Zambia’s Social Cash Transfer Programme
Zambia is a lower-middle-income nation in Southern Africa, with a population estimated at 15.5 million people in 2015. Over half
(58.2 per cent) of its population is located in rural areas, often at low density and following sparse patterns of territorial occupation
that accentuate challenges related to access to public services and markets. In terms of food consumption, 54.4 per cent of the
country is living below the poverty line, with 13.6 per cent living below the extreme poverty line (CSO 2015). The country’s flagship
government-run social protection programme established to tackle this problem is the Social Cash Transfer (SCT), an unconditioned
cash transfer programme.
The SCT has undergone several design changes since its inception in 2003, and this Policy Research Brief provides a brief overview of
those changes in terms of the programme’s targeting, benefit structure and governance, and the findings of impact evaluations between
2003 and 2016.
Since its creation, the programme has had three very different formats. Initially, between 2003 and 2010, it comprised five pilots with
limited connections among them. The design followed the ultra-poor approach (also known as the 10 per cent inclusive model or IM),
since it aimed to cover the poorest 10 per cent of the population of the districts served.
Subsequently, between 2010 and 2014, the programme had two different streams, each with different characteristics. One of them,
called the Child Grant (CG), specifically aimed to benefit households with children, whereas the Multiple Category Transfer Grant
(MCTG) targeted other forms of vulnerability.
Finally, since 2014 the programme has been operating in a third format, known as harmonised targeting because it has established a
single selection criterion (households with high dependency ratios) with the aim of reaching out to different sorts of vulnerable households.
This Policy Research Brief starts by discussing the differences in the targeting mechanisms and coverage figures of the programme’s
three stages. The following section discusses how the benefit structures have varied in the three stages of the SCT, and then there
is a discussion of the different governance structures at each stage. Next follows an analysis of the findings of impact evaluations
for the stages of the SCT that have been subjected to evaluation (i.e. the ultra-poor model between 2003 and 2010 and the CG and
MCTG models between 2010 and 2014). Finally, the text concludes by summing up the core differences due to modifications of the
programme’s design that characterised its three stages, and also by analysing the extent to which impact evaluations and other forms
of assessment might have contributed to changes in its design over the years.
2 Targeting and coverage
The first format of the SCT, used between 2003 and 2010, had a targeting goal of covering the poorest 10 per cent of people in
the districts in which it operated. At that time, beneficiary households were selected solely through community-based selection
mechanisms, whereby community organisations ranked households in terms of their poverty level, albeit without any objective or
standardised criteria to actually measure poverty levels (Habasonda 2009; Chiwele 2010). The sole exception to this methodology
was the Katete pilot, which targeted elderly people and worked somewhat as a universal pension scheme (Habasonda 2009).
For the Kalomo, Kazungula, Chipata and Monze pilots, the sole benchmark guiding social workers to select the poorest
households was a call to prioritise the enrolment of ‘destitute’ and ‘incapacitated’ or ‘labour-constrained households’. Although
many policy and programme documents illustrated these conditions by listing several situations that would meet these criteria,2
these lists were not exhaustive nor ranked the many situations in terms of which were the most pressing or took greater
precedence in terms of eligibility for the programme (Garcia and Moore 2012; Hamonga 2006; Permanent Mission of the
Republic of Zambia in Geneva 2009).
The International Policy Centre for Inclusive Growth is jointly supported by the United Nations Development Programme and the Government of Brazil.
research
brief June 2018 ISSN 2358-1379
62
2 International Policy Centre for Inclusive Growth
As mentioned above, between 2003 and 2010 the SCT was not
a consistent and homogeneous programme with country-wide
coverage. Rather, it comprised five pilots. Indeed, in a sense, that
stage could even be seen as some sort of pre-history of the SCT
(i.e. a stage marked by pilots that led to the institutionalisation of
the SCT at a later stage).
The districts in which pilots were implemented were chosen
based on political considerations, the interest of donors and
the imperative of experimenting in different contexts—from
extremely difficult-to-reach rural areas, such as the Kazungula
district, to urban contexts, such as Chipata, passing through
predominantly rural areas such as those of Kalomo, Monze and
Katete (Garcia and Moore 2012). The first district to receive the
intervention, in 2003, was the Kalomo district. Between 2005 and
2007 the SCT was expanded to four other districts: Kazungula
(2005), Chipata (2006), Monze (2007) and Katete (2007) (Garcia
and Moore 2012; Michelo 2015; van Ufford et al. 2016; Monteiro
Costa, Gyoeri, and Soares 2016; Habasonda 2009; Chiwele 2010;
Tembo et al. 2014; Chimai and Mulenga 2014).
A 2009 report by the Permanent Mission of the Republic of
Zambia in Geneva to the Office of the High Commissioner
provides official figures on the coverage of these pilots
just before the SCT started its second stage, with different
targeting and selection mechanisms used from 2010 to 2014.
According to this source, the five pilots covered a total of
7,337 households and another 4,580 individuals. Because
the Kalomo, Kazungula, Chipata and Monze pilots targeted
households, and not individuals, their coverage figures are
presented in terms of households. The Katete pilot, however,
targeted individuals; hence, its coverage figures are presented
in terms of individuals. The pilot in Kalomo covered 3,573
households, in Monze covered 1,900, in Chipata covered
1,190, and in Kazungula covered 674.
From 2010 to 2014, however, the SCT acquired both more scale
and institutional strength, as its spread across the country’s
districts was meant to standardise its operations (as opposed
to the pilots, which were purposefully set up to operate with
certain differences to find out which arrangement would work
best). Between 2010 and 2014 the programme’s targeting was
subdivided into two streams: the CG and the MCTG.
The CG aimed to enrol all households with children under 5
years old, starting with districts with the highest child mortality
rates (Shagbom’bo, Kalabo, Kaputa, Zambezi and Milenge).
The MCTG aimed to enrol all extremely poor households with
elderly people and orphans, or widows and orphans, as well
as those with members with disabilities, starting with districts
chosen among the poorest ones (Zambezi, Serenje and Luwingo).
It would not be, however, until the next stage of the programme,
from 2014 onwards, that it would count on an objective way to
estimate poverty. Between 2010 and 2014 that particular aspect
was determined solely by the impression of social workers and
community members involved in the selection process, even
though their matching with the categorical criteria was subjected
to more objective validation (van Ufford et al. 2016).
By 2014 the CG stream of the SCT had reached 28,000
households, 94 per cent of which were living below the extreme
poverty line prior to receiving the programme (at that time,
extreme rural poverty was at 74 per cent). As for the MCTG,
by 2014 it had reached 17,700 households, 95 per cent of which
were living below the extreme poverty line prior to receiving
the programme (MCDSW and AIR 2016a; AIR 2016a; MCDSW
and AIR 2016b; AIR 2016b).
The third and current stage of the SCT, which started in
2014, was marked by the decision to keep those beneficiaries
previously enrolled through the CG and MCTG selection criteria,
though new beneficiaries would have to be selected through
a new targeting method. Known as harmonised targeting,
this consisted of targeting extremely poor households with
dependency ratios equal to or greater than three. This solution
was perceived as a way to reach out to the poorest population
being omitted by the previous targeting methods, while
ensuring programme enrolment for many beneficiaries who
would previously have been selected either by the CG or the
MCTG. The new targeting method was not only meant to render
the programme more progressive, but it also had the potential
to make its administration easier by no longer having to operate
a programme with two different targeting and selection
processes (van Ufford et al. 2016).
In this way, since 2014 the SCT has expanded its coverage
by targeting extremely poor households considered
labour-constrained due to not having any members who are
fit to work, or by having dependency ratios equal to or greater
than three (dependent members include those younger than
19, those older than 64 and those aged 19–64 with a chronic
illness or disability). The current model is set to gradually cover
all districts of Zambia, starting with the poorest ones (with the
highest poverty ratios) according to the 2010 Living Conditions
Monitoring Survey (ibid.).
Another innovation is that, since 2014, the SCT has incorporated
a proxy means test (PMT) in its selection process; therefore, a
household’s poverty level is no longer solely decided by the
impressions of community workers and social assistants. Rather,
families have been ranked according to poverty scores based
on a form that collects information on their socio-demographic
characteristics and living conditions. This does not mean that
community and social workers’ impressions and deliberations
are no longer part of the selection process, since they still play
a fundamental role in screening households to be ranked by
the PMT, and even in requesting reviews of cases they consider
incorrectly assessed. According to the Ministry of Community
Development and Social Welfare (MCDSW 2015), the selection
process since 2014 can be described as follows:
a. Volunteer social workers in the community (members of
the Community Welfare Assistance Committee—CWAC),
with the support of local leaders, identify households that
meet the first two eligibility criteria: i) residency; and
ii) incapacity/labour constraint.
b. Designated interviewers, mostly teachers, known as
enumerators, visit households identified in the first
stage to collect more information about them.
c. The information collected from households during
enumeration (stage 2) is entered into the SCT Management
and Information System (MIS) to produce a list of eligible
3
Policy Research Brief
households limited to the 15 per cent with the highest
vulnerability scores in the PMT.
d. The list of eligible households generated by the MIS is
shared with the community, allowing community members
to propose removing better-off households from the list
(leading to a final list of beneficiaries covering 10 per cent
of the area’s population).
e. Selected households are notified and provided with
information about the operation of the programme and
their entitlements.
TABLE 1
Targeting and selection process of the SCT over its three stages, disaggregated by specific pilot (for the first stage)
or programme stream (for the second stage)
Stage 1 (2003-2010)
Pilot or
stream (year
of launch)
District level targeting Household-/individual-level
targeting Selection process
Kalomo
(2003)
District chosen politically, with influence
of donors and the goal of experimenting
in different local contexts
Destitute and incapacitated/
labour-constrained households
Social workers and community
volunteers deliberate about those
perceived to be the most destitute
and incapacitated
Kazungula
(2005)
District chosen politically, with influence
of donors and the goal of experimenting
in different local contexts
Destitute and incapacitated/
labour-constrained households
Social workers and community
volunteers deliberate about those
perceived to be the most destitute
and incapacitated
Chipata
(2006)
District chosen politically, with influence
of donors and the goal of experimenting
in different local contexts
Destitute and incapacitated/
labour-constrained households
Social workers and community
volunteers deliberate about those
perceived to be the most destitute
and incapacitated
Monze
(2006)
District chosen politically, with influence
of donors and the goal of experimenting
in different local contexts
Destitute and incapacitated/
labour-constrained households
Social workers and community
volunteers deliberate about those
perceived to be the most destitute
and incapacitated
Katete (2007)
District chosen politically, with influence
of donors and the goal of experimenting
in different local contexts
Poor elderly people
Social workers and community
volunteers do the active search for
poor elderly people
Stage 2 (2010-2014)
CG Starting with districts with the highest
child mortality rate
All district households with
children under 5 years old
Location by social and community
workers and volunteers, and
enrolment of those who match the
easy-to-observe categorical criteria
(i.e. having children under 5 years old)
MCTG Starting with districts chosen among
the poorest ones
Extremely poor households
with elderly people and
orphans, or widows and
orphans, as well as those with
members with disabilities
Social and community workers’
perceptions of households’
poverty levels defines a first list
of potentially eligible households;
final eligibility depends on
checking whether these
households satisfy the
objective categorical criteria
laid out by the programme
Stage 3 (2014 onwards)
Harmonised
targeting
Starting with the
poorest districts
Extremely poor households
considered labour-constrained
due to not having any members
who are fit to work, or by having
dependency ratios equal to or
greater than three (dependent
members include those younger
than 19, those older than 64 and
those aged 19–64 with chronic
illness or disability)
Social and community workers’
perceptions of households’
potential eligibility define a list
of households, whose objective
categorical criteria are then
checked, and who are ranked
in terms of their poverty levels
estimated through a PMT
Source: Authors’ elaboration based on Garcia and Moore (2012); MCDSW and AIR (2016a); AIR (2016a); MCDSW and AIR (2016b); AIR (2016b); and van Ufford et al. (2016).
4 International Policy Centre for Inclusive Growth
Official estimates indicate that around 25 per cent of the
Zambian population are expected to be eligible considering
only the categorical criteria of the SCT’s harmonised
targeting format, while 20 per cent are expected to be
eligible under the categorical criteria and due to being poor
based on the PMT and community-based assessment (van
Ufford et al. 2016). Tabulations prepared on request for this
study revealed that, by December 2016, the harmonised
targeting stream of the SCT reached over 239,000 households
(7–8 per cent of the population, and, assuming no inclusion
errors, 20 per cent of the extremely poor population of the
country). A conversation with programme stakeholders
indicated that the government intended to double this
coverage in 2017.
TABLE 2
Coverage of the SCT over its three stages, disaggregated by specific pilot (for the first stage) or programme stream
(for the second stage)
Stage 1 (2003-2010)
Pilot Coverage Year of reference
Katete 4,580 persons 2009
Kalomo 3,573 households 2009
Monze 1,900 households 2009
Chipata 1,190 households 2009
Kazungula 674 households 2009
Stage 2 (2010–2014)
Stream Coverage Year of reference
CG 28,000 households 2014
MCTG 17,700 households 2014
Stage 3 (2014 onwards)
Current format Coverage Year of reference
Harmonised targeting 239,000 households 2016
Source: Author’s elaboration based on: Permanent Mission of the Republic of Zambia in Geneva (2009); MCDSW and AIR (2016a); AIR (2016a); MCDSW and AIR (2016b);
AIR (2016b); and tabulations received from the MCDSW at the end of 2016.
Benefit and payment structure
The pilots which comprised the first stage of the SCT were rolled
out mostly with the purpose of experimenting with different
designs to find out which model was most efficient and could,
therefore, be considered for a further stage of scaling up the
initiative. As such, the five pilots carried out between 2003 and
2010 differed from each other in terms of the benefit values,
benefit formula, the frequency of payments and even the
strategy for distributing the benefits to beneficiaries. Before
presenting these features, it is important to highlight that,
from 2012 onwards, the Zambian currency, the Kwacha (ZMW),
underwent a reform that divided values by 1,000 (therefore,
post-2012 values of, for instance, ZMW50,000 become simply
ZMW50). This explain the extreme variations in terms of nominal
values before and after that date. Nevertheless, all values
presented here also have their equivalent in USD PPP 2011.3
The first SCT pilot to be rolled out, in Kalomo district, awarded
beneficiaries a basic grant of ZMW40,0004
and a ZMW10,000
bonus to households with children. Due to logistical constraints,
however, the distribution of benefits did not take place every
month. Instead, the accumulated value for two months was paid
bimonthly. At first the programme envisaged paying the benefits
via banks, for beneficiaries living 15km or closer to the bank, and
organising pay-points for those living farther away. However,
it soon realised that even beneficiaries living close to the bank
struggled to access their benefits; therefore it started organising
pay-points even for those living close to banks (Garcia and Moore
2012; Schuring, Michelo, and Boonstoppel 2007).
The Kazungula district pilot awarded a higher basic benefit, of
ZMW50,000, and an additional bonus of ZMW20,000 per child in
the household. The accumulated value for two months was paid
bimonthly, predominantly via pay-points (Garcia and Moore
2012; Schuring, Michelo, and Boonstoppel 2007).
In Chipata district the pilot awarded a basic monthly benefit
of ZMW50,000, with a ZMW10,000 additional bonus to
households with two or more individuals, a ZMW10,000
additional bonus to households with children enrolled in
primary school and a KZMW0,000 additional bonus for
households with children enrolled in secondary school
(Garcia and Moore 2012; Schuring, Michelo, and Boonstoppel
2007). Unlike all the other pilots, payments were meant to
take place every month, though in practice the initiative was
marked by endemic delays (Habasonda 2009). It started with
a mixed payment routine, through both banks and pay-points,
but shifted towards a more predominant use of pay-points
latter on (Garcia and Moore 2012; Schuring, Michelo, and
Boonstoppel 2007; Habasonda 2009).
The Monze district pilot awarded the very same benefit as
that of the Kalomo district pilot, also on a bimonthly basis and
predominantly distributing benefits to beneficiaries via paypoints.
A slight difference between this pilot and the Kalomo
district one is that Monze’s beneficiaries were supposed to
commit, either in writing or orally, to keeping their children
in school and following a schedule of health visits. However,
households were not subjected to any sanction if they failed
5
Policy Research Brief
to observe these commitments, and, in practice, not even this
form of soft conditionality turned out to be taking place (Garcia
and Moore 2012; Schuring, Michelo, and Boonstoppel 2007;
Seidenfeld and Handa 2011; Handa, Seidenfeld, and Tembo 2012).
For the Kalomo, Kazungula, Chipata and Monze pilots, it is
interesting to note that the value of the basic grant (originally
ZMW30,000 per month, which was later increased to ZMW40,000
or ZMW50,000 depending on the pilot) was meant to cover the
price of a 50kg bag of maize, which would allow a household
of six individuals to consume an additional meal per day,
presumably their second (Garcia and Moore 2012; Schuring 2010).
These four pilots were meant to reassess the eligibility of
households every three years. For households no longer eligible
for the programme, a lump sum grant of ZMW500,000 was
awarded on graduation (Garcia and Moore 2012). This lump
sum payment would not apply to the Katete district pilot, since
it was a universal old-age pension; therefore, beneficiaries were
not expected to graduate. Its benefit consisted of a flat grant
of ZMW60,000 per month, with the accumulated value of two
months’ worth of benefits paid bimonthly, predominantly via
pay-points (Garcia and Moore 2012; Schuring, Michelo,
and Boonstoppel 2007).
In terms of the average benefit awarded by each pilot per
household per month, the Katete pilot provides the highest value,
ZMW75,000 (USD36.41 PPP 2011), followed by the Chipata pilot,
at ZMW60,000 (USD29.13 PPP 2011) (Permanent Mission of the
Republic of Zambia in Geneva 2009; Garcia and Moore 2012).
There are no such specific figures available for the Kalomo, Monze
and Kazungula pilots, but the average for these three initiatives
taken together is ZMW47,500 (USD23.10 PPP 2011) (Permanent
Mission of the Republic of Zambia in Geneva 2009).
From 2010 to 2014, when the SCT evolved from several pilots into a
unified, single programme operating through two streams (the CG
and the MCTG), the benefit formula became a flat grant of ZMW60
per month (USD22.74 PPP 2011) per household, increasing to
ZMW70 in 2014 (USD23.00 PPP 2011), paid bimonthly. Payments
were delivered via pay-points. Although from 2014 onwards the
SCT underwent major design changes with the introduction of
harmonised targeting, its benefit remained the same at ZMW70
per household per month (van Ufford et al. 2016).
TABLE 3
Summary of benefit and payment structure across the stages of the SCT, disaggregated by kind of pilot (for the first stage)
and programme stream (for the second stage)
Pilot
(year of
launch)
Benefit formula/structure
(nominal values as of 2009)
Household-/
individual-level
targeting
Selection process
Stage 1 (2003–2010)
Kalomo (2003)
ZMW40,000/USD19.42 PPP 2011 (basic benefit) +
ZMW10,000/USD4.85 PPP 2011 (bonus for households
with children)
Bimonthly
Banks, and pay-points for those
farther than 15km from a bank;
but soon even those close to
banks began receiving the
benefit via pay-points
Kazungula
(2005)
ZMW50,000/USD24.28 PPP 2011 (basic benefit) +
ZMW20,000/USD9.71 PPP 2011 (for each child) Bimonthly Predominantly pay-points
Chipata
(2006)
ZMW50,000/USD24.28 PPP 2011 (basic benefit) +
ZMW10,000/USD4.85 PPP 2011 (bonus for households
with two or more individuals) + ZMW10,000/USD4.85
PPP 2011 (bonus for households with children enrolled in
primary school) + ZMW20,000/USD9.71 PPP 2011 (bonus
for households with children enrolled in secondary school)
Monthly
(but marked by
endemic delays)
Only changed to pay-points
later (when it started being
subjected to an assessment)
Monze (2006)
ZMW40,000/USD19.42 PPP 2011 (basic benefit) +
ZMW10,000/USD4.85 PPP 2011 (bonus for households
with children); soft conditionalities never rolled out
Bimonthly Predominantly pay-points
Katete (2007) ZMW60,000/USD29.13 PPP 2011 per elderly person Bimonthly Predominantly pay-points
Stage 2 (2010–2014)
CG ZMW70/USD23.94 PPP 2011 (flat) Bimonthly Predominantly pay-points
MCTG ZMW70/USD23.94 PPP 2011 (flat) Bimonthly Predominantly pay-points
Stage 3 (2014 onwards) (nominal values as of 2015)
Harmonised
targeting ZMW70/USD23.94 PPP 2011 (flat) Bimonthly Predominantly pay-points
Source: Authors’ elaboration based on Garcia and Moore (2012); MCDSW and AIR (2016a); AIR (2016a); MCDSW and AIR (2016b); AIR (2016b);
and van Ufford et al. (2016).
6 International Policy Centre for Inclusive Growth
3 Governance
Throughout its three stages, the SCT has always been an
unconditional cash transfer programme, managed by the
MCDSW. The institutions responsible for each aspect of
enrolment, selection and payment, however, have changed
and evolved alongside the three phases of the programme.
Up until 2007-2008 these roles were mostly performed by
local partners contracted by donors, such as the UK
Department for International Development (DFID), the
German Technical Cooperation Agency (GTZ) and CARE.
From 2007-2008 onwards, and especially from 2010
onwards, these responsibilities were transferred to the local
government, more specifically to local Community Welfare
Assistance Committees (CWACs). At the subnational level,
the Public Welfare Assistance Scheme (PWAS) designates
District Welfare Assistance Committees (DWACs) responsible
for establishing, supporting and monitoring Area Coordinating
Committees (ACCs) that oversee the CWACs (Chiwele 2010).
The CWACs perform case management not only for the
grants and in-kind support they deliver on their own
(which are actually of very limited reach and budget),
but also for the SCT. However, given that the Public Welfare
Assistance Committee’s (PWAC) entire ground staff consists
of volunteers, there are severe limitations to its operation,
especially when considering that the volunteers of the PWAS,
CWACs and ACCs are often almost as vulnerable as the people
who benefit from the programme.
FIGURE 1
The PWAS structure
District Welfare
Assistance Commi?ees
(DWACs)
Community Welfare
Assistance Commi?ees
(CWACs)
Area Coordina?ng
Commi?ees (ACCs)
Reach out to poten?ally eligible households and
mo?vate them to apply to available social programmes
Occasional involvement with supplementary
ini?a?ves to provide case management and
referral support to SCT beneficiaries
Organise the community-based stage
of the programme’s selec?on process
Support and monitor payment opera?ons
Source: Authors’ elaboration based on Chiwele (2010).
Baseline reports from impact evaluations of the CG and MCTG
programmes indicate that targeting of these initiatives is good
(AIR 2016a; 2016b). However, another targeting assessment
suggests that CWACs tend to prioritise identifying their relatives
and neighbours in contexts where a limited budget prevents
coverage of all eligible households in a district (Beazley and Carraro
2013). Since 2014, data collection and validation have become the
task of enumerators, who are teachers designated to enrol and
validate the information declared by households identified by the
CWACs as potentially eligible for the SCT (van Ufford et al. 2016).
In terms of funding, there has been a significant expansion,
as well as a gain of financial ownership by the government.
The original pilots were mostly funded by donors (DFID, GTZ
and CARE), but, starting mostly in 2010, the government has
been expanding its share. Interviews with MCDSW stakeholders
revealed that in 2016 the government contributed 85 per cent
of the programme’s budget. For 2017 the government planned
to double its social protection budget (a 108 per cent increase)
and to further scale up the SCT (UNICEF and ZIPAR 2016).
Impact evaluations
Of the five pilots that formed the SCT from 2003 to 2010,
only the Kalomo and Monze pilots were subjected to impact
evaluations, and the Kalomo impact evaluation was actually
a comparison of beneficiaries’ conditions before and after
receiving the grant, since the study had no control group.
Once the SCT shifted its format to that prevailing from 2010
to 2014, it underwent another impact evaluation exercise.
But since its latest design change, from 2014 onwards,
it has not undergone any impact evaluation.
The main findings of the Kalomo impact evaluation (a comparison
of beneficiaries’ conditions before and after the intervention)
7
Policy Research Brief
included desirable improvements in terms of school enrolment,
access to more than one meal a day, more satisfaction with the
meals eaten, access to a more varied diet, less reported illness,
a decrease in the number of households depending on external
sources of income (such as relatives and neighbours), a reduction
in household debts and increased asset ownership, investment
and consumption. There was a decrease in the practice of
begging among the communities covered, and there was
no significant evidence of misuse of the SCT benefit money.
The sole aspect that did not cause the expected impacts was
that of school absenteeism, which increased in the short term
(even though there were signs of improvement in the long term)
(Schuring, Michelo, and Boonstoppel 2007).
Regarding the Monze district pilot, the main findings of its
impact evaluation include improvements in terms of livestock
ownership, access to fertilisers and a greater amount of cash
crops. There have also been positive impacts such as increased
enrolment and on-time school entry. Beneficiaries were also
found to experience positive impacts on their expectations
of quality of life and willingness to delay gratification.
Worryingly, there were no impacts on food expenditure
or food composition, nor on any health outcomes.
Besides sampling limitations, the authors of the impact
evaluation study also noted that the lack of positive impacts
on health and nutrition might be a result of the systematic
delays in payment delivery, such that beneficiaries would
often receive a large amount of money covering many
months when they were not paid, which would stimulate
them to invest the money (or pay debts with it), instead
of using it for routine consumption related to health and
nutrition. Another important shortcoming identified by
the impact evaluation had to do with severe delays in the
support meant to be given to CWAC members to enable
them to conduct their tasks (Seidenfeld and Handa 2011).
The main findings of the impact evaluation of the CG and the
MCTG streams of the SCT include positive effects on income,
depth of poverty, food security, housing and living conditions.
Furthermore, positive impacts were found on children’s access
to clothes and other basic material needs, productivity and
asset ownership, as well as on the likelihood of children to
be enrolled in school while also reducing their probability of
dropping out of school. The CG stream also revealed positive
impacts on reducing diarrhoea incidence among children
under 5 years old and improving infant and young child
feeding in beneficiary households.
The MCTG impact evaluation included a module on adolescents
aimed at assessing impacts on knowledge, attitudes and
practices (KAP) related to HIV. This measured impacts on age
of sexual debut, age-disparate sexual partnerships, condom use
and mental health. None of them, however, revealed any benefit
from the SCT (MCDSW and AIR 2016a; AIR 2016a; MCDSW and
AIR 2016b; AIR 2016b).
Maybe more important than the impact evaluation studies per se
were some coverage and process evaluation studies that took place
and led to alterations to the programme’s operations and overall
design (mostly on its targeting and selection process). The targeting
assessment of the CG and the MCTG by Beazley and Carraro (2013),
for instance, revealed that the categories used did not
have the highest correlation with extreme poverty. According to
van Ufford et al. (2016), this study was fundamental to motivating
the shift towards the harmonised targeting approach, including the
introduction of a PMT to rank people in terms of their poverty level,
and the introduction of enumerators to collect data for the PMT
among households identified by CWAC members.
4 Conclusions
The trajectory of Zambia’s SCT from a set of pilots to its current,
uniform design reveals improvements following a process
of experimentation regarding the best selection criteria and
processes vis-à-vis the available state capacity to handle the
SCT’s core operations. Back in 2003, the pilots that would
later turn into the SCT were delegating many crucial roles to
associations and non-state actors. Furthermore, the selection
criteria were largely based on community-based mechanisms
without even a clear and objective benchmark. Over time, the
State seems to have gained the capacity to assume control
of the SCT’s core operational functions, and has been able to
gradually standardise the SCT and make the selection process
more transparent and progressive.
From its second to its current, third phase, the many categorical
criteria that mediate eligibility were simplified by adopting one
single categorical selection criterion capable of encompassing
many of the vulnerable groups previously listed as eligible
according to the CG and, especially, the MCTG. In many respects,
one can think of the current SCT selection criterion as a revamped
and easier-to-manage version of the MCTG, which, instead
of listing all eligible family types, uses a broader category
(the dependency ratio) likely to reach out to previously covered
families and many others also found to be in a situation of
vulnerability. The progressivity of this new approach also benefits
from a complementary PMT that has been used since 2014.
Zambia’s SCT also holds the distinction of having
implemented these institutional and design improvements
without compromising the expansion of its coverage, and
while increasing the government’s financial ownership of
the initiative. Currently, the SCT is one of the cash transfers
in the subregion with the largest coverage and greatest
government ownership.
Though relatively little time has passed since the SCT was last
subjected to an impact evaluation study, it seems strategic to
undertake another such study, as the design of the programme
has been significantly altered since 2014. The impact evaluation
could benefit from the programme’s plan to expand its coverage
in the near future, including the expansion towards districts that
are not yet being covered. This could make it easier to build a
counterfactual group.
Certain studies and assessments have proven crucial to many
of the SCT’s design changes, most notably that taking place
from 2014 onwards. To a large extent, this was influenced by
targeting assessments of the CG and the MCTG. Not only was
the targeting of the programme altered as a result of this study,
but even its selection process changed, as poverty assessments
shifted from community impressions to a PMT-based ranking.
The very collection of information by enumerators to run this
PMT emerged as a way to avoid selection biases found by the
study to be introduced by CWAC members.
8 International Policy Centre for Inclusive Growth
More such relations between programme assessments and
its operational features should be encouraged, to avoid the
trap of running studies whose results are only considered by
programme managers as long as they attest to the good quality
of the initiative. Assessments and impact evaluations are not
pieces of propaganda, and their findings should be given political
relevance even (and specially) when they imply the need for
adjustments to the programme. In the past, impact evaluation
studies of the Kalomo and Monze district pilots, for instance,
led to no programme adaptation whatsoever, even though
the studies suggested that the programme was not capable
of fostering positive health outcomes. Similarly, the impact
evaluation of the MCTG that revealed no impacts on HIV-related
KAP has not led to any significant restructuring of the programme
to make it more effective at promoting such outcomes.5
The need for institutional plans to incorporate evidencebased
critiques of the programme’s design can be illustrated
by the SCT pilots taking place between 2003 and 2010. Even
though they were carried out with the explicit purpose of
experimenting and learning so as to decide on the best
design for a national programme, it turned out that only two
of them were subjected to impact evaluations. And none of
them really contributed to the decisions about how to scale
up the initiative from 2010 onwards. The Chipata pilot, which
adopted one of the most creative payment structures, was
not evaluated nor considered adequately in this process. It is
plausible that, had all the pilots been subjected to evaluation
and their findings taken seriously, the SCT could probably have
started to operate, back in 2010, in the format it assumed in
2014, instead of operating for four years in formats that were
later found to be undesirable.
Culled from The International Policy Centre for Inclusive Growth, is jointly supported by the United Nations Development Programme and the Government of Brazil
