can’t or don’t need to collect complete consumption Consumption expenditure data is notorious for being difficult to collect in household surveys – often requiring a disproportionate amount of the total interview time Shorter consumption modules: cheaper and easier – Reduce time to complete the interview length of fieldwork Respondent and interviewer burden/fatigue “cost” to interviewers of questionnaires with lots of pages Shorter consumption modules can be used to – To rank households: identify poorer from richer households or – To approximately identify “poor” households
Non-monetary indicators There are other dimensions of poverty apart from insufficient income or consumption Use of proxy indicators Correlates of consumption can be used to evaluate welfare status of households and rank poverty status – Assets – Health status – Education level – Access to public services – Housing conditions – Level of indebtedness – Unemployment
Outline Partial consumption/Income Asset Indices Subjective wellbeing Multidimensional poverty
Partial Consumption/income Shorter list of consumption items Subset list of items from mean of 75 food items in LSMS (mean of 130 total consumption items) Collapsed list of items into broader categories. Example: Susenas short form collapses over 100 items into 15 items Take top 10 food items with respect to share in total food expenditure. – A shorter food module, down to 1 page from several. But excluding rare or infrequently consumed items may not reduce interview time by much at the mean. – Tanzania 49 minutes for 7-day food recall, 58 food items 41 minutes for 7-day food recall, 17 food items (subset)
Asset index Can be used to characterize household economic status in the absence of expenditure and income data An aggregate index is constructed based on consumer durable assets owned by household members, along with a set of housing characteristics – Note: this is not wealth in the formal sense of the value of household assets owned minus liabilities. Data used to construct asset indices are simple to collect and are frequently available Stata code – factor var_1 var_2 var_3 predict new_var It is better to choose assets with same unit. Be careful when mixing different unit like counts and binary variables
Asset index The main proponents of this measure are Filmer and Pritchett – Filmer and Pritchett overcome the lack of data on income and consumption expenditures in DHS surveys by constructing a proxy for long-run household wealth, using survey information on assets and using the statistical technique of principal components. Deon Filmer and Kinnon Scott. 2012.“Assessing Asset Indices” Demography 49 (1):359-92 Compare these alternative approaches to welfare measurement – Overlap between “consumption” and “index” approaches – Implication for the analysis of welfare gradients – Correlates of the congruence between approaches 10
Overlap in the classification in the poorest quintiles Reasonable, but not “perfect”, match between rankings: Asset indices highly related to each other, less congruence with PCE Per Capita HH expend Predicted per capita HH expend. PC index, all indicators PC index, assets only IRT index Share weighted average Count index PC value of durable goods (1)(2)(3)(4)(5)(6)(7)(8) Proportion of the population classified in the poorest 20 percent by per capita expenditure who are in the poorest 20 percent according to other welfare indices Albania10.470.420.41 0.370.380.47 Brazil10.680.640.620.630.570.63 Nicaragua10.560.510.460.50.480.490.52 Uganda10.520.480.430.510.470.48 Vietnam10.540.490.50.470.490.480.49 Note: Blank entry indicates that data are not available. Cross country averages are weighted
Asset index Findings – Reasonable, but not “perfect”, match between rankings – Inferences about inequalities in education, health care use, fertility, child mortality, labor market outcomes are quite robust – Rankings are most similar in settings with small transitory shocks to expenditure, or with little random measurement error in expenditure. – In settings where private goods such as food are the main component of expenditures, asset indices and per capita consumption yield the least similar results. Asset index may do well when we are not able to measure consumption very well. For example in urban areas where we can have a lot of underreporting of consumption. In such instances, assets which are more observable could do better. 12
Subjective Wellbeing (SWB) SWB concerns peoples’ self-reported assessment of their own well-being. – sometimes questions are asked about the household from one individual. – Could be historical or refer “current” situation Three broad approaches have been identified when measuring subjective well-being; – Evaluative Cantril ladder – Experience – Eudemonic
Subjective Wellbeing (SWB) SWB provides a wider focus than income/consumption levels alone. Could be directly related to poverty or less clearly linked to poverty – Self-reported poverty (Are you poor on scale of 1-6) – Self-reported wellbeing (less clearly linked to the word "poor") for example subjective reporting of health status, welfare, satisfaction of life, etc Subjective poverty line (SPL) ( Pradhan and Ravallion 2000) – questions on the perceived adequacy of (food or total) household consumption – SPL is ‘the level of total spending above which respondents say (on average) that their expenditures are adequate for their needs’.
Subjective Wellbeing (SWB) Example of SWB questions – Overall, how satisfied are you with your life nowadays? – Do you feel poor? – Overall, how happy did you feel yesterday? – Minimum income question (MIQ): is a subjective survey question designed to directly assess the delicate balance between what people earn and what they spend. The question reads, "Living where you do now, what is the smallest income you and your family would need (before any deductions) to make ends meet each month?“ – Cantril ladder: Rate your current life on a ladder scale for which 0 is ‘the worst possible life for you’ and 10 is ‘the best possible life for you’.
Subjective Wellbeing SWB does not necessarily correlate with increased standard of living The Easterlin paradox: Average happiness do not increase as countries grow wealthier Studies have found that in some rich countries, happiness has not grown along with living standards – Relative nature – Increase living standards temporary happiness
Subjective Wellbeing How well do subjective wellbeing correlate with the objective income measure? – Cojocaru and Diagne (2013) find weak correlation between subjective relative income and household relative welfare position as measured by consumption or assets in Central Asia and Europe. – Alem et al (2012) also find that while consumption poverty has declined in Ethiopia, subjective poverty remains largely unchanged
Subjective Wellbeing (SWB) Problem of different scales across people - one person’s standard of welfare may be different from another person’s standard – Married individuals – Single – Urban People may not be consistent overtime. (Kahneman 2010) – not explained by any material change – the response on Friday may be diff from Monday
Subjective Wellbeing Responses can be affected by – Wording of question – Context of question Some factors can affect perception – Previous poverty status can affect current perception – Relative economic standing – Being engaged in any kind of income-generating job Hypothetical vignettes: – Respondents score a set of vignettes describing different scenarios related to the topic (such as poverty status) for a hypothetical person/household – Vignettes can help reveal the respondents’ own scale
Subjective Wellbeing: References A. Deaton and A. Stone. 2013. “Two happiness puzzles” Forthcoming, American Economic Review Alexandru Cojocaru, and Mame Fatou Diagne. 2013. “ How Reliable and Consistent Are Subjective Measures of Welfare in Europe and Central Asia? Evidence from the Second Life in Transition Survey” World Bank Working Paper Yonas Alem, Gunnar Köhlin and Jesper Stage. 2012. “ The Persistence of Subjective Poverty in Urban Ethiopia” Working paper in economics, University of Gothenburg
Subjective Wellbeing: References Beegle, Kathleen, Kristen Himelein and Martin Ravallion. 2012. “Frame-of-Reference Bias in Subjective Welfare Regressions.” Journal of Economic Behavior and Organization 81:556-570. Daniel Kahneman1 and Angus Deaton (2010). “High income improves evaluation of life but not emotional well-being” King, G., C. J. L. Murray, J. A. Salomon, and A. Tandon. 2004. “Enhancing the Validity and Cross-Cultural Comparability of Measurement in Survey Research.” American Political Science Review, 98.1, pp. 191-207
Subjective Wellbeing: References Pradhan, M. and Ravallion, M. (2000) “Measuring poverty using qualitative perceptions of consumption adequacy,” The Review of Economics and Statistics, 82(3), pp. 462–71. Daniel Kahneman and Alan B. Krueger (2006) “Developments in the measurement of subjective well-being” The Journal of Economic Perspectives, 20, 3-24 Gero Carletto and Alberto Zezza (2006) “Being Poor, Feeling Poorer: Combining Objective and Subjective Measures of Welfare in Albania” King, G. and J. Wand. 2007. “Comparing Incomparable Survey Responses: Evaluating and Selecting Anchoring Vignettes.” Political Analysis, 15, pp. 46-66.
Multidimensional Poverty Poverty is not just about income, it is made up of several factors – Poor health – Limited access to education – Life expectancy – Lack of income – Inadequate living standard – Limited access to information and knowledge – Lack of empowerment A multidimensional measure of poverty can incorporate these different factors/indicators to capture the complexity of poverty and better inform policies to relieve it.
Multidimensional Poverty Index Developed by James foster and Sabrina Akirie – Under the Oxford Poverty & Human Development Initiative and the United Nations Development Program – Newer version of the human development index (HDI) They use two forms of cutoffs to identify the poor – First- Traditional dimension-specific deprivation cutoff, which identifies whether a person is deprived with respect to that dimension. – Second- Poverty cutoff which delineates how widely deprived a person must be in order to be considered poor. The methodology allows for – Decomposability - Poverty identification for subgroups of the population – Allows identification of dimensional deprivations that contributes the most to poverty for any given group
Multidimensional Poverty Multidimensional poverty index (MPI) – Uses 3 dimensions Health: Nutrition and Child mortality Education: Years of schooling and Schooling attendance Living standards: Cooking fuel, Sanitation, Water, Electricity, Floor, Assets MPI reveals not only how many people are poor but also the composition of their individual poverty A person is multidimensionally poor if he or she is deprived in at least one third of weighted indicators Provides one number by which countries can be ranked using a simple tool – MPI = H * A – H: Percentage of people who are “multidimensionally” poor – A: Average intensity of MPI poverty across the poor (%)
Multidimensional Poverty Limitation of multidimensional indices Uses relative weights for each dimension, which are generally chosen somewhat arbitrarily by the analyst. Ravallion (2011) suggests a “dashboard approach” with a set of “multiple indices” that measure various dimensions of poverty rather than a single ‘multidimensional index’” Ferreira and Lugo (2012) look at the pros and cons of the single index versus “dashboard approach”
Multidimensional Poverty: References Sabina Alkire and James Foster (2011) “Counting and multidimensional poverty measurement”, Journal of Public Economics 95: 476–487) Francisco Ferreira and Maria Ana Lugo (2012) “Multidimensional poverty analysis: Looking for a middle ground” ECINEQ working papers Ravallion, Martin (2011): “On multidimensional indices of poverty”, Journal of Economic Inequality, 9 (2): 235-248.
A few final thoughts Other options for measuring poverty When conducting smaller surveys: – It is good to have a handful of comparable questions to a larger/national survey. – It helps with external validity. – Allows survey to survey imputation