LINKING PRACTITIONER POVERTY MEASURES TO INTERNATIONAL POVERTY LINES

1
LINKING PRACTITIONER POVERTY MEASURES TO
INTERNATIONAL POVERTY LINES

• Small Enterprise Foundation
• University of Natal

2
Presentation Overview

• Summary of SEFs approach, definition and
  measurement of poverty
• Description of the Participatory Wealth Ranking
  Method (PWR)
• Summary of the poverty assessment of SEF and the
  relationship between the Poverty Assessment Tool
  (PAT) and the PWR
• Present preliminary work in modeling the PAT to
  national income and expenditure surveys
• Explore observable indicators of poverty which
  serve as good proxies for money metric measures
• Match these indicators to national income and
  expenditure surveys

3
SEFs MISSION

• To work aggressively towards the elimination of
  poverty by reaching the poor and very poor with a
  range of financial services to enable them to
  realise their potential
• There are many levels of poverty. Most poverty
  reduction measures dont reach the poorest, but
  the middle-poor. We need to actively target the
  poor.

4
Participatory wealth ranking process
3 Facilitators for half a day for an average
village of 600 households

• Mapping the village
• List of households
• Set up reference groups
• Card sorting and wealth
• ranking
• Triangulation
• Scoring
• Selecting the poorest

6 Facilitators for one day with 3 reference
groups per 100 households
Half hour in office
5
Characteristics of the very poor
Group 1The very poor

• Children don't go to school
• Lack of food
• Poor shelter - many people in one room
• Unemployed
• Lack of clothes to wear
• Beggar/ must hunt food/ hand outs
• Big families
• Don't meet families basic needs
• Deserted by husbands
• No source of income/ no-one else to help
• Struggles for food and shelter, no thought of
  luxuries
• Children are dirty

• No opportunities to achieve anything
• Uncertainty eg. food for tomorrow
• Works for other village people
• Polygamous families
• No assets
• Unmarried daughters with children
• Bad health
• Many children to support
• Pensioner with very big responsibilities
• Christmas is like any other day
• Get food by ploughing
• Children steal from others - especially food

6
Characteristics of the poor
Group 2 Very poor, but a bit better than the
poorest

• Casual labourers - farm/domestic
• Pensioner with big responsibilities
• Buy some food daily not good food
• Can't satisfy their needs - money doesn't last
• Working but not earning enough
• Have shelter - bit better than poorest
• Poor clothes - but better than poorest
• Problems sending children to school - often dont
  go beyond primary level
• Small businesses

• Lot of credit
• Don't have much things only OK when have job
• Problems sending kids to school
• Problems with emergencies
• Lot of children to support
• Deserted by husbands
• Big families
• Better house, but owner died/ redundant - so
  conditions have gone down
• Better clothes than the poorest
• Better shelter - especially during the rains

7
Characteristics of the poor

• Group 4 People who are not poor
• better off rich
• Professionals / Government workers
• Have cars
• Run businesses
• Better housing
• Good jobs/ secure jobs
• Herds of cattle
• Good food
• Afford luxuries/ don't lack money
• Good clothes
• Can afford to educate to tertiary level
• Hire other people for their business/ at home
• Everyone has a bed
• Look after families with no hardships
• Children go to better schools
• Shop owners
• Well off with little responsibilities

• Group 3 Poor
• Dreams of luxury
• Can buy basic food and other basic needs
• Have businesses
• Pensioner with little responsibilities
• Employed, but low salaries
• Children go to school and have uniforms, but
  cant afford tertiary education
• Manage to support families
• Employed, but low income
• Poor income, but little responsibilities
• Better housing
• Fewer children to support
• Some clothes

8
Output of PWR

• Map of village
• List of households
• Poverty score for each household
• Descriptive poverty definition for each rank
• Cut off line for the very poor
• Eligible for the TCP program

9
Can practitioner poverty measures be linked to
national income and expenditure measures?

• Extent to which households categorized as poor
  using PWR are also categorized as poor using the
  Poverty Assessment Tool (PAT)
• Extent to which households in national surveys
  categorized as poor using the components of PAT
  are categorized as poor using money metric
  measures
• Components of local perceptions of poverty that
  determine the position given in the PWR, which
  are measurable may be matched to data contained
  in national surveys PAT
• Extent to which households in PAT national
  surveys that share these characteristics are
  categorized as poor using a money-metric approach
  the PAT indicators
• Community wide characteristics that influence the
  extent to which a limited easily observable
  indicator corresponds with levels of welfare
• Extent to which poverty thresholds are meaningful?

10
Brief overview of the Poverty Assessment of SEF -
2000

• 1. Compare the poverty profiles - programme
  participants and a control group
• 2. Contrast poverty levels between the clients
  drawn from the poverty targeted (TCP) and
  non-poverty targeted sub-group (MCP)
• 3. Assess the validity of proxy indicators in
  place of money metric measures

11
The distribution of poverty scores of clients and
non-clients in the TCP and the MCP
12
Conclusions of the SEF Poverty Assessment

• Research Findings
• The poverty targeting strategy used in TCP
  central to ensuring poverty outreach
• Small loan sizes (the self-targeting method used
  in MCP) did not deter the non-poor
• Self-targeting in this context, is an inadequate
  mechanism to attract the poor.
•  

13
Match between the Poverty Assessment (PAT) and
the Participatory Wealth Ranking (PWR)
14
PAT as a proxy for money metric measures of
poverty
Cumulative Frequency Distributions for Poor and
Non-Poor cohorts (defined using total household
expenditure per capita)  
15
Percent breakdown by poverty tercile, Northern
province and the Rest of South Africa
60
50
40
30
20
10
Percent
Northern Province
0
Rest of South Africa
highest
middle
lowest
POVGROUP
16
Human Development Index
17

Drivers of local perceptions of poverty in the
PWR

• Aggregated per person household assets
• Per person value of appliances
• Percent of adults who did not attend school
• Source of drinking water
• Roofing material
• Correlation is significant at the 0.01 level
  (2-tailed)

18

19
PWR poverty quintiles by percent of adults
in hh without education
n199
14
12
10
8
Mean off hh no school
6
4
2
5
4
3
2
1
PWR quintiles
20
Shared characteristics of poverty between the
PAT, PWR and the National Income and Expenditure
Survey 2000

• Percent of adults in households who did not
  attend school
• Source of drinking water
• Roofing material
• Quality of exterior walls
• Unemployment dependency ratio
• Correlation is significant at the 0.01 level
  (2-tailed)

21
(No Transcript)
22
Total household income quintiles
by quality of external walls
50
40
mud/plastic/asb
30
iron
20
wood
10
concrete
Percent
0
bricks
5
4
3
2
1
Total household income in quintiles
Weighting using 1996 Census estimates, Simkins,C
2002
Income and Expenditure Survey 2000
23
Spatial distribution of the population
by mean percent of hh members no school
16
14
12
10
8
Mean hh - no school
6
4
2
0
G rural
M-rural
LP-rural
FS rural
G Urban
NC rural
EC rural
M -urban
NW-rural
LP-urban
FS urban
KZN-rural
NC urban
EC urban
WC -rural
NW urban
WC-urban
KZN -urban
Rural - urban divide in provinces
Weighting using 1996 Census estimates, Simkins,C
2002
Income and expenditure Survey 2000
24
Spatial distribution of the population
by mean annual hh income
80000
70000
60000
50000
Mean annual hh income
40000
30000
20000
10000
0
M-rural
G rural
FS rural
LP-rural
G Urban
NC rural
EC rural
M -urban
LP-urban
NW-rural
FS urban
KZN-rural
NC urban
EC urban
WC -rural
NW urban
WC-urban
KZN -urban
Rural - urban divide in provinces
Weighting using 1996 Census estimates, Simkins,C
2002
Income and expenditure Survey 2000
25
Conclusions

• Individual indicators only partially explain
  persistence, community wide indicators also
  important
• The PWR and PAT apply broad based definitions and
  have been shown to be robust measures of poverty
  yet they are relative measures
• Linking these measures to national and
  international poverty lines requires both
  critical assessment of absolute poverty
  thresholds relative measurement
• One possible avenue is an asset index which would
  address the causes of poverty

26
Conclusion

• PAT and PWR classify some 70 percent of
  households in the same way in terms of their
  level of welfare. This overlap tends to be
  stronger at the poorer end of the distribution,
  reaching 75 percent of those categorised as being
  poor. PWR tends to be more conservative in
  identifying households as poor compared to PAT.
• When a similar indicator is constructed in a
  national data base, households classified as poor
  using the PAT indicator are also classified as
  poor using a conventional money metric measure
  based on income or expenditure and a poverty line
  or threshold.
• Since PWR poor households match PAT poor
  households, and PAT poor households match poverty
  line poor households, the majority of PWR poor
  households are also poverty line poor households.