1 Measuring poverty 2 Multidimensional poverty 3 Poverty
- Slides: 19
1. Measuring poverty 2. Multidimensional poverty 3. Poverty Dynamics 4. Panel Data 5. Inference with Panel Data 6. International Poverty Comparisons 7. Vulnerability 8. Tackling Poverty Module 3 Poverty Dynamics JONATHAN HAUGHTON jhaughton@Suffolk. edu APRIL 2020
Objectives 1. Explain why we need to measure poverty over time (“poverty dynamics”) 2. Enumerate the main sources of data on poverty dynamics 3. Construct a transition matrix using panel data 4. Identify the chronically, persistently, and transient poor April 2020 JH: COURSE ON POVERTY MEASUREMENT 2
What, why “poverty dynamics” Simply: the measurement of poverty over time Useful to: ◦ Monitor, evaluate, the effects of policies, shocks, projects ◦ Distinguish between households that are persistently poor (so need more income) and occasionally poor (may need insurance, transfers) April 2020 JH: COURSE ON POVERTY MEASUREMENT 3
How? Ask about the past Rarely done, because of concerns about accuracy ◦ e. g. Survey after a crisis, ask questions about situation before and during crisis April 2020 JH: COURSE ON POVERTY MEASUREMENT 4
Case: Krishna et al. ◦ 316 households in 20 villages in W. Kenya; also a similar approach in Rajasthan. ◦ Asked villagers to identify who among them was poor 25 years earlier. Verified with individuals. ◦ Transition matrix: April 2020 JH: COURSE ON POVERTY MEASUREMENT 5
How? Repeated Cross-Sections Commonest situation: ◦ Cross-section data at two or more points in time ◦ Example: Dollar and Kraay (2002) ◦ 418 “episodes” from 137 countries over 4 decades allowing comparison of poverty over 5 -year spans ◦ Conclusion: “Growth is good for the poor” (in the long run) ◦ Example: Chen and Ravallion (2008) ◦ Use information from 675 surveys for 116 countries to estimate evolution of poverty in LDCs, 1981 -2005 ◦ For details, see Povcal. Net. April 2020 JH: COURSE ON POVERTY MEASUREMENT 6
Example: Cambodia Two similar surveys; poverty lines adjusted for inflation; Poverty fell modestly. April 2020 JH: COURSE ON POVERTY MEASUREMENT 7
Example: Peru Calculate poverty rates by group of household Report relative risks ◦ e. g. P 0 = 10% nationally, 15% for laborers, so relative risk of being poor is +50% for laborers. April 2020 JH: COURSE ON POVERTY MEASUREMENT 8
How: Panel Data For panel data, we survey the household (or individual) more than once. ◦ E. g. 1993 Vietnam Living Standards Survey interviewed 4, 800 households. Of these, 4, 305 were interviewed again in 1998. ◦ E. g. Institute for Crop Research in the Semi-Arid Tropics (ICRISAT) surveyed 240 households annually from 1975 -1985 in six villages in southwestern India. More details in Module 5 April 2020 JH: COURSE ON POVERTY MEASUREMENT 9
Transition matrix: poverty, Rwanda April 2020 JH: COURSE ON POVERTY MEASUREMENT 10
April 2020 JH: COURSE ON POVERTY MEASUREMENT 11
Transition Matrix: quintiles Often uses quintiles, as on the next slide for Vietnam, 1993 -1998. ◦ Note the considerable mobility; 59% changed quintile between 1993 and 1998. ◦ “Shooting stars” rose at least 2 quintiles. ◦ Interesting to ask why. [Haughton et al. 2001] ◦ Signal or noise? Pritchett et al. think half of observed mobility may reflect measurement error April 2020 JH: COURSE ON POVERTY MEASUREMENT 12
April 2020 JH: COURSE ON POVERTY MEASUREMENT 13
April 2020 JH: COURSE ON POVERTY MEASUREMENT 14
Issue: Inconsistency in method over time April 2020 JH: COURSE ON POVERTY MEASUREMENT 15
Issue: deflation When comparing two surveys, adjust for inflation. ◦ The Achilles heel of inter-temporal poverty comparisons ◦ India: official price indexes over-inflated poverty line ◦ 4 percentage points too much, 1993/94 to 1999/2000 ◦ So reduction in poverty was understated ◦ Also, urban cost of living overstated ◦ +15%, not +36%, relative to rural ◦ It matters! April 2020 JH: COURSE ON POVERTY MEASUREMENT 16
Price adjustments April 2020 JH: COURSE ON POVERTY MEASUREMENT 17
Issue: Questionnaire comparability ◦ Southwest China Poverty Monitoring Survey ◦ Mean income per capita: 855 to 993 yuan, 1995 to 1996; +16%! ◦ 1995 survey: one-time recall; 1996 survey: diary ◦ India: National Sample Surveys ◦ 1993/4: P 0 = 36%; down to 26% by 1999/2000. ◦ But latter questionnaire had fewer consumption items; shorter recall period for food; longer recall period for some durables. ◦ Deaton (2001): if survey instrument unchanged, 28% poverty rate in 1999/2000 April 2020 JH: COURSE ON POVERTY MEASUREMENT 18
Further reading On price indexes, Haughton & Khandker, Chapter 16 ◦ Base-weighted Laspeyres index is commonly used, but overstates inflation. Tornqvst index better. Deaton (2001) has a good discussion. On designing questionnaires ◦ Grosh and Glewwe (2000), Designing Household Survey Questionnaires for Developing Countries April 2020 JH: COURSE ON POVERTY MEASUREMENT 19
- What is absolute poverty
- Arrays in pascal
- Turing macine
- Multidimensional gradient
- Ssas multidimensional vs tabular
- Multidimensional analysis and descriptive mining of complex
- Index
- Multidimensional turing machine
- Mddb database
- Multidimensional reporting
- Multidimensional scaling - ppt
- Multidimensional or hypervolume niche
- Multidimensional space in data mining
- Multidimensional expressions
- Escalamiento multidimensional no métrico
- Processing multidimensional array
- Verilog array indexing
- Marketing multidimensional
- Multidimensional database definition
- Integrative approach to psychopathology