Chanda Bihari, goat farmer in Varanasi, India.
Measuring, Managing and Ending Poverty
Training on
Progress out of Poverty Index®
SPTF 2015, Siem Reap, Cambodia
Trainer - Muhammad Awais – Responsible Inclusive Finance Lead - Good Return
Skype: mawaisq | Email: awais@goodreturn.org
Grameen Foundation Solution Areas:
Insufficient &
inconsistent
income
Inability to
tolerate
shocks and
manage risks
Lack of
essential,
actionable
information
Needs are not
understood
and suboptimal
services are
delivered
Solution
Areas
Information
Services
Financial
Services
Progress out of Poverty Index® by
Grameen Foundation
Poverty Tools
and Insights
2
Good Return | World Education Australia
• Partner with:
–
–
–
–
Financial Service Providers
Microfinance Associations
Financial Sector Regulators
Private Sector
• Through partnership we seek to
develop innovative solutions for:
– Responsible and Inclusive Finance
(Digital Finance, Consumer
Protection & Poverty Measurement)
– Financial Capability and
Consumer Empowerment
– Smallholder Agriculture
• Programs in 8 Asia Pacific
Countries
Agenda
Session 2
Why poverty measurement is important
in microfinance?
Development goal by Microfinance institutions (MFIs) legal form – MIX
Social Performance Report 2009-10
Development goal by regions – MIX Social Performance Report 2009-10
Development goal and outcomes tracking – MIX social
performance report 2009-10
%MFIs
with goal
%MFIs
tracking
outcomes
Poverty
reduction
2009 Social Performance Report –
Pakistan Microfinance Network (PMN)
21 institutions reported for PMN 2009 Social
Performance Report
Social Performance Report 2009 of PMN
1. Poverty reduction is the second most popular development
objective
2. Very few institutions have a poverty assessment strategy in
place or have reliable data on the poverty profile of their
clients.
3. Institutions assume they are lending to the ‘poor’, so they are
meeting their objective.
Social Performance Report 2009 of PMN – Poverty
Tracking
1.
To reduce poverty, its vital to track poverty
2.
3.
Very few Institutions are tracking the poverty level of clients
Most institutions can’t report data on whether the program has
helped clients move above a poverty threshold
Reliable and actionable data on poverty is not being collected or
collated by institutions
Tools in use are subjective and make local or international
benchmarking difficult
Significant data gaps, especially in terms of tracking client
poverty
4.
5.
6.
Why to measure and report poverty?
1. Not to forget the poor but to keep them as part of the agenda
2. Be true to vision, mission and objectives [of poverty
alleviation]
3. Identification of poor to target interventions
4. Achievement of Millennium Development Goals (halve # <
$1/day)
5. Financial Institutions reporting to the Microcredit Summit
Campaign i.e. 100 million microfinance clients cross $1/day
line by 2015
6. Donors (report % poor, decide how to allot funds)
7. To make poverty outreach transparent and verifiable and thus
more explicit and intentional
8. Monitoring, evaluation and impact measurement of policies,
programs and institutions
Two approaches of poverty measurement
1. Subjective poverty measurement: The subjective assessment
of the wellbeing of an individual by himself/herself or by other
people. Subjective tools used by MFIs are Participatory Wealth
Ranking (PWR), Housing Index, etc.
2. Objective poverty measurement: The use of mainstream and
uniformly applicable concepts e.g. national poverty line, food
poverty line or $1/day. Tools include (in the microfinance
sector) Progress out of Poverty Index (PPI), Poverty
Assessment Tool (PAT), etc.
Methods of client poverty measurement in
microfinance
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
Loan size
Subjective
Income
poverty
Expenses
measurement
Net income
Participatory Wealth Ranking (PWR)
Housing Index
Land Index
Social criteria
Access to basic services
Empowerment (social, political, etc)
Poverty Assessment Tool (PAT)
Progress out of Poverty Index (PPI)
……………
(This is not an exhaustive list)
Objective
poverty
measurement
Q&A
Session 3
Understanding the Progress out of Poverty
Index (PPI)
Material for this session is based on
http://www.microfinance.com/English/Papers/Scoring_Poverty_Cambodia_2011_EN.pdf
developed by Mark Schreiner from Microfinance Risk Management L.L.C.
Introduction
Progress out of Poverty Index (PPI)
PPI for Cambodia is objective, country-specific,
and easy-to-use poverty scorecard. Pro-poor
programs can use one Poverty Scorecard that
applies to the whole county (both urban and
rural) to estimate the likelihood that a
household has a per-capita consumption below
a given poverty line.
Three basic uses of the PPI
1. Measure poverty - number or percentage of
households under a poverty line at a point in time
2. Track changes in the poverty of households over time
3. Segment [poor] households to target services
PPI do not measure impact!
Worldwide PPIs for 54 countries
Asia Pacific: 1.China; 2.Pakistan; 3.Indian; 4.Nepal; 5.Bangladesh; 6.Sri Lanka; 7.Cambodia; 8.Vietnam;
9.Indonesia; 10.Philippines; 11. Myanmar; 12. Timor-Leste; 13. Myanmar & 14. Fiji
Why use the PPI?
•
The direct approach to poverty measurement via survey is
difficult and costly. The 2011 Cambodia Socio Economic Survey
(CSES) runs on 58 pages.
•
Participatory Wealth Ranking (PWR) is subjective and relative.
Poverty measurement based on land ownership or housing
quality is blunt with unknown precision. These can’t be
consolidated and benchmarked.
•
The indirect approach (the approach of PPI) via poverty scoring
is simple, quick, and inexpensive. The PPI uses 10 nonfinancial but verifiable indicators.
What are Proxy Indicators?
2011 Cambodia
Socio Economic
Survey (CSES)
o By taking one
spoon, you confirm
if soup is properly
cooked.
o Taste of spoon
represent the
whole pot – No
need to drink the
whole pot!
o Spoon is proxy
of the whole pot!
PPI for
Cambodia –
A proxy of
CSES 2011
Features of the PPI
1. Objective: Based on national survey data:
1. Highest-quality expenditure measurement
2. Quantitative and observable indicators
2. Universal: For all poverty alleviation programs
3. Practical: Accepted and actually used:
1. Few indicators, inexpensive to collect
2. Simple enough to understand and compute on paper in the
field in real time without software
4. Known targeting accuracy
PPI Implementation
PPI
OF
A
HOUSEHOLD
IN
Cambodia
9
0
0
5
3
0
0
0
0
0
17
PPI Measures Poverty for:
Five olddefinition
Eight
governmentdefinition
Eight World
Bank-definition
100% of national
100% of national
100% of national
150% of national
150% of national
150% of national
200% of national
$1.25/day 2005 PPP
$2.50/day 2005 PPP
200% of national
Median
$1.25/day 2005 PPP
200% of national
Median
$1.25/day 2005 PPP
$2.00/day 2005 PPP
$2.50/day 2005 PPP
$2.50/day 2005 PPP
$5.00/day 2005 PPP
$5.00/day 2005 PPP
Poverty lines with similar name vary in values due to the
definition of poverty. For example
KHR4,081
100% national poverty line (World-Bank-definition)
KHR3,863
100% national poverty line (Government-definition)
KHR3,328
100% national poverty line (old-definition)
Example: Poverty Rate of a Client
Score
Household has a
score of 17
Household Poverty Rate (%):
Poverty likelihood – known bias
73.8 – (+1.1*) = 72.7%
100% National (World
Bank-definition)
Poverty
likelihood
for the
100%
national
poverty
line (World
Bankdefinition)
Known Bias for the World-Bank-definition poverty lines
*Figure 8 – page 165 of the http://www.microfinance.com/English/Papers/Scoring_Poverty_Cambodia_2011_EN.pdf
developed by Mark Schreiner from Microfinance Risk Management L.L.C.
Example: Change in the Poverty Rate of a Client
Score
After one
year the
same
household
has a score
of 32
100 National
(world Bank)
Poverty rate of the
household
decreased* by 39.5%
for the 100%
national poverty line
(World BankDefinition)
* [34.3 – (+1.1)] - [73.8
– (+1.1)] = - 39.5%
Estimating Group Poverty Rate
The estimated poverty rate of the group is the average poverty
likelihood of its members subtracting off known bias i.e.
Cambodia – Sample 3 clients – 1st March 2015
Client
Score 01/03/15
Poverty likelihood for 100% National Poverty Line
(World Bank-definition)
001
16
73.8
002
20
60.7
003
42
10.5
Average Poverty Likelihood (%)
48.3
[(73.8+60.7+10.5)/3]
Known bias*
+1.1
Group (average) Poverty Rate (%)
48.3 – (+1.1) = 47.2
Track Changes in the Group Poverty Rate
Cambodia – Sample 3 clients – 1st March 2015 and 1st March 2016
Client
Score 1/3/15
Score 1/3/16
Poverty likelihood for 100% National
Poverty Line(World Bank-definition)
1/3/15
1/3/16
001
16
20
73.8
60.7
002
20
43
60.7
10.5
003
42
47
10.5
5.5
48.3 – (+1.1) = 47.2
25.6 – (+1.1) = 24.5
Group (average) Poverty Rate (%) =
(Average poverty likelihood – known
bias)
(Average) poverty of the group dropped by 22.7%
PPI Fundamentals: Portfolio analysis on 1st Mar 2015
Score
For example, a
MFI has 4,000
clients
• 3,545 clients
have a score of
12 each
• 335 clients
have a score of
34 each
100% national
(World Bankdefinition)
No. of
clients
(A)
Poverty
Likelihood –
100% national
(WB-definition
(B)
# clients
below 100%
national (WBdefinition (A X
B)
3,545
88.6%
3,141
335
34.3%
115
120
0.0%
0
Total
3,256
Poverty Rate (%) of MFI before correction of bias:
(3256/4000) X 100 = 81.4%
Less bias (+1.1%)
• 120 clients
have a score of
66 each
Poverty Rate (%) of MFI after correction of
bias: 81.4 – (+1.1) = 80.3%
OR 3,212 out of 4,000 clients are below the 100%
national poverty line (World Bank-definition)
Portfolio analysis
Tracking changes over time
Suppose the same group of
clients, from the previous
example, is re-tested one
year later and the portfolio
poverty rate is 51.0% below
the 100% national poverty
line (World Bank-definition)
Total Clients
4,000
Year 1
3,212
Year 2
2,040
# Clients Below the
100% national
poverty line (World
Bank-definition)
80.3% 51.0%
Poverty Rate
(80.3% - 51.0%)
= 29.3%
Change in Poverty
Rate (%)
(3,212 - 2040) =
1,172
# Clients cross the
100% national
poverty line (World
Bank-definition)
PPI Exercise
0
5
10
15
min
PPI Construction
A set of 100 indicators are selected using past experience and common sense
2011 Cambodia
Social Economic
Survey (CSES)
58 pages/500
items
In the areas of
1. Household composition
2. Education
3. Housing
4. Ownership of durable assets
5. Employment.
1.
2.
3.
4.
5.
Inexpensive to collect
Easy to answer quickly
Simple to verify
Change over-time with a change in
poverty status
Applicable in all regions of Cambodia
These indicators are ordered by strength to predict poverty on its own
3,586 households (HHs) surveyed in 2011 CSES
PPI construction from the random half of the 3,586
HHs
The scorecard of
16 candidate
indicators is field
tested. This
scorecard is also
reviewed by
potential users in
terms of cultural
sensitivity, ease of
understanding,
ease of verification
and potential to
change over time.
A 10 indicators
scorecard is
finalized.
o One-indicator
scorecard
o Two-indicators
scorecard
o 16 candidate
indicators
scorecard
Final 10indicators
Scorecard
o Calculation of score
[range] of household[s]
o Household per-capita
consumption < a
poverty line
o Association of above
two
+
Likelihood Table
P
=P
I
o Random half of
the 3,586 HHs is
used for
validation
Adopted from the http://www.microfinance.com/English/Papers/Scoring_Poverty_Cambodia_2011_EN.pdf
developed by Mark Schreiner from Microfinance Risk Management L.L.C.
Adopted from the http://www.microfinance.com/English/Papers/Scoring_Poverty_Cambodia_2011_EN.pdf
developed by Mark Schreiner from Microfinance Risk Management L.L.C.
Data and Definitions of Poverty
Three Definitions of Poverty and Four Rounds of CSES
2004
CSES
2012
CSES
Three Definitions
of Poverty:
1. Old
2. Government
2011
CSES
o
o
o
3. World Bank
2009
CSES
Old-definition data exist only for 2004 and 2009
Government-definition data exist only for 2009, 2011 and 2012
World Bank-definition data exist only for 2004, 2009 and 2011
2004 all-Cambodia
Poverty line KHR1,852
Poverty rate (HH) 30.2%
Old-definition
poverty
Food standard
“Observed cost – for
people in the middle
quintile of expenditures
– of a reference food
basket providing 2,100
calories”
“Sum of minimum
consumption
standards for food
and non-food items”
Non-food standard
2009 all-Cambodia
Poverty line KHR3,328
Poverty rate (HH) 11.7%
“Observed non-food
consumption of
households whose
total consumption is
at food line”
Government-definition poverty
“Sum of minimum consumption standards
for food and non-food items”
Food standard differ from old-definition:
o Use of a food basket of 2,200 calories
o Derives food-basket quantities from data on
people in the 5th to 30th percentiles of total
consumption
o Prices food items in a different way
o Sets the non-food standard as the average nonfood consumption of people in the third decile
o Adds a token value for clean water (rather than no
such addition)
2009 all-Cambodia
Poverty line KHR3,863
Poverty rate (HH) 19.2%
Non-food standard differ from old-definition:
The measure of consumption excludes:
o Use-value of owner-occupied housing and other
non-rental arrangements
o Luxury/vice items that the poor rarely consume
2011 all-Cambodia
Poverty line KHR4,399
Poverty rate (HH) 15.9%
2012 all-Cambodia
Poverty line KHR4,540
Poverty rate (HH) 15.2%
Adopted from page 20-33 of the http://www.microfinance.com/English/Papers/Scoring_Poverty_Cambodia_2011_EN.pdf
developed by Mark Schreiner from Microfinance Risk Management L.L.C.
World-Bank-definition poverty: “Sum of
minimum consumption standards for food
and non-food items”
World-Bank-definition differ from old-definition,
it uses 2,200 calories standard. Other than this
World-Bank-definition resembles to olddefinition.
2004 all-Cambodia
Poverty line KHR2,512
Poverty rate (HH) 47.4%
2009 all-Cambodia
Poverty line KHR4,081
Poverty rate (HH) 20.1%
2011 all-Cambodia
Poverty line KHR4,637
Poverty rate (HH) 16.4%
Adopted from page 20-33 of the http://www.microfinance.com/English/Papers/Scoring_Poverty_Cambodia_2011_EN.pdf
developed by Mark Schreiner from Microfinance Risk Management L.L.C.
PPI Measures Poverty for:
Five olddefinition
Eight
governmentdefinition
Eight World
Bank-definition
100% of national
100% of national
100% of national
150% of national
150% of national
150% of national
200% of national
$1.25/day 2005 PPP
$2.50/day 2005 PPP
200% of national
Median
$1.25/day 2005 PPP
200% of national
Median
$1.25/day 2005 PPP
$2.00/day 2005 PPP
$2.50/day 2005 PPP
$2.50/day 2005 PPP
$5.00/day 2005 PPP
$5.00/day 2005 PPP
Poverty lines with similar name vary in values due to the
definition of poverty. For example
KHR4,081
100% national poverty line (World-Bank-definition)
KHR3,863
100% national poverty line (Government-definition)
KHR3,328
100% national poverty line (old-definition)
Estimates of Poverty Likelihood
Score range from 0 (most
likely < poverty line) to 100
(least likely < poverty line)
A score of 25 has a
poverty likelihood of
46.6% because 3,664
out of 7,867
households (46.6%) in
2011 CSES - who score
25-29 - have
consumption
below the World Bankdefinition national
poverty line
Likelihood of a Household to Fall between two Adjacent Poverty Lines
23.3% - 7.5% = 15.8%
Targeting
Nontargeted
Clients
Of
MFI
Targeting cut-off
100% national
poverty line (World
Bank-definition)
Targeted
Targeted
Non-targeted
Leakage
Exclusion
At and above poverty line
mistakenly targeted
At and above poverty line
correctly non-targeted
Above
poverty
line
Targeting cut-off
100% national
poverty line (World
Bank-definition)
Below
poverty
line
Inclusion
Under-coverage
Below poverty line
correctly targeted
Below poverty line mistakenly
non-targeted
How to set poverty cut-off?
Hit Rate =
1
X
HHs correctly included (Inclusion)
-
0
X
HHs mistakenly under-covered (Under-coverage)
-
0
X
HHs mistakenly leaked (Leakage)
+
1
X
HHs correctly excluded (Exclusion)
Using 3rd
column set
a cut-off to
achieve a
desired
poverty
rate.
Measuring changes in poverty rates over time
using the old definition of poverty with the old
2004 and new 2011 PPI
Use Revised PPI (based on 2011 CSES)
from this point on!
To Compare Poverty Rate Between Old PPI
and New PPI, use the (Similar) Poverty Line of
Old-Definition!
PPI is updated
2013
01.03.2013
Poverty measurement
of 3 HHs using 2009
PPI
Same PPI
2015
01.03.2015
Poverty
measurement of 3
HHs using 2014 PPI
2016
01.03.2016
Poverty
measurement of 3
HHs using 2014 PPI
Portfolio analysis using the 100% national poverty line
PAST (1st Mar 2013)
2009 PPI
HH
Now (1st Mar 2015)
2015 PPI
Future (1st Mar 2016)
2015 PPI
Score
Pov. Like.
(%)
Score
Pov. Like. (Olddefinition) (%)
Pov. Like.
(WB-Def.) (%)
Score
Pov. like. (WBdef.) (%)
1
15
56.1
21
30.4
60.7
22
60.7
2
20
45.3
26
20.8
46.6
27
46.6
3
25
34.3
31
14.5
34.4
37
20.2
Bias
-0.8
-
-0.7
+1.1
-
+1.1
Poverty Rate (%)
46.0
-
22.6
46.1
-
41.4
Estimated change between:
Past and now (hybrid)
22.6 – 46.0 = - 23.4 percentage points
Now and future (non-hybrid)
41.4 – 46.1 = - 4.7 percentage points
Past and future (grand spliced)
- 23.4 + (- 4.7) = - 28.1 percentage points
PPI is updated
2013
01.03.2013
Poverty measurement
of 3 HHs using 2009
PPI
Same PPI
2015
2016
01.03.2015
01.03.2016
Poverty measurement
of 4 HHs (3 old plus 1
new) using 2014 PPI
Poverty measurement
of 4 HHs (same old 4)
using 2014 PPI
Portfolio analysis using the 100% national poverty line
HH
PAST (1st Mar 2013)
2009 PPI
Score
Poverty
Likelihood
Now (1st Mar 2015)
2015 PPI
Future (1st Mar 2016)
2015 PPI
Sco
re
Pov. Like.
(Olddefinition)
(%)
Pov. Like.
(WB-Def.)
(%)
(3 HHs)
Pov. Like.
(WB-Def.)
(%)
(4 HHs)
Score
Pov. like. (WB-def.)
(%)
(3 HHs)
Pov. like. (WBdef.)
(%)
(4 HHs)
21
30.4
60.7
60.7
22
60.7
60.6
(%)
1
15
56.1
2
20
45.3
26
20.8
46.6
46.6
27
46.6
46.6
3
25
34.3
31
14.5
34.4
34.4
37
20.2
20.2
4
-
-
35
-
-
20.2
41
-
10.5
Bias
-0.8
-
-0.7
+1.1
+1.1
-
+1.1
+1.1
Poverty Rate (%)
46.0
-
22.6
46.1
39.3
-
41.4
33.3
Estimated change between:
3 HHs: Past and now (hybrid)
22.6 – 46.0 = - 23.4 percentage points
3 HHs: Now and future (non-hybrid)
41.4 – 46.1 = - 4.7 percentage points
3 HHs: Past and future (grand spliced)
4 HHs: Now and future (non-hybrid)
- 23.4 + (- 4.7) = - 28.1 percentage points
33.3 – 39.3 = - 6.0 percentage points
Practical guidelines for PPI Use
Practical Guidelines for the PPI Use
o Main challenge in the scorecard design is not to achieve
maximum statistical accuracy but to increase the
chances of its usage;
o For the PPI, statistical accuracy is balanced with
simplicity, ease-of-use and face-validity;
o PPI score can be computed in the field because it has:
Only 10 indicators
Only multiple-choice indicators
Only simple points – non-negative integers and no
arithmetic beyond addition
o The PPI is readily available to be used;
Practical Guidelines for the PPI Use
o Training, monitoring and evaluation are key to get high
quality PPI data;
o MFI to consider followings for the PPI implementation:
Who will do the scoring
􏰀How scores will be recorded
􏰀What participants will be scored
􏰀How many participants will be scored
􏰀How frequently participants will be scored
􏰀Whether scoring will be applied at more than one
point in time
􏰀Whether the same participants will be scored at
more than one point in time
Practical Guidelines for the PPI Use
o PPI data can be collected in the field by:
Staff
Third parties (community, professional enumerators,
etc.)
A combination of above two
o PPI data can be collected from:
All relevant participants (a census)
􏰀 A representative sample of relevant participants
􏰀 All relevant participants in a representative
sample of relevant branches
􏰀 A representative sample of relevant participants
in a representative sample of relevant branches
Practical Guidelines for the PPI Use
o The frequency of the PPI data collection can be:
 As a once-off project (precluding measuring change)
 􏰀 Every two years (or at any other fixed or variable time interval,
allowing measuring change)
 􏰀 Each time a field worker visits a participant at home (allowing
measuring change)
o To measure change using the PPI
 􏰀Score a new, independent sample, measuring change across samples
 􏰀 Score the same sample at both baseline and follow-up
o PPI data can be collected through:
 􏰀Paper in the field, and then filed at a central office
 􏰀 Paper in the field, and then keyed into a database or spreadsheet at a
central office
 􏰀 Portable electronic devices in the field, and then uploaded to a
database (e.g. using ODK, TaroWorks, etc)
Thank You
PPI CERTIFICATION
What is PPI certification?
o A confirmation that PPI data is reliable and
accurate meeting PPI standards of use
o A recognition of excellence in poverty
measurement with the PPI
What are PPI Standards of Use?
1. Basic Standards of Use: Best practices to
accurately report the PPI data
2. Advanced Standards of Use: Best practices
to integrate the accurate and reliable (PPI) data
into operations
Basic PPI Standards of Use
(report reliable and accurate data)
An organization must meet all of the Basic
Standards to qualify for Basic PPI Certification.
Basic standards cover:
A.
B.
C.
D.
E.
F.
G.
Commitment to poverty measurement
Training
Collection
Data storage and validation
Data analysis
Data use
Reporting
Group Work
1. What standards are not clear?
2. What are most challenging standards?
A. Commitment to poverty measurement
1. The board demonstrates commitment to poverty
measurement data, requiring that management be
responsible for its tracking and requesting periodic updates
on poverty data.
2. Senior management responsible for the PPI understands
the PPI, can explain why the organization implements it, and
monitors its implementation.
3. A written plan or detailed manual for PPI implementation is
in place and includes a clear design of the process.
B. Training
4. The current PPI project manager has been trained on the
PPI, either by an external party or by someone with
significant experience using the PPI within the organization.
5. Staff responsible for surveying have been trained on and
understand the precise meaning of the PPI questions and
responses. In addition, they are informed when data
validation is in place to verify responses to surveys they
collect.
6. If PPI data are manually entered into a database, data entry
staff are trained on and understand the PPI data entry
process.
7. Staff involved in PPI data analysis are trained in accurate
analysis and reporting guidelines.
C. Collection
8. PPI data are collected on a census basis or, if sampling is
used, with samples random and representative of the
population(s) to be analyzed.
9. PPI surveys are completed with the date the survey is
administered, identifying information of the enumerator,
identifying information of the client (such as a unique client
ID number), answers to each question and a final score.
10. The PPI survey and look-up tables used do not deviate from
the original country-specific PPI as found on
Progressoutofpoverty.org or microfinance.com. If a
translated version is used, the translation is in writing and
either matches the national household expenditure survey or
has been professionally translated if no official translation is
available.
C. Collection
11. Enumerators ask each PPI question in a manner reflective
of the original meaning of the question. The best approach
to achieve this is to ask the question exactly as it is
written. Where doing so causes confusion or is otherwise
problematic, the enumerator either visually verifies an
indicator or rephrases the question in a way that does not
distort the meaning of the question. Enumerators carry a
copy of an interview guide with them during survey
collection.
12. The latest available version of the PPI, including the
scorecard and the look-up table, is in use within one year
of release.
D. Data storage and validation
13.Before PPI data is analyzed, a quality control check of
the accuracy of PPI collection occurs by someone other than
the original enumerator to verify that (a) the interview
actually took place and (b) the responses match those
originally recorded. A random and representative sample of
at least 30 or a number representing at least 5%, whichever
is greater, of collected PPI scorecards is reviewed for each
data collection or on at least a quarterly basis. A review
process is in place to identify and correct the source of error
if checked surveys significantly deviate from the original data
collected. A discrepancy rate is calculated by averaging the
percentage of inconsistent responses of each PPI validated.
Batches with a discrepancy rate above 10% should not be
used during analysis until all PPIs have been reviewed.
D. Data storage and validation
14. PPI data are secured to prevent unauthorized access.
15. All collected PPI data, including client scores and look-up
tables, are centrally stored in an electronic manner that
permits analysis. Furthermore, the PPI version used is
clearly tracked in the database. A basic system like a
spreadsheet is acceptable.
16. Data are reviewed and cleaned before analysis.
17. Original PPI surveys, including answers to each question,
are stored, either in physical or electronic form, for backchecking for at least one year.
D. Data storage and validation
18.Before PPI data is analyzed, the PPI data entry process is
checked for accuracy. A random sample of at least 30 or a
number representing at least 5%, whichever is greater, of
encoded PPI scorecards is reviewed. A review process is in
place to identify and correct the source of error if checked
surveys significantly deviate from entered data. A discrepancy
rate is calculated by averaging the percentage of inconsistent
responses of each PPI validated. Batches with a discrepancy
rate above 10% should not be used during analysis until all PPIs
have been reviewed. This standard is not applicable to
organizations that enter PPI data directly into a database
system, e.g., with a mobile data collection tool like TaroWorks.
E. Data analysis
19. If an updated PPI has been adopted after use of an earlier
version and the organization plans to track poverty
movement or compare results from different versions, the
organization is correctly comparing such results according to
the appropriate case:
I. Green reset: Poverty likelihoods are compared across
versions without complication.
II. Yellow reset: Legacy poverty lines developed
specifically for calculating hybrid estimates of change are
used to compare results to a previous PPI. Newdefinition poverty lines should be used to report
outreach.
III. Red reset: The previous PPI must continue to be
collected for comparison purposes while the updated PPI
must be collected to report outreach.
E. Data analysis
20. Computation of the percentage of clients or customers below
a given poverty line is accurately calculated.
20. PPI scores are not used during data analysis.
F. Data usage
22. Poverty data are benchmarked using objective regional
rates, peer benchmarks or internally set benchmarks.
G. Reporting
23. Poverty lines used for analysis align with the objectives of
the organization and/or project. For example, if an
organization strives to reach only the extreme poor, use of a
poverty line that is too high will obscure whether the
organization is reaching the extreme poor and offer less
insight into the population served relative to the
organization’s objective. Furthermore, these poverty lines
are understood by the PPI project manager. Finally, the
organization does not unintentionally or otherwise
misrepresent its poverty outreach by using a poverty line
significantly higher or lower than one appropriate for a stated
target, if such a target exists.
G. Reporting
24. Recent PPI results are reported to the board and senior
management with a frequency appropriate to the PPI
implementation plan (e.g., upon completion of a one-time
data collection or on a quarterly basis for continual data
collection). Reported results are timely, with underlying data
taking no more than four months to be included in a report
for the first time. Reports must indicate the rate of poverty of
incoming clients (i.e., concentration), as well as
benchmarked results.
25. PPI results, particularly when displayed in graphs, include
the number of PPIs analyzed, the poverty line(s) used for
analysis, the dates or time period in which data collection
occurred, and the population represented, e.g., rural
branches, all regions, incoming clients for year 2014, and all
clients.
Advanced PPI Standards of
Use
(visit the website:
http://www.progressoutofpoverty.org/sites/default/files
/Advanced%20Standards%20v2.0.PDF )
Steps to PPI certification
1. Download the PPI Standards of Use at
progressoutofpoverty.org/standards.
2. Gauge certification readiness with the PPI SelfAssessment Tool, available at
progressoutofpoverty.org/resources.
3. Contact a rating agency.
4. Submit the documentation required by the rating agency
and facilitate an on-site evaluation led by rating agency
representatives.
5. If needed, improve your organization’s PPI practices
based on rating agency feedback.
6. Certification is awarded at either the Basic or Advanced
level at the rating agency’s discretion.
Q&A
The Challenges of the PPI
The challenges
•
Simplicity of PPI is difficult to accept for some people.
•
Can poverty be measured with 10 questions? PPI consist
of 10 proxy indicators which have values with scores Plus
poverty look-up tables. PPI is NOT just 10 questions.
•
The gaps between PPI results and the results from MFI’s
own poverty measurement tools. Where does MFI’s
poverty tool fits in terms of poverty line(s)? Comparing
apples with apples.
•
Strong ownership in MFI for its own poverty
measurement tool although it knows that the responses
are inflated and the results are not accurate due to the
subjectivity and non-verifiable nature of the indicators.
The challenges
•
PPI is developed from old (2 or more years) national
socio-economic survey. So people think that PPI is not
reflective of the current situation. Govt. is also using the
results of the latest national survey means it is still highly
relevant.
•
PPI doesn’t fit to all extreme situations. Sur, it doesn’t. Its
not perfect like any other tool.
•
Top management and Board lack understanding of PPI.
•
Social performance is just a talk and not the walk.
•
MFI lacks capacity to interpret PPI results.
•
How does PPI make a business case?
Q&A