Efficiency, equity and feasibility of strategies to
identify the poor: an application to premium
exemptions under national health insurance in
Ghana
Caroline Jehu-Appiah, Genevieve Aryeetey, Ernst Spaan,
Irene Agyepong, Rob Baltussen
Health Policy 2010;95:166-73
Saly, Senegal 2011
Background
• Currently many sub-Saharan African countries are
exploring ways to replace user fees at point of service use
with more equitable alternatives.
• Many are experimenting with social and community
health insurance
• Regardless of what financing mechanisms are used,
exemptions are needed for the poorest
• However there are challenges with effectively identifying
and targeting the most vulnerable groups for
exemptions.
Background cont’
• Ghana passed a National Health Insurance (NHI) act in
2003 and by 2010 66% enrolment (NHIA)
• Empirical evidence shows enrolment among the poor is
low (Asante and Aikins 2008).
• Efforts to identify and enrol the poor through premium
exemptions at an early stage of the NHIS are required.
• Difficulties in identification and targeting of the poor
and uncertainties as to the most cost-effective
approaches to use.
Objectives
• To identify potential strategies to identify the
poor, and assesses their feasibility, efficiency
and equity
▫ Estimate costs of each strategy, and present tradeoffs between
 feasibility (defined as practical ability to identify the
poor in a given context),
 efficiency (defined by cost per exempted poor
individual, and
 equity (defined by error of exclusion)
Methods
• A literature search in Medline. 62 articles selected and classified into 4
broad strategies to identify the poor: (1) means testing, (2) proxy means
testing (PMT), (3) geographic targeting (GT) and (4) participatory
welfare ranking (PWR).
• To estimate the implementation costs, cost models were developed on
the basis of a combination of empirical estimates and expert opinion,
using the 2008 price levels and costs and errors of in- and exclusion of
these strategies
TC  Survey Cost  P0 x Peligible x Pr emium
• Sensitivity analysis was employed to assess the impact of varying
assumptions on study results and study conclusions
Identifying the Poor by Household Income Means Testing (LSMS)
Pros of LSMS’s
• Benchmark in poverty
assessment
• Objective , high quality
assessment of HH income
• Both absolute and relative
poverty analysis of welfare
Cons of LSMS’s
• High cost
• Measure income defined
poverty and not its broader
dimensions
• Measure HH and not
individual welfare
• Do not disaggregate beyond
regional level
• Small sample size
Targeting the Poor by Household Indicators - Proxy
Means testing (DHS,CWIQ)
Pros
• Potential to provide alternative
welfare ranking
• Lower administrative costs
and information on the key
indicators is more widely
available.
• Effectiveness-correctly
predicts poverty status from
80-84% of its participants
(Johannsen 2006).
Cons
• Limited to relative analysis of
welfare
• Exclusion of 16-20% of poor
(Johansen,2006)
• Asset indices say nothing
about absolute poverty
• Cannot be used to monitor
changes in poverty over time
• Measured at HH and not
individual level
Geographic targeting (poverty maps)
Pros
Cons
• Relatively easy to implement
• Leakages to non poor
especially in urban areas
• Possible to map at the district,
sub district levels
• Data availability
• Narrow targeting improves
coverage of poor
• Requires High level of
Econometric Expertise
• Village level targeting more
effective than regional
targeting at reducing leakages
Poverty Mapping
• Maps out poverty
incidence for whole
country
• Allows for Blanket
exemptions of whole
districts based on
incidence of poverty
• At no cost to MOH
Targeting by Participatory Welfare Ranking
(PWR)
Pros
• Simple, transparent
Cons
• Low cost
• Ineffectieve in larger
communities urban areas
• Widely accepted – community
participation and ownership
• High level of skill and
facilitation
• Effectiveness- 82%
• Risk of sampling and
respondent bias
• Useful in combination with
Geographic targeting
• Captures new settlements, street
children, ophans
• Criteria differs from
community to community
therefore results are not
comparable
Gt. Accra Region
12.000
10.000
ThousandS $
8.000
PWR
6.000
4.000
PROXY MEANS
TESTING
2.000
0
Accra Tema
Ga Dangbe Dangbe
West East
GEO
TARGETING
Thousands $
Upper West Region
1.600
1.400
1.200
1.000
800
600
400
200
0
PWR
PROXY MEANS
TESTING
GEO
TARGETING
District
Nadawilli
Strategy Costs
Cost of
survey
(g)
Total Cost Cost per
(j)=(g)+(h) identified
poor
individual
(efficiency
indicator)
(j)=(i)/(f)
incremental
cost per
extra
exempted
poor
individual
PWR
45,799
356,566
402,365
11.63
PMT
70.491
404,930
475,421
12.57
23
536,189
536,189
11.63
7
GT
Tema
Costs of
premium
exemptions
(h)=(f)*φ
PWR
239,855
1,276,787
1,516,642
41.44
PMT
407,393
1,343,473
1,750,866
43.76
69
3,252,962
3,252,962
66.67
171
GT
Discussion
• PMT, PWR and GT achieve efficiency and equity objectives to different
degrees
• PWR appears the least costly and therefore most efficient strategy, but is
also the least equitable.
• GT covers all (poor) individuals in a given area, and is therefore the most
equitable but also the most costly.
• Incremental costs of exempting one extra poor individual range between
US$7 and US$ 171.
• Choice highly dependent on poverty setting and feasibility of
implementation
• Selection of a strategy therefore has to be contextualised, and it is not
advisable to apply a single strategy across the entire country.
Policy dilemma
• TC of paying for insurance premiums exemptions for all
poor households in Ghana is $22 million
• These costs consume 4% of total health resource
envelope for 2008 ($513 million) and
• 16% of total NHI budget ($142 million)
• Policy dilemma and tough choices posed by the gap
between the desired mandate and financing constraints
Study limitations
• Analysis is based on literature review and expert
opinion
• Research is needed to verify assumptions
• Partial assessment of various strategies and
excludes the perceptions of the community and
policy makers on the acceptability of various
approaches
• Thank you