Multidimensional Poverty Measurement and Analysis

Transcription

1 Multidimensional Poverty Measurement and Analysis Sabina Alkire, James Foster, Suman Seth, Maria Emma Santos, José Manuel Roche and Paola Ballon 22 June 25 University of Oxford

2 Multidimensional Measurement Methods: multidimensionalpoverty.org

3 Multidimensional Measurement Methods: multidimensionalpoverty.org

4 Multidimensional Measurement Methods: multidimensionalpoverty.org

5 Multidimensional Measurement Methods: multidimensionalpoverty.org

6 Multidimensional Measurement Methods: multidimensionalpoverty.org

7 Multidimensional Measurement Methods: Contents Chapter Introduction Chapter 2 The Framework Chapter 3 Overview of Methods for Multidimensional Poverty Assessment Chapter Counting Approaches: Definitions, Origins and Implementations Chapter 5 The Alkire-Foster Counting Methodology Chapter 6 Normative Choices in Measurement Design Chapter 7 Data and Analysis Chapter 8 Robustness Analysis and Statistical Inference Chapter 9 Distribution and Dynamics Chapter Some Regression models for AF measures

8 Multidimensional Measurement Methods: Contents Chapter Introduction Chapter 2 The Framework Chapter 3 Overview of Methods for Multidimensional Poverty Assessment Chapter Counting Approaches: Definitions, Origins and Implementations Chapter 5 The Alkire-Foster Counting Methodology Chapter 6 Normative Choices in Measurement Design Chapter 7 Data and Analysis Chapter 8 Robustness Analysis and Statistical Inference Chapter 9 Distribution and Dynamics Chapter Some Regression models for AF measures

9 Multidimensional Measurement Methods: Contents Chapter Introduction Chapter 2 The Framework Chapter 3 Overview of Methods for Multidimensional Poverty Assessment Chapter Counting Approaches: Definitions, Origins and Implementations Chapter 5 The Alkire-Foster Counting Methodology Chapter 6 Normative Choices in Measurement Design Chapter 7 Data and Analysis Chapter 8 Robustness Analysis and Statistical Inference Chapter 9 Distribution and Dynamics Chapter Some Regression models for AF measures

10 Background

11 Motivation: Normative: Capability Sen, A. K. (992). Inequality Re-examined. Oxford: Oxford University Press. Empirical: Mis-match Stewart, F., Saith, R., and Harriss-White, B. (27). Defining Poverty in Developing Countries. Basingstoke: Palgrave Macmillan. Bourguignon, F., Bénassy-Quéré, A., Dercon, S., Estache, A., Gunning, J.W., Kanbur, R., Klasen, S., Maxwell, S., Platteau, J-P., and A. Spadaro (2) Millennium Development Goals: An Assessment, in R. Kanbur and M. Spencer (eds.), Equity and Growth in a Globalizing World. World Bank, ch. 2. Policy: Atkinson, A. B. (23). Multidimensional Deprivation: Contrasting Social Welfare and Counting Approaches, Journal of Economic Inequality, ():5-65.

12 Methods: Dashboard: Ravallion, M. (2b). On Multidimensional Indices of Poverty, Journal of Economic Inequality, 9(2): Composite: Nardo, M., Saisana, M., Saltelli, A., Tarantola, S., Hoffman, A., and Giovannini, E. (28). Handbook on Constructing Composite Indicators: Methodology and User Guide. Ispra, Italy: OECD.

13 Methods: Dominance: Duclos, J. Y., Sahn, D. E., and Younger, S. D. (26a). Robust Multidimensional Poverty Comparisons, The Economic Journal, 6(5): Statistical: Asselin, L. M. (29). Analysis of Multidimensional Poverty: Theory and Case Studies. Dordrecht: Springer. Fuzzy: Lemmi, A. and Betti, G. (26). Fuzzy Set Approach to Multidimensional Poverty Measurement. New York: Springer. Axiomatic: Bourguignon, F. and Chakravarty, S. R. (23). The Measurement of Multidimensional Poverty, Journal of Economic Inequality, (): Counting: Nolan, B. and Whelan, C. (2). Poverty and Deprivation in Europe. Oxford: Oxford University Press.

14 Moving Forward: Axiomatic & Counting Identification: Sen, A. K. (976). Poverty: An Ordinal Approach to Measurement, Econometrica, (2): Decomposability: Foster, J. E., Greer, J., and Thorbecke, E. (98). A Class of Decomposable Poverty Measures, Econometrica, 52(3): Functionings & Counting: Brandolini, A., D Alessio, G., 998. Measuring Well-being in the Functioning Space. Mimeo. Rome. Banco d Italia Research Department.

15 The Alkire-Foster Methodology

16 Multidimensional Data Matrix of well-being scores for n persons in d dimensions X Dimensions = Persons z = ( ) Cutoffs

17 Multidimensional Data Replace entries: if deprived, if not deprived Dimensions X = Persons z = ( ) Cutoffs These entries fall below cutoffs

18 Deprivation Matrix Replace entries: if deprived, if not deprived Dimensions g = Persons

19 Identification Weights Deprivation Matrix Weighted Deprivation Matrix = w w w w w w w g [ ] 3 2 w w w w w = = g Dimensions Dimensions

20 Identification Counting Deprivations Assuming equal weights and c Persons = g 2 d j j w d = = Dimensions

21 Identification Q/ Who is poor? Dimensions c g = 2 Persons

22 Identification Union Approach Q/ Who is poor? A/ Poor if deprived in any dimension c i c Persons 2 = g Dimensions

23 Identification Union Approach Q/ Who is poor? A/ Poor if deprived in any dimension c i c Persons Observations Union approach often predicts very high numbers. Charavarty et al 98, Tsui 2, Bourguignon & Chakravarty 23 etc use the union approach 2 = g Dimensions

24 Identification Intersection Approach Q/ Who is poor? A2/ Poor if deprived in all dimensions c i = d c Persons 2 = g Dimensions

25 Identification Intersection Approach Q/ Who is poor? A2/ Poor if deprived in all dimensions c i = d c Persons Observations Demanding requirement (especially if d large) Often identifies a very narrow slice of population Atkinson 23 first to apply these terms. 2 = g Dimensions

26 Identification Dual Cutoff Approach Q/ Who is poor? A/ Fix cutoff k, identify as poor if c i > k c Persons = g 2 Dimensions

27 Identification Dual Cutoff Approach Q/ Who is poor? A/ Fix cutoff k, identify as poor if c i > k (Ex: k = 2) c Persons = g 2 Dimensions

28 Identification Empirical Example Poverty in India for dimensions 9% of population would be targeted using union % using intersection We need something in the middle (Alkire and Seth 29)

29 k = 2 Censor data of non-poor Aggregation g = Dimensions c 2 Persons

30 Aggregation k = 2 Censored weighted deprivation matrix and censored deprivation score Dimensions c(k) ( ) g k = 2 Persons

31 k = 2 Censored weighted deprivation matrix c(2) Persons Two poor persons out of four: H = /2 Aggregation Headcount Ratio ( ) = 2 g 2 Dimensions

32 Suppose the number of deprivations rises for person 2 Dimensions c(2) Critique ( ) = 2 g 3

33 Suppose the number of deprivations rises for person 2 Dimensions c(2) Two poor persons out of four: H = ½ No change! Violates dimensional monotonicity Critique ( ) = 2 g 3

34 Aggregation Return to the original censored weighted deprivation matrix Dimensions c(2) Persons ( ) = 2 g 2

35 Aggregation - Intensity Need to augment information Dimensions c(k) c(k)/d ( ) = 2 g 2 / 2 / Deprivation shares among poor

36 Aggregation - Intensity Need to augment information Dimensions c(k) c(k)/d A = average deprivation share among poor = 3/ ( ) = 2 g 2 / 2 / Deprivation shares among poor

37 Aggregation: Adjusted Headcount Ratio Adjusted Headcount Ratio = M = HA Dimensions c(k) c(k)/d Persons M = HA = (/2)*(3/) =.375 ( ) = 2 g 2 / 2 /

38 Aggregation: Adjusted Headcount Ratio Adjusted Headcount Ratio = M = HA = μ( (k)) Dimensions c(k) c(k)/d Persons M = HA = (/2)*(3/) =.375 M = μ( (k)) = 6/6 =.375 ( ) = 2 g 2 / 2 / g g

39 Aggregation: Adjusted Headcount Ratio Suppose the number of deprivations rises for person 2 Dimensions c(k) c(k)/d Persons ( ) = 2 g 3 / 3/

40 Aggregation: Adjusted Headcount Ratio Suppose the number of deprivations rises for person 2 Dimensions c(k) c(k)/d g ( 2) = 3 3/ / Persons A = average deprivation share among poor = 7/8 M changes! M = 7/6 =.375 Satisfies dimensional monotonicity

41 Methodology: Adjusted Headcount Ratio Interpretation: conveys information on deprivations Applicability: valid for ordinal data Simplicity: easy to compute Useful properties Subgroup decomposition Dimensional breakdown Expandable: If variables are all cardinal can go further

42 Multidimensional Measurement Methods: Contents Chapter Introduction Chapter 2 The Framework Chapter 3 Overview of Methods for Multidimensional Poverty Assessment Chapter Counting Approaches: Definitions, Origins and Implementations Chapter 5 The Alkire-Foster Counting Methodology Chapter 6 Normative Choices in Measurement Design Chapter 7 Data and Analysis Chapter 8 Robustness Analysis and Statistical Inference Chapter 9 Distribution and Dynamics Chapter Some Regression models for AF measures

43 Countries This methodology can be used to design robust official statistics Official multidimensional poverty statistics are used in Mexico, Colombia, Bhutan, Chile and others. This book can be useful for technical advisors in countries that are exploring or actively designing multidimensional poverty measures The MPPN (Multidimensional Poverty Peer Network) is a network of some countries with such an interest. It can also be used for designing other policy-relevant indices Examples: Well-being, Child Poverty, Empowerment (WEAI)

44 National MPIs - Reflect National Priorities - Vital for policy - Measure to Manage ~ Target, Coordinate, M&E

45 MPI-Colombia Educational Conditions Childhood & Youth Work Health Housing & Public Services Schooling Illiteracy. Used to allocate resources in national development plan School Attendance At the right level Access to infant services No Child Labour Absence of long-term unemployment Formal work Coverage Access to health care given a necessity.. Improved Water Sanitation Flooring Exterior Walls Overcrowding.5.

46 - key aspects:. Reflects the objectives of social policy 2. Coordinates public policy sectors 3. Monitors public policy. Informs Decision making:. Geographic targeting 2. Programme composition 3. Graduation from CCTs

47 International Agencies & Civil Society Use for Programme Quality (inward looking) Diagnostic and prioritization Baseline and M&E tool Applied to distinct domains (poverty, empowerment, child rights, etc) Understanding underlying multidimensional processes Multidimensional measures force institutions to break silos, and to discuss interaction between dimensions!

48 International Agencies & Civil Society Use for Policy & Research or Advocacy (outward looking) Accountability tool to hold governments to account Monitoring multidimensional outcomes (Global MPI, UNICEF MODA, MPI & Post25) Policy analysis for evidence based advocacy

49 Teaching Masters/PhD level courses ü Module on poverty measurement ü Comprehensive overview for teachers ü Unified notation and framework From book website: ü Online chapters ü Video presentations ü PowerPoints ü Paper exercises ü Stata do files for statistical exercises

50 How Can the Book be Used for Teaching?

51 How Can the Book be Used for Teaching?

52 How Can the Book be Used for Teaching?

53 How Can the Book be Used for Teaching?

54 How Can the Book be Used for Teaching?

55 How Can the Book be Used for Teaching?

56 Researchers Empirical research questions Descriptive Inferential Measures: Comparable M for OECD countries. Applied measures: Health poverty, Child/elder poverty Assess the transmission mechanisms between polices and multidimensional poverty - Micro and macro regressions; - Impact evaluation analyses

57 Researchers Methodological research questions Measurement Data analysis Axioms - distributional analysis Standards for robustness analyses Treatment of missing values Treatment of panel data and demographics (dynamic analysis of poverty)

58