An assessment of
Cambodia’s Cash Transfer
program for the Poor and
Vulnerable Households
during COVID-19
The World Bank
June 2021
ACKNOWLEDGEMENTS
This report is authored by Maheshwor Shrestha with contributions from Robert Palacios and
Kenichi Victor Nishikawa Chavez. We thank Sodeth Ly, Marco Nicoli, Vrishali Shekhar, and
Kimsun Tong for their invaluable inputs to this report. We are very thankful to Wendy Karamba,
Clarence Tsimpo Nkengne, and Kimsun Tong for their cooperation and support on the data
collection through the Cambodia High-Frequency Phone Survey.
The study was conducted under the close supervision and guidance of Yasser El-Gammal, Inguna
Dobraja, and Mariam Sherman.
Special thanks to the General Secretariat of the National Social Protection Council of Cambodia
for their guidance and support. The team also thanks Ministry of Planning and the Ministry of
Social Affairs, Veterans and Youth Rehabilitation for helpful guidance and support.
1
TABLE OF CONTENTS
ACKNOWLEDGEMENTS ........................................................................................................................................1
EXECUTIVE SUMMARY .........................................................................................................................................4
1. INTRODUCTION AND CONTEXT ................................................................................................................7
A) SOCIAL ASSISTANCE IN CAMBODIA PRE-COVID-19 .........................................................................................7
B) IDPOOR PROGRAM AND ITS ROLE IN TARGETING SOCIAL ASSISTANCE ..............................................................8
C) IMPACTS OF COVID-19 IN CAMBODIA..............................................................................................................9
2. CAMBODIA’S RESPONSE TO THE COVID-19 PANDEMIC .................................................................. 14
A) CASH TRANSFER PROGRAM ............................................................................................................................. 15
B) COVERAGE OF THE CTP-COVID .................................................................................................................... 16
C) ADEQUACY OF THE PROGRAM ......................................................................................................................... 20
D) CAMBODIA’S RESPONSE IN THE CONTEXT OF GLOBAL AND REGIONAL RESPONSE ............................................ 21
3. IMPLEMENTATION OF THE CTP-COVID ............................................................................................... 23
A) INSTITUTIONAL ARRANGEMENTS..................................................................................................................... 23
B) DELIVERY SYSTEM FOR CTP-COVID ............................................................................................................. 26
4. RESULTS AND IMPACTS ............................................................................................................................. 28
A) REACH OF THE PROGRAM AMONG INTENDED BENEFICIARIES .......................................................................... 28
B) BENEFICIARY USE AND EXPERIENCE WITH THE CTP-COVID .......................................................................... 31
C) USAGE OF THE TRANSFERS AND WELFARE OUTCOMES .................................................................................... 37
5. POLICY RECOMMENDATIONS ................................................................................................................. 42
A) SHORT-RUN POLICIES ...................................................................................................................................... 42
B) MEDIUM-TERM POLICIES ................................................................................................................................. 43
6. ANNEX: ADDITIONAL RESULTS ............................................................................................................... 47
2
INDEX OF FIGURES
Figure 1: The economy has contracted for the first time in 25 years ........................................................................... 10
Figure 2: Cambodia’s growth drivers are decelerating dramatically ........................................................................... 10
Figure 3: Incomes from key livelihood sources decline since the pandemic ............................................................... 11
Figure 4: Coverage of the IDPoor program by consumption decile ............................................................................ 17
Figure 5: A quarter of the households receive some form of social assistance throughout the pandemic ................... 19
Figure 6: CTP-COVID cash transfer dominates social assistance after June 2020 ..................................................... 20
Figure 7: CTP-COVID transfer amounts are sizable shares of the consumption of the poor ...................................... 21
Figure 8: Income per-capita and coverage of the cash transfer programs ................................................................... 22
Figure 9: Economic growth, coverage, and expenditures in cash transfer programs in EAP ...................................... 23
Figure 10: Almost all of the intended IDPoor families received the CTP-COVID relief transfer ............................... 28
Figure 11: High coverage of social assistance among eligible IDPoor households ..................................................... 29
Figure 12: Awareness and card validity issues were most prevalent among those who did not receive the transfers . 30
Figure 13: Grievance mechanism and Conflict Resolution of the CTP-COVID ......................................................... 31
Figure 14: Frequency and amounts of CTP-COVID transfers were in line with expectations .................................... 32
Figure 15: Average monthly transfer per family has been relatively stable across the program period ...................... 32
Figure 16: Most CTP-COVID beneficiaries found the registration and payment process to be easy .......................... 34
Figure 17: Only some beneficiaries expressed preference to save the CTP-COVID transfers .................................... 35
Figure 18: A sizable share of the IDPoor families live in villages without a Wing payment agent ............................ 36
Figure 19: A significant share had to travel 30 mins or more to receive the transfers ................................................. 36
Figure 20: Most households used the transfers to purchase food and essentials ......................................................... 37
Figure 21: Beneficiaries consider CTP-COVID transfer to be very important ............................................................ 38
Figure 22: CTP-COVID transfers significantly muted income shocks for recipient households ................................ 39
Figure 23: CTP-COVID receipt is associated with better HH welfare for some outcomes ......................................... 40
Figure 24: Higher benefit amounts under CTP-COVID is associated with higher measures of wellbeing ................. 41
Figure 25: Schematic of varying benefit levels by poverty and vulnerability ............................................................. 45
INDEX OF TABLES
Table 1: Amounts of payments under the CTP-COVID .............................................................................................. 16
INDEX OF BOXES
Box 1: Cambodia High Frequency Phone Surveys (HFPS) ........................................................................................ 13
Box 2: The On-demand IDPoor identification process. ............................................................................................... 18
3
EXECUTIVE SUMMARY
Pre-COVID social assistance system was nascent in Cambodia but had some strong features
that facilitated quick expansion of assistance during COVID-19. Until 2015, Cambodia spent
around 0.09 percent of its GDP on social assistance – one of the lowest in the world and much
lower than comparator countries. The launch of the cash transfer for pregnant women and children
under two (CTP-PWYC) in June 2019 marks the first nationwide cash transfer program and a
positive step towards providing greater social assistance. The program is targeted to the identified
poor (IDPoor) women and provides cash payments through a mobile payment vendor. The IDPoor
program, launched since 2006 in rural areas and gradually expanded to urban areas by 2017,
identifies families as poor and provides them an IDPoor card (equity card) which entitles them to
poverty targeted government programs such as the free health insurance for the poor. The existence
of the IDPoor program to identify poor families and the mechanism established for CTP-PWYC
to provide cash payments through a mobile payment vendor were crucial features that enabled the
government to provide support during COVID-19.
The “Cash Transfer for the Poor and Vulnerable Households during COVID-19�? (CTP-
COVID) program leveraged those features to quickly launch and implement the support
program. COVID-19 had large and negative impacts on the Cambodian economy which halted
an impressive growth trajectory achieved in the past 15 years. Key exports, including garments
and tourism, shrunk which adversely affected livelihoods. The government introduced a relief
package of programs among which the CTP-COVID – a cash transfer program targeted to the
IDPoor families – remains the largest. The CTP-COVID leveraged the existing IDPoor database
of identified poor families as well as the implementation and delivery arrangement of CTP-PWYC
to provide monthly cash payments to the poor households. The on-demand beneficiary
identification of the IDPoor program was also scaled-up to allow newly poor and the new poor to
be included in the program throughout its implementation. By June 2020, the program was ready
to disburse cash in the hands of beneficiaries.
The CTP-COVID had a relatively low coverage of beneficiaries, but the benefit levels of the
program were adequate. The IDPoor program only identified about 540,000 households as poor
by June 2020. This was only about 17 percent of all households in the country – much fewer than
those adversely affected by the pandemic. Even with the expansion in coverage through the on-
demand identification process, which identified an additional 153,000 households within a year,
only one in five households in the country were covered by the program. Coverage was lower
than other countries in the region with the exception of Lao PDR. On the other hand, the benefits
provided under the CTP-COVID were sizable and constituted about 20 to 29 percent of the
consumption of the poorest decile. Simulation results show that the transfers have the potential to
reduce pre-COVID level of poverty by 3.3 percentage points compared to the pre-pandemic levels
of poverty.
The experience of the beneficiaries and the impacts of the program were largely positive .
Based on the primary data collected by the World Bank and the available administrative data, this
report finds that:
• The CTP-COVID reached a large share of the intended population. Over 95 percent of the
beneficiaries with valid IDPoor card received the transfers. This is a praiseworthy success
in implementation and benefit delivery.
4
• Beneficiaries had very positive experience with the registration and payment processes.
Most beneficiaries found the process for beneficiary registration and the process for
withdrawing payments relatively easy. Grievances with the program were not that
common, although not unheard of – particularly in the initial months.
• Travel time to withdraw benefits was a hurdle for a sizable minority. More than a fifth of
the beneficiaries live in villages that are at least 5 kilometers from the nearest payment
agent. Limited travel option for the poor could mean that this could present significant
hurdles. Over a quarter of traveled at least 30 minutes to withdraw payments. Geographic
disparities are large with longer travel times required in remote provinces in the plateau
and mountains.
• The cash transfers helped the beneficiaries to cope with adverse income shocks. Four out
of five beneficiary households mentioned that the transfers were either ‘important’ or ‘very
important’ for them. The transfers also muted overall income declines for the beneficiary
households. The report also finds suggestive evidence of improved welfare among the
recipients. Beneficiary, especially those who received higher amount of transfers, report
improved household wellbeing, are more likely to optimistic about their wellbeing over the
next few months and are less likely to report food insecurity.
Policy recommendations. Though implemented as an explicitly temporary program, the CTP-
COVID program offers important lessons for social assistance in Cambodia in the immediate term
where COVID-19 continues to adversely affect many families in Cambodia as well as the post-
COVID social assistance in the medium term. In the short-term, the report recommends to:
(i) Continue the support to poor families. As the recent and ongoing episode of the outbreak
shows, the pandemic and its effects are far from over. The economy will take longer than
expected to fully recover. In such a situation, the government should continue to support
poor and vulnerable families. Such support will prevent families from falling deeper into
poverty and helps them recover quickly when the economy bounces back.
(ii) Expand the coverage of the support to include more families. The government should find
ways to expand the IDPoor program to include not just the poorest, but also the near-poor,
the vulnerable, and the low-income families that are very likely to be adversely affected by
the pandemic and related lockdown restrictions. The introduction of a new cash transfer
program to low income families affected by the lockdown is a positive step in this direction.
(iii) Explore using mobile technology to quickly identify and register potential beneficiaries.
While the IDPoor process were established for a ‘regular’ time where commune officials
some flexibility in their work program, the additional burden of identifying new
beneficiaries may be burdensome for local implementers of the program. Mobile
technologies can be leveraged to facilitate identification and registration processes by
automating many of the time-consuming processes.
In the medium-term, stronger set of social assistance programs are required. Specifically, the report
recommends to:
(iv) Develop the IDPoor program to serve as a social registry. The pandemic has highlighted
the need for a strong social registry that all current and future social programs can use to
identify potential beneficiaries. The government should plan to develop the strong platform
of the IDPoor program to convert this into a social registry by expanding its coverage to a
large segment of the population, modernizing its scoring system to allow a granular level
of classification, and linking it to various administrative and program databases.
5
(v) Introduce strong and adequate social assistance programs. Particularly immediately after
the pandemic, social assistance will need to be much higher than its pre-pandemic level
both in terms of coverage and adequacy. In addition, tailoring benefit eligibility and
amounts to need (measure of poverty and or vulnerability) as well as benchmarking benefit
levels to the consumption of the poor would be recommended.
(vi) Introduce shock responsive elements to social assistance programs. The CTP-COVID was
an example of an ad-hoc shock responsive program. However, the government should
recognize that shocks, both idiosyncratic and covariate, occur regularly in Cambodia.
Hence, shock responsiveness should be in-built to the new package of social assistance,
the ‘family package’ currently under consideration, which allows for quick horizontal and
vertical expansion in response to shocks.
6
1. INTRODUCTION AND CONTEXT
a) SOCIAL ASSISTANCE IN CAMBODIA PRE-COVID-19
SA system is in its nascent stage in Cambodia
The Royal Government of Cambodia (RGC) exemplifies an economy of growth and transition
away from agriculture over the past few decades. Despite being world leaders in poverty reduction,
Cambodia’s statistics conceal the high levels of multidimensional poverty and vulnerability levels.
Although Cambodia has a deep history of government support for poverty alleviation, social
assistance was integrated into the National Strategic Development Plan (NSDP) only recently
when the National Social Protection Policy Framework (NSPPF) which covers a period of 2016-
25 was approved. The NSPPF recognizes the critical role played by social protection in preventing
and reducing poverty, vulnerability, and inequality, improving living standards, and paving a better
future for the economy and all of its citizens. For these reasons, the government envisions building
a social protection system that is inclusive, effective, and financially sustainable (Royal
Government of Cambodia, 2017).
Initial social assistance programs were geared towards improving the human capital, and
consequently welfare, of the poor and the vulnerable. This includes establishment of the Health
Equity Fund (HEF) to provide free health care to identified poor families, nutrition programs for
pregnant women and children, scholarship programs for primary schools, and school feeding
programs. Most of these programs, implemented at small-scale pilots (except the HEF), sought to
improve the health and human capital of children and were targeted towards poor and vulnerable
families. These programs were complemented by Food Reserve Program to prevent food insecurity
in disaster affected areas. The limited scale of these programs meant that a large-scale social
assistance programs were lacking in Cambodia prior to 2019.
Pre-2019 levels of Social Assistance were very low in Cambodia
Lack of large-scale social assistance programs (except scholarships and subsidized health
insurance) meant that Cambodia had one of the lowest expenditures in social assistance programs.
Until 2015, Cambodia spent around 0.09 percent of its GDP on social assistance – one of the lowest
in the world, much lower than the world average of 1.6 percent, or the regional average of 1.5
percent for developing and transition economies (World Bank, 2018). The expenditure was also
much lower than the average among its regional neighbors: 0.84 percent in Indonesia during 2013–
2015, 0.47 percent in Thailand during 2010–2011, and 0.67 percent in the Philippines during
2013–2014. Only 2 percent of the poorest quintile of Cambodia’s population were covered by
social assistance programs. Prior to 2019, Cambodia’s social assistance could be characterized
with abysmally low levels of funding, poor and fragmented coverage and targeted mostly towards
rural areas.1
1
OECD Development Pathways Social Protection System Review of Cambodia, 2017
7
The Cash Transfer Program for Pregnant Women and Children under two represents a positive
step towards providing broader social assistance
Motivated by the promising results from early pilots in promoting maternal and child health and
nutrition, RGC announced the nationwide Cash Transfer Program for Pregnant Women and
Children under two (CTP-PWYC) in June 2019.2 To maximize the impact of the government
support, the program targets women and children from identified poor (IDPoor families) as access,
health-seeking behavior, as well as maternal and child outcomes are worse for the poor.3
The program provides a total of $190 to pregnant women (and children under two) conditional on
antenatal visits (up to four visits), institutional delivery, and postnatal (up to ten) visits to health
facility. Beneficiaries receive $50 for delivery and $10 for each health visit. Transfers are provided
to beneficiaries through agents of a mobile-payments vendor.4
The CTP-PWYC reflects Cambodia’s commitment towards investing in human capital by
targeting early years of life and nudging health-seeking behavior amongst women. Supporting
evidence from early stage pilots demonstrated improved nutritional outcomes amongst targeted
women and her children and a higher utilization of maternal health services.5 The CTP-PWYC has
successfully transformed the landscape of the social assistance by gradually increasing coverage
from 58,324 beneficiaries within the first six month of announcement in 2019 to 143,000 in 2020,
and intends to cover over 800,000 beneficiaries by 2024. The RGC intends to spend around US$
40 million for the program between 2020 to 2024. This program is expected to significantly boost
social assistance expenditure as implementation picks up steam.
However, although a landmark program in Cambodia’s social assistance trajectory, the CTP-
PWYC is limited in terms of overall coverage and adequacy of the benefits. The program is
targeted to IDPoor families which currently covers fewer that one out of every five households in
the country. The program benefits are also small; transfers amount to only 1.8 percent of the
average consumption. The transfers are inadequate, by itself, in providing families with enough
additional resource to cover the nutrition gap they may face. However, the transfers can still be
valuable in nudging better health-seeking behaviors from and outcomes of pregnant women and
children from poor families.
b) IDPOOR PROGRAM AND ITS ROLE IN TARGETING SOCIAL ASSISTANCE
Cambodia’s social assistance hinges heavily on the identification of the poor (IDPoor) program
As seen above, both the HEF and CTP-PWYC use the IDPoor program as a primary way to identify
beneficiaries and target support. The IDPoor program, managed by the Ministry of Planning, was
first established in 2006 in rural areas primarily to identify the poor families for the free health
insurance for the poor provided under HEF. Between 2014 and 2017, the program was rolled out
to all urban areas as well. Poverty assessments are carried out by Village Representative Groups
(VRGs) and the Commune/Sangkat Working Group (CSWG).
2
The program was formally established through a sub-decree (#245) launched in December 2019.
3
According to DHS 2014, Cambodia has one of the highest maternal mortality rates in the region at 161 per 100,000 live births in 2015 and one
of the highest infant mortality rates at 22.8 per thousand live births. Infant mortality rates are higher in households at the bottom quintile than the
top quintile (62 vs 16 per thousand live birth).
4
Beneficiaries enroll using their IDPoor cards and mobile phones. Upon meeting the condition, beneficiaries receive a PIN code in their mobile
phones with which they withdraw the transfer from the nearest agent of the mobile-payment vendors.
5
World Bank (2016). Kingdom of Cambodia: Impact Evaluation of the Maternal and Child Health and Nutrition Cash Transfer Pilot. Report#
AUS20675.
8
In the IDPoor assessment process, community-based selection and validation are carried out using
a structured questionnaire, which provides the basis for computing a household IDPoor score. By
early 2020, around 540,000 households (about 15 percent of all households in the country) were
classified as either poorest/destitute (IDPoor 1) or poor (IDPoor 2) and provided with an IDPoor
card. The IDPoor card, also known as the Equity card, is used by many government and non-
governmental program to target assistance to the poor.
Until recently, the assessment of the household happened in a third of the provinces each year.
That is, households in any province could be assessed at most once every three years. Working
groups in the villages would first come up with a list of potentially poor households and administer
the questionnaire to them. If the households have qualifying scores, their information is vetted by
the community through an iterative process. If households are eventually classified as IDPoor 1 or
IDPoor 2, An equity card is provided to them and their information is digitized and stored in a
database maintained by the Ministry of Planning.
An on-demand identification process (OD-IDPoor) was piloted in 2018 and was being rolled out
to all rural areas as of early 2020. The key distinction between the regular process and the OD-
IDPoor process was that the communes would lead the beneficiary identification and interviews
and that the data would be collected through tablets and directly synced with the central IDPoor
database.
An assessment of the ID poor program by the World Bank in 2019 found that the program does a
reasonable job of identifying those in need of support. However, simple technical improvements
such as switching to a regression-based scoring system could result in large improvements in
performance. Similarly, the assessment recommended an expansion of the coverage of the ID poor
program as well as introduction of a more granular classification system to identify the poor as
well as the vulnerable and the near poor.6
So, when the pandemic hit, Cambodia was at a nascent stage of social assistance delivery system.
There was one social assistance program with national coverage; delivery system for providing
assistance was new; and there was one database, the IDPoor database, that identified one-sixth of
the families in the country as poor.
c) IMPACTS OF COVID-19 IN CAMBODIA
The first case of COVID-19 in Cambodia was reported on January 27, 2020. Authorities were
quick to react to the health crisis by taking appropriate measures and were able to contain the
spread of the disease. Till February 2021, the country had done a remarkable job in limiting the
health impact of the pandemic. But that did not prevent the country from facing the adverse impacts
of the pandemic beyond that of the initial restrictions to travel and to economic activities. With the
recent uptick in the pandemic since March 2021 And the restrictive measures being taken to
contain the pandemic, the economic outlook for the country continues to look grim.
6
Together with the World Bank and other development partners, the Ministry of Planning has been engaged implementing some of these
recommendations.
9
COVID-19 has led to large adverse impacts on the Cambodian economy
The pandemic has stalled an impressive growth trend over the past two decades (Figure 1). The
economy contracted by 3.1 percent in 2020, a remarkable downturn which halted the impressive
7.7 percent average annual growth rate in the country between 1995 – 2019. The magnitude of
contraction was bigger than the initial estimates forecasts suggesting that the pandemic cut deep
into the economic lives of Cambodians. In the recent months, disease caseload has begun to
increase sending large parts of the country into severe lockdown restrictions. Though there were
signs of recovery in early part of 2021, the resurgence in disease and the mitigation measures
suggests that the recovery in 2021 is expected to be, at best, uneven. The negative impacts of the
pandemic are expected to continue well into 2021 with hopes for an optimistic V-shaped recovery
fading.
Figure 1: The economy has contracted for the first Figure 2: Cambodia’s growth drivers are
time in 25 years decelerating dramatically
Source: World Bank staff calculations based on National Source: World Bank staff calculations based on National
Accounts data. Accounts data.
The pandemic has thwarted the main drivers of Cambodia’s celebrated growth trajectory that is
heavily dependent on tourism, garment and footwear manufacturing, trade, and construction
(Figure 2). These sectors cumulatively contributed towards 42 percent of country’s GDP and
around 39.5 percent of total employment in 2019. These sectors were responsible for 52 percent
of the GDP growth in 2019.
Global demand shocks, supply chain disruptions and nationwide lockdowns have negatively
impacted exports, particularly that of garment which accounts for two-thirds of the manufacturing
exports. In 2020, garments and footwear sector contracted by 6.1 percent, leading to the contraction
of the manufacturing sector by 2.8 percent, even though the non-garment manufacturing sector
grew by 10 percent. Likewise, the construction sector, another important industry contributing 11
percent to the GDP in 2019, contracted by 1.8 percent.
In addition, the adverse impact of international travel restrictions has had deep repercussions on
the tourism industry. Cambodia saw an 80 percent decline in international arrival in December
2020 compared to the previous year. A loss of US$3 billion in tourism receipts has been estimated
for the year 2020. Consequently, the hospitality sector (hotels and restaurants) shrank by a
whopping 32.2 percent. Together with trade, which contracted by 5.4 percent, this contributed to
the contraction of the services sector which shrank by 6.3 percent.
10
COVID-19 has adversely affected the livelihood and income sources of the individuals
Unfortunately for Cambodia, the hardest hit economic sectors are also the most labor-intensive:
the garment sector accounts for 17 percent of non-farm employment, transport and hospitality
accounts for 11 percent, and construction accounts for 4 percent (World Bank, 2020). The broader
contraction in these sectors translates to a significantly large proportion of the workforce without
jobs or a lowered earning at the jobs that they hold. In addition, returning migrants from Thailand7
have added to the unemployment numbers and negatively impacted the livelihoods of remittance-
dependent households. Unemployment projections reveal a loss of 390,000 to 570,000 jobs in
2020, raising the unemployment rate from 0.7 percent in 2019 to 3.2 percent–4.4 percent in 2020
(ADB, 2020).8 Worse, the nascency of the social safety nets in Cambodia would mean that a lot
of these individuals are left to deal with the pandemic on their own pushing the near-poor into
poverty, and the poor deeper into poverty.
Figure 3: Incomes from key livelihood sources decline since the pandemic
Source: World Bank staff estimates from the Cambodia HFPS
Note: Sample restricted to LSMS+ households surveyed in Feb-March 2021
The high-frequency phone surveys (HFPS) conducted by the World Bank helps assess the impacts
of COVID-19 on household’s economic activities, incomes, access to essential services, exposure
to shocks and coping mechanisms.9 The data from HFPS households validates the adverse impact
of the pandemic on employment and livelihoods. As of February 2021, employment has yet to
return to pre-pandemic levels when 82 percent of respondents were working. Throughout the
various rounds of the surveys conducted between May 2020 and February 2021, employment rates
7
80,000 migrant workers have been estimated to have returned since the COVID-19 outbreak
8
ADB, 2021. “Employment and Poverty Impact Assessment: Cambodia�?.
9
See Box 1 for details on the survey. For the purpose of the analysis of this report, the period before January, 2020 is referred to as the “pre -
pandemic period�? and the period before June,2020 as the “pre-COVID-19 cash relief period�?.
11
remained below 72 percent. Seasonality in farming and employment led to a larger drop to 65
percent in October but employment resumed back to being around 72 percent in December 2020
and February 2021.
The decline in employment and the reduction in demand for goods and services has lowered
incomes for many households. Two-fifths of all households reported a decline in their total
household income in February 2021 relative to the pre-pandemic levels in January 2020. Another
40 percent reported no increase in total incomes.
The income source that was hit the hardest was the income from non-farm business activities. And
overwhelming 57 percent reported a decline in income from nonfarm business activities relative
to pre-pandemic levels, and another 40 percent reported no change. About 78 percent households
indicated that the reduction in consumer demand as the main reason for the income loss.
Household labor incomes also fell due to the pandemic. Only 11 percent of the households saw an
increase in their income from wages since the onset of the pandemic. The income reduction from
wages were particularly large for those reporting the reduction. Similarly, a large share of
households reported a loss in income from farming as well as remittances from migrant members
of their households. Though specific numbers differ, the reduction in incomes from various sources
was reported throughout all survey rounds from May 2020 to February 2021 which indicates a
sustained decline in incomes due to the pandemic.
Besides losses in employment and incomes, the pandemic also contributed to a reduction in human
capital acquisition through repeated school closures. Survey results show reduced child
involvement in learning activities during the times of school closures from March 2020 and
gradual reopening from August through November 2020. Schools closed again in December,
adversely affecting child learning. These learning gaps is expected to worsen the already low levels
of human capital in Cambodia. Thankfully, at least till March 2021, the health impact of the
pandemic had been very low, thanks to quick preventative actions. Consequently, the access to
markets, health facilities, and other basic essential services remained undisrupted.
The inadequate coverage of a social safety net that safeguards against impact on livelihoods,
income and human capital losses have led to a worrying deterioration in the households’ perception
about their economic status and well-being. Nearly half the households report that they have a
lower current and future well-being and economic status as compared to the previous year
underscoring the long-lasting impact of the pandemic on the households.
In this context, this report assesses the RGC’s response to the pandemic through its Cash Transfer
for Poor and Vulnerable households during COVID-19 (CTP-COVID) program. The report
elaborates on the features of the program, its design and implementation, comparing it to the
pandemic response from other countries in the region and globally and bringing in insights from
the household surveys conducted throughout the year (Cambodia HFPS, see Box 1). The report
then combines the lessons from the implementation of this program and the results from the survey
to provide policy recommendation for improving social assistance during the current pandemic
period as well as beyond the pandemic.
12
Box 1: Cambodia High Frequency Phone Surveys (HFPS)
To monitor the socio-economic household level impacts of COVID-19 in Cambodia, the World
Bank, in collaboration with the National Institute of Statistics (NIS), designed and implemented
the High Frequency Phone Survey (HFPS) of households to assess the impact of the COVID-
19 pandemic on economic activities, income levels, access to essential services, exposure to
shocks and coping mechanisms. For Cambodia, a total of 5 survey rounds are planned, with
households being called back every 1 to 2 months.
This allowed for a regular data generation and promoted evidence-based decision making within
the government as well as other development partners. The data provides rich information about
households that can be useful in assessing the impacts of the pandemic at a microlevel. The
survey covers important and relevant topics, including knowledge of COVID-19 and adoption
of preventive behavior, economic activity and income sources, access to basic goods and
services, exposure to shocks and coping mechanisms, and access to social assistance.
The Cambodia COVID-19 HFPS consists of two separate samples: (a) Living Standards
Measurement Study Plus (LSMS+) and (b) IDPoor households. LSMS+ is a nationally
representative household survey implemented October–December 2019 by the National
Institute of Statistics (NIS) with technical and financial support from the World Bank. LSMS+
consists of 1,512 households, of which 1,364 have a phone number. The same sample was
followed up in each round without any replacements for households that could not be followed-
up. Sampling weights were adjusted according to the steps outlined in Himelein (2014) to obtain
unbiased nationally representative estimates from the sample. LSMS+ sample weights were
computed to ensure representativeness at the national and urban/rural level.
To closely monitor the government’s Cash Transfer for Poor and Vulnerable Households during
COVID-19 (CTP-COVID), which was launched in June 2020, an additional sample from the
beneficiary list of the cash transfer program for women and children under two, who were all
from IDPoor families, and hence eligible for the CTP-COVID were included as a separate
sample. From the beneficiary list of about 10,000, about 1,000 IDPoor households were sampled
for survey. The IDPoor sample was divided into five strata: (i) Phnom Penh and other urban
areas, (ii) Plain, (iii) Tonle Sap, (iv) Coastal, and (v) Plateau and Mountain. The sample was
randomly selected proportional to the number of IDPoor households in each province. With
each successive round of the survey, previous sample was followed up. Additional sample was
drawn to replace the IDPoor sample as well as the LSMS+ sample using the sample sampling
strategy.
Surveys spanned May 2020 to March 2021. The first round of survey for the LSMS+ sample
was conducted in May 2021 whereas that of the IDPoor sample was conducted in June 2020.
For both samples, the second-round survey was conducted in Aug-Sept 2020, third round in
Oct-Nov 2020, fourth round in Dec 2020 – Jan 2021, and the last round in Feb-March 2021.
Data collection was completed before the recent uptick of the pandemic that began in April 2021
and the lockdown restriction put in place to contain it.
13
In the final round, 378 households from the LSMS+ sample were successfully interviewed with
290 households interviewed in all five rounds. Likewise, 1309 IDPoor households were
interviewed in the last round with 397 households interviewed in all five rounds. Sampling
weights for the IDPoor sample were computed to ensure the representativeness of IDPoor
households at the national and regional level in each round.
The HFPS was implemented using Computer Assisted Telephone Interview (CATI) techniques
and the questionnaire was programmed using the Survey Solutions CAPI software package.
Enumerators used mobile phone devices. Enumerators were given data bundles, allowing for
internet connectivity and for daily data transfer and synchronization with the server. Field
supervisors reviewed the survey responses with enumerators via one-on-one calls daily and
addressed concerns that arose immediately following enumerators interview. At the same time,
a research analyst was in charge of checking the uploaded data daily to identify errors so as to
inform the field supervisors and enumerators.
Source: Base on: Karamba et. al (2021), The Socioeconomic Impacts of COVID-19 on Households in Cambodia: Results from the High-
Frequency Phone Survey of Households Round 4 (17 December 2020-12 January 2021), Phnom Penh. Additional information on the surveys
also available at https://www.worldbank.org/en/country/cambodia/brief/monitoring-the-impact-of-covid-19-on-households-in-cambodia
2. CAMBODIA’S RESPONSE TO THE COVID-19 PANDEMIC
Expecting the large and negative impacts of the pandemic, the RGC quickly introduced several
fiscal stimulus packages in its annual budget of 2020 as part of its response countering the income
and financial losses amongst the distressed individuals. The total fiscal cost of RGC’s COVID-19
response is estimated at US$698 million with a breakup of US$504 million allocated to COVID-
19 health response supporting the poor and vulnerable, wage subsidy and training and support to
the Rural Development Bank and SME Bank; and US$194 million provided as tax reliefs to the
garment, footwear, and tourism industry. In particular, it includes wage subsidies to the those
engaged in the worst hit industries namely construction, garment manufacturing and tourism
(unemployment benefits amounting to US$70 per month - US$40 paid by the government and
US$30 paid by the factory of suspended workers of garment/footwear/travel goods sector; and 20
percent of minimum wage entirely paid by the government to suspended workers of the tourism
sector for skill upgradation); expanding the Cash for Work for development of public and rural
infrastructures; financial assistance to small and medium scale enterprises;
However, the early response measures largely excluded most of the affected population and
primarily, the poor and the vulnerable population who tend to face the brunt of the crisis. To
support such population, the RGC launched the “Cash Transfer for the Poor and Vulnerable
Households during COVID-19�? (CTP-COVID) program which remains the primary focus of this
report.
14
a) CASH TRANSFER PROGRAM
The RGC established and launched the “Cash Transfer for Poor and Vulnerable Households during
COVID-19�? (CTP-COVID) program through a sub-decree on June 12, 2020. The procedural,
institutional arrangements and governance structures for the program was articulated in
accompanying prakas released by the Ministry of Social Affairs, Veterans and Youth
Rehabilitation (MoSVY), the lead implementing agency, on June 15, 2020. Beneficiaries were
expected to receiving cash transfers beginning June 25, 2020, within a couple of weeks of program
establishment.
The CTP-COVID aims to provide cash transfer to the poor and vulnerable population. The choice
of cash as an instrument of assistance was informed by the urgency to deliver assistance in the
hands of the beneficiaries. In addition, recent experience of the RGC with the launch of the CTP-
PWTC also provided an efficient way to deliver cash to the hands of the beneficiaries. As with the
CTP-PWTC, beneficiaries are enrolled with a mobile-payment vendor using their IDPoor cards.
Beneficiaries cash the transfer payments from the nearest agent of the mobile-payment vendor
every month for the duration of the program (see section 4 for more details). The existence of a
large payment provider with agents across the country, and the ease of delivering cash made it
much more convenient for the government to use cash as the instrument for providing assistance.
The CTP-COVID provides cash transfers to the IDPoor families. The existence of a database that
identifies poor families, the IDPoor database, was the biggest factor behind using IDPoor as the
targeting mechanism for assistance. The advantage of this database is that the information on who
is identified as ‘poor’ is readily available. In addition, the IDPoor database includes information
on the level of poverty (IDPoor 1 and IDPoor 2), as well as information on household
demographics. This made it easier for the government to clearly identify the beneficiaries as well
as calculate the tentative fiscal cost of implementing the program. It was widely noted that the
IDPoor database does not identify those who were newly poor or were adversely affected by the
COVID-19 crisis and needed assistance. To partially address this concern, the OD-IDPoor system
was launched nationwide together with the program so add (or remove in some cases) beneficiaries
to the program. Indeed, during the first nine month of program implementation, the number of
IDPoor families increased from 560,000 to 710,000 families by March 2021.
The CTP-COVID provides cash amounts according the vulnerabilities experienced by the IDPoor
households. IDPoor 1 household (extremely poor) receive more than IDPoor 2 (poor) households
reflecting the underlying need of the extremely poor. Similarly, the minimum amount received by
a household in Phnom Penh is higher than other urban areas, which in turn is higher than rural
areas. This reflects the higher cost of living in urban areas and in the capital. On top of this,
households get an additional amount for each additional member, reflecting the association
between poverty and household size. Households get additional top-up for each child, elderly, and
disabled member. This reflects additional vulnerabilities associated with having a children, elderly,
and disabled member in the household. Likewise, households with a HIV positive member get
additional transfer. Table 1 shows the exact amounts received across these categories.
The CTP-COVID was initially announced for a period of two months but was extended multiple
times with the latest extension (announced in March 2021) extending the program till June 2021.
15
That is, each beneficiary household received uninterrupted cash transfer every month from June
2020 to June 2021. The program may further be extended given the ongoing adverse impact of the
pandemic.
Table 1: Amounts of payments under the CTP-COVID
Type Target Urban (Phnom Penh) Other Urban Rural
IDP 1 IDP 2 IDP 1 IDP 2 IDP 1 IDP 2
Each HH $30 $30 $30 $30 $30 $20
Each HH member $13 $9 $10 $7 $6 $4
Additional top-ups for vulnerable members
Each child 0-5 $10 $7 $10 $7 $6 $4
Each disabled
$10 $7 $10 $7 $6 $4
member
Each elderly (60+)
$10 $7 $10 $7 $6 $4
member
Each person living
$10 $7 $10 $7 $6 $4
with HIV
Source: Prakas released by the Ministry of Social Affairs, Veterans and Youth Rehabilitation (MoSVY) on June 15,
2020
b) COVERAGE OF THE CTP-COVID
Due to the way the program is set-up, the coverage of the CTP-COVID is essentially the same as
the coverage of the IDPoor program. Hence, the program inherits the advantages and limitations
related to the coverage of the IDPoor program. We first assess the coverage of the ID poor program
and move on to the coverage of social assistance more broadly.
The Cambodia Socio-economic Survey (CSES,2019-20) is used as a benchmark to assess the
performance of the IDPoor program by examining the coverage of the IDPoor or Equity cards
ownership across those population groups in need of support: deciles of per-capita consumption.
A comparison of the IDPoor card ownership from the CSES data to the actual number of families
owning the IDPoor cards indicates significant underreporting of Equity card ownership in the
CSES data. In the survey, only 10 percent of the households reported owning a card as opposed to
around 15 percent (early 2020) to 20 percent (by March 2021) as per the number of families in the
IDPoor database. To address this, our analysis includes both the coverage according to the survey
as well as coverage adjusted for the larger number of IDPoor families in the administrative data
(Figure 4).
16
Figure 4: Coverage of the IDPoor program by consumption decile
Source: World Bank Staff estimates based on the CSES 2019/2020.
Note: The adjusted figure matches the share of IDPoor in the population as per the administrative coverage of the IDPoor
database in March 2021 (20 percent). The share of IDPoor households in the CSES 2019/2020 only 10 percent. The adjustment
is done proportionally so that the percent increase is the same across all consumption deciles.
As mentioned earlier, the IDPoor process includes community validation and judgement, along
with poverty scorecard to identify households that are “deserving�? of poverty targeted support. As
Figure 4 shows, the coverage of those in bottom decile is much higher than those at the top decide
suggest the some degree of success of the targeting process on average. However, it still leaves
some room for improvements with regards to reducing exclusion and inclusion errors. Only 56
percent of those in the poorest decile are included (exclusion error) and 7 percent of the richest are
also included (inclusion error). While no targeting system is perfect, previous analysis by the
World Bank (2020) shows that technical improvements – such as switching to a regression based
PMT scoring system – and increased coverage of the IDPoor program can address many of these
issues.10
Recognizing the importance of increased coverage in reaching more of the poor, the government
rapidly expanded the coverage of the IDPoor program after the program was announced. During
the inception of the CTP-COVID, the IDPoor database classified around 540,000 households
(about 15 percent of all households in the country) as poor (IDPoor 1 and IDPoor 2). The inherently
dynamic On-Demand Identification Process (OD-IDPoor), which allows for continual updating of
the database with newly poor households including those impoverished by the COVID-19 crisis,
was implemented with zeal (see Box 2). With the support from the local governing bodies and the
commune councils, additional 153,050 families were added to the beneficiary database of the
IDPoor program during the first 11 months of program implementation (July 2020- May 2021).
The OD-IDPoor program continues to identify more poor families through this process. As an
10
The Ministry of Planning is currently working with the World Bank and other development partners (GIZ and UNDP) to improve the IDPoor
instrument and the scoring system.
17
illustration of the advantage of increased coverage, Figure 4 shows that when the coverage of the
program increases from 10 percent (survey estimate) to 20 percent (current estimate), the coverage
of the poorest decile increases from 31 percent to 56 percent.
Box 2: The On-demand IDPoor identification process.
The quick scale-up of the on-demand IDPoor (OD-IDPoor) process helped the government to
expand the coverage of the identified poor families by over 153,000, or 27 percent, over the
course of the first year of CTP-COVID implementation. Local autonomy in beneficiary
identification and fully digitized data collection and beneficiary determination processes
enabled the government to continually identify additional IDPoor families throughout program
implementation.
The following steps presents a stylized summary of the OD-IDP process:
1. Commune Councils prepares an initial list of potentially poor families each month. This
initial list is generated based on local knowledge about families in the community.
Information on the potentially poor families comes from village chiefs, civil society
organizations working in the communes, or other social program implementers. This
step is solely done at the discretion and judgement of the commune councils.
2. The communes collect data on the potentially poor families using smart devices (tablets)
provided to each commune. The devices have a specific application pre-loaded with
relevant data collection instruments. Use of smart devices not only makes the OD-
IDPoor process faster, but also makes it less prone to errors. Intuitive user interface,
instructive error messages, logical skip patterns, and automatic score calculations helps
simplify the process and reduce errors. The local implementers have been trained to
effectively use the tablets and the application.
3. The tablets automatically determine beneficiary status and the data transferred to the
national IDPoor database. Once the data is entered into the application, the application
automatically determines the status of the household as ‘extremely poor’ (IDPoor 1),
‘poor’ (IDPoor 2), or ‘non-poor’ based on the scores. The tablets are automatically
synced with the national IDPoor database and the information is updated in real time.
4. For households classified as IDPoor, their pictures are taken and the process of obtaining
the IDPoor card is initiated.
For the CTP-COVID program, the OD-IDPoor process is directly linked with the beneficiary
registration process (Section 3) so that newly identified poor families can start receiving the
program benefit right away.
Consequently, the coverage of social assistance (including other social assistance beyond the CTP-
COVID) reflects similar pattern to that of the coverage of the IDPoor program. Only 24 percent of
all households reportedly received any kind of social assistance during the pandemic (Figure 5).
This is indicative of very sparse coverage of a social safety net particularly during the pandemic
which exposed a large proportion of the population to economic hardships. As with the coverage
of the IDPoor program, only about half of those in the bottom quintile received any social
18
assistance whereas about 8 percent of those in the top quintile also received social assistance. This
reflects similar degree of success (and failures) to target the needy as with the IDPoor program.
Similar pattern exists when looking at coverage by self-identified socio-economic status. The
lower coverage of social assistance presumably hurt the poorest the most, and their exclusion can
have long-term detrimental impacts.
Figure 5: A quarter of the households receive some form of social assistance throughout the
pandemic
Source: World Bank staff estimates from the Cambodia HFPS survey
Note: Sample restricted to the LSMS+ Sample
Noticeable inter-regional disparity is present with the rural households being more likely to receive
social assistance. Likewise, households in the low lying Tonle Sap basin or in the plateaus and
mountains have a higher likelihood of receiving social assistance than the households in the plains,
coasts, or the urban areas. This is likely driven by higher poverty rates, and hence greater need of
support, in those regions.
After the launch of the CTP-COVID, it has dominated the social assistance landscape in the
country. In May 2020, prior to the launch of the program, only 13 percent of the households
received any social assistance mostly in-kind form (Figure 6). These were probably driven by
small-scale social assistance programs implemented at the local level either by the local
governments, development partners, NGOs, or other bodies. As the CTP-COVID gained
implementation momentum, it occupied a larger share of social assistance received by the
households. By February 2021, 96 percent of the households that received any social assistance,
received it through CTP-COVID.
Nevertheless, the overall coverage of the CTP-COVID program remained quite low and covered
less than a fifth of the households in the country. the program eligibility was determined by ID
Poor program and hence inherits the limitation of the program. Though a larger share of the poorest
19
were included, many of the poorest were also excluded from the program. More importantly, the
share of the population affected by COVID-19 is much larger than the reach of the program. As
we will see later, Cambodia's coverage of the CTP-COVID is also lower compared to global and
regional competitor countries.
Figure 6: CTP-COVID cash transfer dominates social assistance after June 2020
Source: World Bank staff estimates from the Cambodia HFPS survey
Note: Sample restricted to the LSMS+ Sample
c) ADEQUACY OF THE PROGRAM
The amount of cash provided through the CTP-COVID exceeds the typical amount of cash
provided through any other social assistance program in the country. For instance, the CTP-PWYC
provides a total of $190 over the course of two years, much smaller than the amounts provided
under this program.
To examine the adequacy of the transfers, we look at what share of consumption the transfer
amounts to across different consumption deciles in the CSES 2019/2020 data. The benefits
schedule as indicated in Table 1 and the household composition in the country (according to the
CSES data) implies an average monthly transfer of $15 per person for IDPoor 1 households and
$12 per person for IDPoor 2 households.
These transfer amounts are sizable proportion of the consumption of Cambodian households
(Figure 7). For instance, the transfers to IDP1 families translates to about 57 percent of total food
consumption and 29 percent of total consumption of households in the poorest decile. The amounts
are 40 percent and 20 percent of food and total consumption respectively of households in the
20
second poorest decile. Likewise, the transfers to IDP2 families translates to about 44 percent of
food consumption and 23 percent of total consumption of households in the poorest decile.
Figure 7: CTP-COVID transfer amounts are sizable shares of the consumption of the poor
Source: World Bank staff estimates from the CSES 2019/2020
Though it is difficult to estimate the loss in consumption resulting from the COVID-19 crisis in
Cambodia, the analysis provides some comfort that the transfer amounts would be adequate to
sustain a minimal standard of living for most of the recipient households. The average CTP-
COVID benefit for IDPoor 1 is equivalent to 21 percent of the proposed national poverty line (44
percent of the proposed food poverty line). Likewise, the average benefit for IDPoor 2 households
is equivalent to 17 percent of the proposed national poverty line (35 percent of the proposed food
poverty line). Simulation exercises show that this program had the potential to the poverty rate by
3.3 percentage points compared to the pre pandemic level of consumption and poverty.11
d) CAMBODIA’S RESPONSE IN THE CONTEXT OF GLOBAL AND REGIONAL RESPONSE
Before moving to the implementation mechanisms and the results of the Cambodia's CTP-COVID,
we reflect Cambodia's response in the context of broader global and regional response to the
pandemic. Globally, over 215 countries have implemented about 1,414 social protection measures
11
The simulation exercise assumes a coverage of 20 percent in the population and has a similar inclusion/exclusion pattern as in Figure 5. The
exercise also assumes that beneficiaries with lower consumption receive the IDPoor 1 benefits and others receive IDPoor 2 benefits. As an aside,
if the IDPoor 1 benefits levels were made universal and consumption patterns were as of CSES 2019/2020 (mostly without COVID-19 impacts),
the poverty headcount would fall by 9.5 percentage points. Similarly, a universal IDPoor 2 level benefits would reduce the poverty headcount by
7.9 percentage points. However, the actual simulated effect are much smaller as the IDPoor program leaves out many poor households as discussed
above.
21
(social assistance, social insurance, and labor market programs) to combat the adverse impact of
the pandemic.
Social assistance measures, particularly cash transfers, were the most popular social protection
response during the first year of the COVID-19 pandemic. Social assistance schemes
predominantly account for 62 percent of the global response and 90 percent of emergency response
in low-income settings with cash transfers emerging as the most widely used form of social
assistance and currently covers close to 16 percent of the total global population by December
2021.12,13 Cash transfers accounted for 34 percent of global responses and targeted 1.55 billion
people or 14 percent of the total global population. More than 100 countries announced that they
would scale up these payments but not all were able to do so and others did not meet their initial
targets. Every country in East Asia, excepting Lao, has mounted a cash transfer response during
the pandemic.
Figure 8: Income per-capita and coverage of the cash transfer programs
Source: G2Px database
There were three types of cash transfer responses. The first was vertical expansion which refers to
adding benefits (42 programs in 33 countries) or payments (37 existing programs in 26 countries)
among existing beneficiaries of existing programs. The second was the horizontal expansion which
refers to scaling up of existing program (12 programs in 10 countries). The third is creating
12
Ugo Gentilini, Mohamed Almenfi, and Pamela Dale, 2020, Social Protection and Jobs Responses to COVID-19: A Real-Time Review of
Country Measures
13
Over 117 countries employed labor market programs in response to the pandemic: wage subsidies (122 programs in 93 countries), skill training
(29 schemes) and adjustment to labor market regulations (34 schemes). Social insurance interventions were implemented by 136 countries and
largely took the form of unemployment insurance (103 schemes), paid sick leave (85 schemes), subsidization of social security contributions (77
schemes), pensions (57 schemes) and healthcare insurance (24 schemes).
22
completely new programs as in the case of Cambodia. These horizontal expansions and new
programs were unprecedented in many countries with some programs reaching more than 80
percent of the population. Figure 8 shows the coverage achieved by 86 developing countries.
Cambodia was able to move quickly with its cash transfer response, exceeding its initial targets
for expanding coverage by adding new households to the IDPoor registry in the first few months.
However, the expansion was from a low initial base and the expansion was limited because only
about 17 percent of the population was in the database. This was higher than many countries but
lower than more than half the countries in Figure 8.
Some of the largest expansions were in East Asia and there does appear to be a relationship
between the size of the response and the economic impact of the pandemic both in terms of the
population covered and the magnitude of spending. This can be seen in Figure 9 which shows that
the responses were highly correlated to the decline in real GDP in 2020. Cambodia’s coverage
expansion and increase in spending was modest compared to most of the region, but its recession
was also less severe than most.
Figure 9: Economic growth, coverage, and expenditures in cash transfer programs in EAP
Source: IMF (2021), Cambodia National Accounts, G2Px database
Two things stand out from this international comparison. The first is that Cambodia's response in
terms of the coverage is lower than most comparator countries globally as well as regionally. It is
also lower than what would have been expected given the economic shocks of the pandemic that
Cambodia faced. The second is that, the additional resources that Cambodia brought in for its
social assistance measure (as a share of GDP) is lower than other countries in the region, but
somewhat consistent with the lower economic impact of the pandemic in the country.
3. IMPLEMENTATION OF THE CTP-COVID
a) INSTITUTIONAL ARRANGEMENTS
23
The CTP-COVID program follows similar institutional arrangements and implementation
mechanism as the CTP-PWYC program. Similar to the CTP-PWYC, the program is jointly
managed and implemented by the Ministry of Planning (MoP) that is responsible for the
identification and targeting of the beneficiaries through the IDPoor program and the Ministry of
Social Affairs, Veterans and Youth Rehabilitation (MoSVY), the lead implementation agency in
collaboration with the National Council for Social Protection, Ministry of Economy and Finance
(MEF), Ministry of Interior, Ministry of Planning, and the Sub-national administrations. The
following describes the key responsibilities of each of the key agencies.
Roles and responsibilities of various agencies
The National Social Protection Council (NSPC). The NSPC facilitates collaboration with
relevant line ministries and agencies to implement the program, provides policy direction and
strategies to implement the program, and conducts high level monitoring of program
implementation.
Ministry of Social Affairs, Veterans and Youth rehabilitation (MoSVY). MoSVY is the lead
implementing agency and is responsible for leading, managing, and coordinating program
implementation. MoSVY also prepares action plans, budget plans, and develops implementation
mechanisms for the program. MoSVY is intensively involved in all steps of the program
implementation including outreach to potential beneficiaries, collaborate with payment providers,
coordinate with relevant line ministries and agencies on various aspects of program
implementation, and provide appropriate technical and implementation support to its provincial
departments (PDoSVY). It also prepares regular progress reports on program implementation.
Ministry of Economy and Finance (MEF). MEF controls the overall budgetary aspect of the
program. More specifically, it decides on budget plan for program implementation, coordinates
budget allocation, reviews performance reports and budget execution, and participates in
monitoring and evaluation of program implementation.
Ministry of Planning (MoP). As discussed above, MoP manages the IDPoor program and the
database and is responsible for all aspects of beneficiary identification. Specifically, it provides
IDPoor household data for budgetary purposes including benefit determination of the households,
supports commune/sangkat in implementing mechanisms for OD-IDPoor program including
entering the collected data into the IDPoor database, addresses issues and complains on beneficiary
identification processes, and issues IDPoor cards to qualifying households. MoP also collaborates
with line ministries and agencies and development partners to support relevant aspects of program
implementation, monitoring, and evaluation.
Ministry of Interior (MoI). MoI manages the sub-national level of program implementation and
monitoring. It includes strengthening and adapting existing mechanisms for program
implementation, conduct outreach and aware programs, and guides sub-national administration to
appropriately conduct beneficiary identification based on the technical instructions from the MoP.
Sub-national administrations. The sub-national administration of MoI (capital/provincial
administration, municipality/khan/district administrations), MoSVY (PDoSVY), MoP (Provincial
Department of Planning) respectively carry out and implement their respective functionalities at
24
the local level and assist commune/sangkat administration to implement the program. The
commune/sangkat administrations are responsible for program implementation at the ground level.
Their key responsibilities are to conduct outreach among potential beneficiaries and stakeholders,
validate the identity of potential beneficiaries whose data is being collected by the IDPoor
program, confirm the provision of IDPoor card to qualified households in the regular as well as
through the on-demand process, receive and address grievances, monitor program implementation
and resolve any issue that may arise. Commune/sangkat administrations also report the progress
of program implementation to other sub-national administrations who in turn report to their
respective line ministries.
Program monitoring
The program monitoring mechanism utilizes the existing mechanism of the CTP-PWYC program.
Program monitoring is led by MoSVY and supported by PDoSVY and commune/sangkat
administrations. Daily program implementation is monitored through an online information
management system. MoSVY is also responsible to conduct surveys to assess the impacts of the
program and prepare results and implementation report for the NSPC. The PDoSVY, in addition
to monitoring daily implementation progress, conducting monitoring surveys, coordinating local
stakeholders for program implementation, also conducts regular monitoring visits and record the
results in the online monitoring system. PDoSVY visits at least 10 commune/sangkats and at least
1 percent of the beneficiaries each month for monitoring. The commune/sangkats monitor the
participation of the beneficiaries in the program, monitor identity verification and ensure accuracy
of information entered in the IDPoor database, and report challenges and issues that cannot be
addressed at the local level to PDoSVY.
Grievance redressal mechanism
The mechanism to address grievances depends upon the stage in the delivery chain to which the
grievance pertains to. Grievances after the verification and registration process is handled through
the same mechanism put in place by MoSVY for CTP-PWYC. Complainant may submit written
or verbal complain to the commune/sangkat administration, or PDoSVY, or MoSVY. Complaints
received by commune/sangkat are reviewed, investigated, and addressed by commune/sangkat
chief within a day. Any complaints that cannot be addressed are passed on to PDoSVY. Complaints
received by PDoSVY are handled by the program executing official of the PDoSVY. Any
complaints that cannot be addressed at this stage are passed on to MoSVY through the information
system. Any complaints received by MoSVY are reviewed, investigated, and addressed by the
working group responsible for program implementation. The resolution provided by MoSVY is
final. After a complaint is resolved, the complainant is notified through commune/sangkat
administration through the existing grievance mechanism established for CTP-PWYC.
Any grievances related to the IDPoor cards or the beneficiary identification process is addressed
by the grievance mechanism for IDPoor program established by the MoP. Any issues related to
the verification and validation of IDPoor cards are addressed by the commune/sangkat
administrations.
25
b) DELIVERY SYSTEM FOR CTP-COVID
The delivery chain of the CTP-COVID is adapted from the delivery system for the CTP-PWYC
program. The following steps describe the process is some detail.
Outreach. Since this is a national program, outreach and communications is done by all relevant
agencies at various levels. Program eligibility and beneficiary criteria are clearly communicated
across the country. Specific approaches included messages, interviews, and public service
announcements broadcasted through TV, radio, as well as online through social media platforms.
Beneficiary identification. Beneficiaries are identified via the IDPoor system. Households that
already have a valid IDPoor card (IDPoor 1 or IDPoor 2) are automatically eligible for the
program. Additional beneficiaries are identified via the OD-IDPoor system. Commune/sangkat
councils prepare a list of potentially poor households based on the information gathered at the local
community levels through village chiefs or other civil service organizations operating in the
community. As mentioned earlier, detailed information is collected and centrally stored by the
MoP for all IDPoor cardholders in the IDPoor database.
Verification. Commune/sangkat chief verify the names of the IDPoor cardholders in their
communes. Potential beneficiaries with a valid IDPoor card go to the commune/sangkat hall for
verification. Recently issued IDPoor cards have a photo which is used for verification. For those
with an IDPoor card without a photo, Khmer ID cards along with family books or residential books
with details of the family members are used for verification.
Registration. Once the IDPoor cardholders and their family members are verified, the
commune/sangkat enters the IDPoor card numbers, along with photos of the cardholder and their
family members along with other relevant information in the digital application (app) developed
for this program. The digital system then sends the information over to the payment vendor (Wing).
Payment account setup. Wing, a leading mobile payments provider in Cambodia, was selected
as the sole payment service provider for the program due to its network of more than 9 thousand
payment agents across the country as well as its experience in providing payments under the CTP-
PWYC. The KYC (know-your-client) related verification is done by the commune council
members during the registration phase in presence of the wing agent along with the council
members and eligible beneficiaries. A virtual account is then activated by Wing for each
beneficiary. This account is not a fully-fledged mobile money account and does not require a phone
number to establish. This account is solely dedicated for the purposes of providing regular
payments for the CTP-COVID. The IDPoor card of the beneficiaries are linked to the account,
which they can use to collect the transfer payments every month.
Benefit amount determination. Benefits received by each eligible household is straightforward
and as indicated in Table 1. Commune/sangkat officials communicate the benefit amounts to be
received by the household during the registration phase and explain to them the procedure,
date/time, and location to cash out the payment. Wing also has access to the relevant fields of the
IDPoor database so that the exact transfer amount can be transferred to the beneficiary accounts.
26
Benefit delivery. Benefits are distributed each month through Wing agents located across the
country. On the designated date of disbursement, beneficiaries go to the agent’s counter along with
their IDPoor cards. The agent verifies the identity of the beneficiary using the IDPoor card number
and the information present in the database. Only eligible household members who were registered
with a photo ID can withdraw cash. The agents then take a picture of the beneficiaries with their
IDPoor cards and uploads into the information management system as a proof of payment delivery.
This process repeats every month. During their first visit, upon proper verification, the Wing agents
ask beneficiaries to generate a password for their account. In the subsequent visits, beneficiaries
cash out the transfers with their IDPoor card and using the account password.
As seen in this section, the institutional arrangement and the delivery system for the CTP-COVID
program builds on existing programs and mechanisms. A few strengths of this approach are:
• Existence of a beneficiary identification mechanism in the form of the IDPoor program.
The regular mechanism put in place across the country for several years meant a readily
available database to target assistance for the CTP-COVID. The process for identifying
beneficiaries had been in place for several years, and the newly piloted on-demand system
was scaled up to identify more beneficiaries during the program duration. As the on-
demand system relies on tablets to collect data, information is updated in real-time with
the central database.
• Strong coordination across various implementing agencies. Successful implementation of
the program requires successful coordination and cooperation across various implementing
agencies. Thanks to the strong commitment of the central government and various agencies
towards providing quick assistance to the poor and vulnerable, such coordination and
cooperation was possible. The coordination extended from program implementation in the
ground to real-time data sharing through APIs and online information management system
that spanned across different agencies and stakeholders.
• Existing mechanism for cash delivery through a mobile payment vendor. The mechanism
that was set up for CTP-PWYC helped establish a good working relationship on the ground
among the IDPoor program, local commune/sangkat administrations, and MoSVY. This
mechanism allowed the government to quickly put in place a complex institutional
arrangement for the CTP-COVID program. In addition, the pre-existing relationship with
Wing, the payment vendor, for the CTP-PWYC quickly allowed the government to adapt
the arrangement to the CTP-COVID.
• Existing mobile payment provider with large coverage. One of the most important features
that allowed for quick disbursement of the payments is the existence of a mobile payment
provider with a large coverage. Mobile payments meant that cash could be disbursed from
the central authorities to the vendors and then onwards to the agents in the village
instantaneously. The wide coverage of the payment vendor across the country meant an
easy access for the beneficiaries to a payment cash-out point. As we will see later, there
are some minor but sizable gaps in access to payment points which could be improved.
However, given the need to deliver payments urgently, the choice of scaling up existing
relationship with the payment provider is a pragmatic one.
27
4. RESULTS AND IMPACTS
a) REACH OF THE PROGRAM AMONG INTENDED BENEFICIARIES
Administrative data show high reach of program among intended beneficiaries
Since the inception of the CTP-COVID relief transfer in June 2020 the number of eligible IDPoor
households consistently improved on account of the OD-IDPoor process which, by May 2021,
covered over 715,000 families (about 20 percent of the households in the country). Within the
population for which the relief transfers were meant for, 94 to 97 percent of the eligible households
received the CTP-COVID transfers (Figure 10). This suggests an excellent implementation of the
CTP-COVID with only 3 to 6 percent of the eligible IDPoor families excluded from receiving the
transfers.
Figure 10: Almost all of the intended IDPoor families received the CTP-COVID relief transfer
Source: Administrative reporting data obtained from MoSVY
In addition, the administrative cost of the CTP-COVID was also low. The program provided $44.6
per family per month with a total fiscal cost of about $29.0 million per month.14 From an efficiency
perspective, the program attempts to minimize leakages with operational expenses constituting an
insignificant proportion of the total fiscal cost and almost all the expenses are directly transferred
in the form of cash benefits to the beneficiaries. The average share of total expenditures to the
14
Based on administrative reporting data till May 2021.
28
payment vendor (Wing) during the first nine months of program implementation is a negligible
0.67 percent.
Survey data corroborates the high coverage among the eligible IDPoor households
The almost universal reach of the CTP-COVID transfers is not an artifact of administrative data
and reporting. Similar pattern emerges in the household survey data as well (Figure 11). For
instance, in February 2021, 96 percent of households with a valid ID poor card reported receiving
any social assistance, very similar to the administrative numbers. The proportions are only slightly
smaller in the earlier rounds of the survey but are unlikely to be statistically different from the
administrative figures. This lends confidence to the efficiency with which the government
implemented the CTP-COVID.
Figure 11: High coverage of social assistance among eligible IDPoor households
Source: World Bank staff estimates based on Cambodia HFPS
Note: Eligible sample are households with a valid IDPoor card.
Since the HPFS had a sample of the beneficiaries of the CTP-PWYC program, a look into the
reasons why some of them did not receive the CTP-COVID transfers reveals some outstanding
issues with program implementation. The data show that, among those who did not receive a
transfer, a fourth of them did not receive because they were not aware of the program or did not
believe that the program was intended for them (Figure 12). Another 46 percent had issues with
their IDPoor cards (either they did not have the card or had a card that was expired or invalid).
These potential beneficiaries did not even register for the program because of these reasons. These
reasons were particularly prominent in urban areas of Phnom Penh, the coastal areas, as well as
the poor regions that include the plateau and the mountains. Among those who tried to register for
29
the program, most did not meet the program criteria or were disqualified. Disqualification or not
meeting the program criteria what is particularly an issue among the households with the lowest
socioeconomic status with these reasons.
This suggests that, even though the CTP-COVID did a good job of reaching those who were
registered, better outreach and flexibility in the enrollment process could have benefited many and
improved the overall reach of the program.
Figure 12: Awareness and card validity issues were most prevalent among those who did not
receive the transfers
Source: World Bank staff estimates based on Cambodia HFPS.
Note: Only IDPoor sample from Feb-Mar 2021 used for estimation.
Grievances were not very common among non-beneficiaries
The Cambodia HFPS asks the IDPoor sample (beneficiaries of the CTP-PWYC who are expected
to be ‘poor’) that did not receive transfers under the CTP-COVID on whether they complained
about this to the authorities. Only 12 percent complained about not receiving the benefits to the
authorities (Figure 13). Two thirds complained to the village chief and rest complained either to the
commune chief (22 percent) or to the commune council (10 percent). About 44 percent of the
complainants in Feb-Mar 2021 indicated that the complaint what successfully resolved: either they
received or were in the process of receiving the transfers (23 percent) or were provided with the
reasons of not receiving the transfers (22 percent). About 37 percent indicated that the authorities
did not take any action to resolve their complaints.
30
In all areas except in Phnom Penh and urban areas, the share of household that complained fell
over the course of the year. This suggests that the share of complaints fell as some of them were
successfully resolved to the satisfaction of the complainant. Indeed, the rate of successful
complaint resolution also fell from 64 percent in Oct-Nov 2020 to 44 percent in Feb-Mar 2021
indicating that more difficult cases remained which could not be successfully resolved. The share
of positive resolution of the complaints remained consistently lower in Phnom Penh in urban areas.
This indicates that grievances and redressal processes are less effective in larger and more dense
urban areas.
Overall, this indicates that, while grievance mechanisms were put in place for the program, its
usage has been somewhat limited. This could indicate a lack of awareness among potential
beneficiaries about the grievance mechanisms that they could resort to. Even among the grievances
that were received, a sizable share resulted in no visible action taken towards resolving them. This
could have further lowered the complaints from potential beneficiaries as some of them did not
expect it to lead to a positive resolution.
Figure 13: Grievance mechanism and Conflict Resolution of the CTP-COVID
Source: World Bank staff estimates from Cambodia HFPS.
Note: Only IDPoor sample used for estimation.
b) BENEFICIARY USE AND EXPERIENCE WITH THE CTP-COVID
Transfer frequencies and amounts are consistent with program design
Data from the Cambodia HFPS show that the beneficiaries have been receiving cash transfers as
expected. Around 68 percent of the recipients had received nine installments by the time of the
Feb-Mar 20021 surveys, as would have been expected. Only around 9 percent of the households
received less than five or fewer installments possibly due to late registration. Some of this variation
is also driven by the transition of households in and out of the IDPoor database due to continual
verification and updating of the IDPoor database which happened throughout the program period.
At the time of the survey, an average recipient household had received US$ 366.
31
Figure 14: Frequency and amounts of CTP-COVID transfers were in line with expectations
Source: World Bank staff estimates based on Cambodia HFPS.
Note: Estimation restricted to the IDPoor sample
Figure 15: Average monthly transfer per family has been relatively stable across the program period
Source: World Bank staff estimates based on Cambodia HFPS.
32
The average transfer received by the households has been relatively stable throughout the program
year (Figure 15). In Feb-Mar 2021, average recipient household had received US$ 45.8 per month,
similar to the average monthly transfers they had received in the previous months. This is
statistically indistinguishable from the administrative figure of US$ 44.6 per month per beneficiary
household.15 As one would expect, the average amounts are slightly higher in urban areas (and
coastal areas) compared to other regions of the country. There is also some indication that families
with higher pre-pandemic socioeconomic status received slightly smaller amounts compared to
those with lower socioeconomic status.
A regression analysis also confirms that transfer amounts correspond broadly to the design of the
program (Appendix Table 1). For instance, larger families received more benefits than smaller
families. Households with 8 or more family members received US$ 19 more per month compared
to households with 3 or fewer members. Similarly, families with more children and more elderly
members received higher benefit through the CTP-COVID program although the association is
weak. Likewise, families with higher pre-pandemic socioeconomic status received marginally
lower amounts of benefit through the program. The association between geographical region and
benefit amounts, however, is not as strongly correlated in the data once other factors are controlled
for. Overall, the CTP-COVID relief transfer amounts in the data are generally in line with what
would be expected given the program design.
Beneficiaries found the registration and payment processes easy
The high reach of the CTP-COVID transfers and the efficient disbursement of the transfers is a
consequence of the clear and easy process of registration to the program and the payment system
as well as the ease with which beneficiaries could access the payments. Data collected in Feb-
Mar 2021 shows that 94 percent of the beneficiaries found the registration process to be either
‘easy’ or ‘very easy’ (Figure 16).
The rates were slightly lower for Phnom Penh and urban areas as well as in in the plateau and
mountainous provinces. This suggests some difficulties and conducting the registration process in
dense urban areas as well as in remote provinces. The ease of the processes to receive payments
as well as withdrawing cash was close to universal. The rates are again slightly lower in the plateau
and mountainous provinces where 5 percent of the beneficiaries did not find the process of
receiving the transfer to be easy. Together, this suggests an efficient payment delivery mechanism
with some very minor gaps, especially in very remote areas. The efficiency of the payment delivery
and withdrawal mechanism is also reflected in the fact that fewer than 1 percent of the beneficiaries
reported having to pay any fee in order to withdraw cash. It appears that the entirety of the cash
transfers went directly in the pockets of the beneficiaries as intended without any significant
leakages.
15
Based on administrative data on expenditures and beneficiary numbers till February 2021.
33
Figure 16: Most CTP-COVID beneficiaries found the registration and payment process to be easy
Source: World Bank staff estimates based on Cambodia HFPS, Feb-March 2021 round.
Beneficiaries do not indicate a strong preference for savings the transfers
Given the temporary nature of the CTP-COVID transfers, it would be reasonable to expect
beneficiaries to want to see the transfers for future usage. However, the Feb-March 2021 survey
found that only 3 percent of the beneficiaries expressed a preference to save the transfer amounts
instead of cashing them out in full (Figure 17). This rate is much lower than the share of
beneficiaries expressing a preference to save the transfers in earlier rounds of the survey which
range from 7.5 to 12 percent. There is some variation across regions and socioeconomic status, but
no clear and consistent pattern emerge that is robust across different rounds of data collection.
The overall low preference for saving the transfer could stem from the fact that the transfer
amounts are just adequate for the beneficiaries who use it to purchase their daily necessities. The
decline in preference for savings from August 2020 through January 2021 suggests that seasonality
in consumption or potentially other factors could be at play as well. One potential reason could be
the expectation about the duration of the program benefits. Perhaps, in earlier rounds, beneficiaries
may have expected the program to end soon and we have decided to extend their relief transfers
for longer period. However, after repeated extensions of the program, the beneficiaries may have
expected the program to continue in the near future and have reduced their desire to stretch the
transfers for a longer period.
34
Figure 17: Only some beneficiaries expressed preference to save the CTP-COVID transfers
Source: World Bank staff estimates from Cambodia HFPS.
Travel time to withdraw benefits is a hurdle for a sizable minority
Though the cash transfers are done through a mobile payment vendor (Wing) and that the transfer
of funds from the government to the vendor and from the vendor to their agents in the community
is done digitally, the last step in the delivery chain is still manual. That is, beneficiaries still have
to physically travel to their closest agents in order to cash the transfers. This physical process can
sometimes act as a bottleneck in an otherwise efficient delivery system. Much will depend upon
the availability of local agents in remote villages. While 61 percent of the ID poor families live in
villages with a payment agent, a significant 22 percent live in villages that are at least 5 kilometers
from the nearest payment agent. While this may not be a large distance, limited travel options for
the poor could impose significant burden and constraint in terms of travel time to cash the benefits.
Indeed, the time taken to travel and collect the payments reflect this pattern. An average household
spends about 18 minutes to travel to the nearest payment agent in order to cash the benefits, which
is a significant improvement from 23 minutes it took in Aug – Sep 2020 (Figure 19). As expected,
average travel times are much shorter in Phnom Penh and urban areas as well as in the plains.
However, beneficiaries in the coastal provinces as well as provinces in the plateau and the
mountains travel over 23 minutes on average to cash the payments.
As the figure above showing the distribution of payment agent demonstrates, there is a thick tail
of beneficiaries who spend considerable amount of travel time to catch the benefits. Over a quarter
of beneficiaries need at least 30 minutes to travel to the nearest payment agent. Again, the
geographic heterogeneity is quite large here as well. Only 16 percent of beneficiaries in Phnom
Penh and urban areas travel 30 or more minutes to cash their payments. However, over 43 percent
35
of beneficiaries in the coastal provinces and the provinces in remote plateau and mountain areas
travel 30 or more minutes to cash out their payments. Some provinces, particularly in the Tonle
Sap region, have seen a drastic reduction in the share of beneficiaries traveling 30 minutes or more.
In Aug-Sep 2020, 42 percent of the beneficiaries in this region traveled 30 minutes or more. The
share came down to 25 percent by Feb-Mar 2021. However, travel times in remote provinces have
stayed more or less the same throughout the year.
Figure 18: A sizable share of the IDPoor families live in villages without a Wing payment agent
Source: World Bank staff estimates from the data on location of payment agents.
This presents a sizeable gap in the penetration of payment agents, particularly in remote provinces.
The higher degree of unease with the payment processes of beneficiaries from these regions
discussed above perhaps stems from the lack of payment agents close to their villages.
Figure 19: A significant share had to travel 30 mins or more to receive the transfers
Source: World Bank staff estimates from Cambodia HFPS.
Note: Only IDPoor sample used for estimation.
36
c) USAGE OF THE TRANSFERS AND WELFARE OUTCOMES
As seen above, barring mostly minor issues, the CTP-COVID was successful in providing relief
transfers to most of the eligible households. In this section, we examine how the beneficiaries used
the relief transfers and how it affected their incomes and overall well-being.
The CTP-COVID was mostly used to purchase food and other essentials
Almost all the beneficiary households reported using the CTP COVID transfer to purchase food
(Figure 20). The proportion of beneficial using the transfers to buy food is universal throughout the
year as well as across geographical region or socioeconomic status. Likewise, four fifth of the
beneficiary households use the transfers purchase essential items. The share of beneficiary
households purchasing essential item has increased from 58 percent in Aug- Sep 2020 to 78 percent
in Feb-Mar 2021. The share of beneficiary households using the relief transfers to pay back loans
has hovered around 10 to 15 percent throughout the year. This pattern is consistent with the
behavior where beneficiary households first spend the transfers to fulfill their food needs and pay
back loans and, if anything remains, use it to purchase household essentials. Over time, the
continued relief transfers made it possible for many households to spend on household essentials.
Only a small share (less than 4 percent) of the beneficiary households spent the transfers for other
purposes.
Figure 20: Most households used the transfers to purchase food and essentials
Source: World Bank staff estimates from Cambodia HFPS.
Note: Only IDPoor sample used for estimation.
Stated importance of the transfer to the family and their well-being
The beneficiaries considered the relief transfers to have made large differences in their incomes
and well-being. When asked about how important the transfers were to the beneficiaries’
wellbeing, an overwhelming 78 percent mentioned that the relief transfers were either ‘extremely
important’ (for 38 percent) or ‘very important’ (for another 40 percent) to their families (Figure
21). Less than 0.2 percent mentioned that the relief transfers were not important to them at all. The
37
perceived importance of the transfers was similar across geographic regions as well as
socioeconomic status and remained consistently high throughout the year.
Figure 21: Beneficiaries consider CTP-COVID transfer to be very important
Source: World Bank staff estimates from Cambodia HFPS.
Note: Estimation restricted to IDPoor sample
The transfers also made a large difference in the beneficiaries’ wellbeing. 42 percent mentioned
that the relief transfers made a huge difference add another 37 percent mentioned that it made a
large difference. Here as well, the recognition that the transfers made a large difference remain
consistently high throughout the year as well as across geographic regions and socioeconomic
status.
Subjective importance of the transfers is corroborated by actual impact of the transfers on
household income
The Cambodia HFPS corroborates the subjective assessment of the beneficiary households that
the transfers really helped mute income shocks for the recipient households. Figure 22 compares
reported income changes from June 2020 (prior to CTP-COVID transfers) to Feb-Mar 2021
between families that received the transfers with those that did not. The comparison controls for
differences in important covariates such as household size, number of children and elderly in the
household, as well as perceived socioeconomic status prior to the pandemic.
As seen in the figure, most households (independent of whether they received the transfers) have
experienced a decline in incomes from June 2020 to March 2021 across major income sources.
For instance, farm incomes fell by 4 percent for households receiving the transfers and 5 percent
for households not receiving the transfers. The differences are statistically insignificant. Similarly,
both groups experienced a decline in wage incomes: families that received the relief transfers saw
their wage incomes decline by 7 percent, statistically similar to the decline of 5 percent experienced
by families that did not receive the transfers. Likewise, income from household enterprises fell by
about 5 to 7 percent during this period with no statistical difference between households by receipt
of the transfers. Interestingly, remittance income fell by 2 percent for households that received the
relief transfers whereas it increased by 1 percent for families that did not.
38
Figure 22: CTP-COVID transfers significantly muted income shocks for recipient households
Source: World Bank staff estimates from Cambodia HFPS Feb-Mar 2021 round.
Note: The figure plots the income changes for households between June 2020 and Feb-Mar 2021 across each measure of
income indicated in the vertical axis. The bars show the regression estimate of averages for households with and without CTP-
COVID transfers after controlling for household size, demographics, perceived socioeconomic status prior to the pandemic,
and sample composition. The error bars show 95 percent confidence intervals.
More importantly, households that did not receive the transfers saw their overall incomes decline
by 15 percent whereas households that received the transfers experienced a decline of only 5
percent. The difference in income loss of about 10 percentage points is both large and statistically
significant. This difference is clearly driven by the transfers received from the government through
the CTP-COVID. The evidence shown here is consistent with the relief transfers significantly
lowering the adverse impact of the pandemic on household incomes.
There is some suggestive evidence that the transfers led to an increase in household welfare
Ideally, the increase in incomes due to the transfers are expected to increase household welfare for
the recipient households. However, measuring the causal impact of the transfers on household
welfare is tricky. The central problem comparing households that receive the transfers with don't
that do not receive the transfers is that households that receive the transfers are fundamentally
different from those that do not receive the transfers. This is exactly the design of the program:
that only households with a valid IDPoor card are eligible for the transfers whereas others are not.
Having a valid IDPoor card indicates vulnerabilities that other households do not have. This makes
the comparison uninformative on the actual impact of the transfers.
39
Nevertheless, estimating the causal impact of CTP-COVID transfers is very informative. to adjust
for the potential biases that arise from a simple comparison, we make some adjustments to make
the comparison more causal. First, we restrict estimation only to the IDPoor sample. This sample
consists of beneficiaries from the CTP-PWYC program who either have or had an IDPoor card till
recently or were classified as being poor by the health facilities even though they did not have an
IDPoor card. Similarly, we control four other factors that affect the likelihood a household receives
the CTP-COVID transfers such as: household size, household demographics, and measures of
household socioeconomic status prior to the pandemic. These adjustments address, at least
partially, the concerns around a causal comparison.
Figure 23: CTP-COVID receipt is associated with better HH welfare for some outcomes
Source: World Bank staff estimates from Cambodia HFPS (Feb-March 2021 round).
Note: Estimation restricted to IDPoor sample. The figure plots the welfare measures indicated in the vertical axis. The bars
show the regression estimate of averages for households with and without CTP-COVID transfers after controlling for household
size, demographics, perceived socioeconomic status prior to the pandemic, and sample composition. The error bars show 95
percent confidence intervals.
With these caveats, we find that relief transfers under CTP-COVID have made significant
outcomes on some welfare indicators. In particular, as Figure 23 shows, households that received
the CTP-COVID transfers are 2 percentage points more likely to have improved wellbeing since
January 2020 (pre-pandemic) and 3 percentage points more likely to have improved wellbeing
since Jun 2020 (pre-program). However, the same extent of impact do not appear for other
measures. For instance, the index of food insecurity for transfer households is statistically identical
to that of non-transfer households. Furthermore, the optimism about household wellbeing in the
40
near future is actually slightly lower, although statistically similar, for the households that received
the CTP-COVID. These likely reflects the underlying differences between recipient and non-
recipient households to begin with. Households that receive the transfers are likely to be poorer,
even after the adjustment, and have worse counterfactual food security and prospects. The transfers
may have helped them bring up to a level that is equal to the households that did not receive the
transfers (and were wealthier in the first place).
Figure 24: Higher benefit amounts under CTP-COVID is associated with higher measures of
wellbeing
Source: World Bank staff estimates from Cambodia HFPS.
These comparisons show that there is some indication of the positive impact of the CTP-COVID
transfers, but statistical concerns outlined above prevent us from drawing a strong conclusion on
the welfare impacts of the CTP-COVID transfers.
Next, we compare welfare outcomes based on the amounts of transfers received by the households
thus far. The comparison is somewhat cleaner as it only includes households that received some
CTP-COVID transfers at some point over the past year. When we examine the outcomes of these
households, we clearly see a pattern: beneficiary households that received more money in transfers
have much better outcomes along al four welfare measures (Figure 24).
41
In summary, the analysis conducted here finds some suggestive evidence of a positive welfare
impact of the CTP-COVID transfers, and some evidence that larger transfer size led to larger
welfare gains. However, lack of appropriate data and, more importantly, appropriate counterfactual
group prevents us from making a definitive statement.
5. POLICY RECOMMENDATIONS
This report finds that the CTP-COVID program was implemented efficiently and managed to
deliver cash to the hands of the intended beneficiaries. Building on existing mechanisms developed
for complementary programs, the government was able to quickly adapt them for the purposes of
this program. the program beneficiaries where, by and large, happy with the implementation of the
program and found the processes to register and receive payments relatively easy. The transfers
under the program undoubtedly helped the recipient families too mute the adverse income shocks
due to the pandemic. The report also finds some suggestive evidence that household well-being
also improved as a result although data limitation does not allow us to fully explore such impacts.
The program provides valuable lessons towards providing strong safety net during the crisis and
for instituting a stable longer-term social assistance after the pandemic period ends. In what
follows, we describe the policy lessons emerging from the study.
a) SHORT-RUN POLICIES
Continue the support to the poor families
The CTP-COVID program has shown how effective social assistance can be in combating the
adverse effects of the pandemic. The recent increase in COVID-19 caseload in the capital as well
as the rest of the country shows that the pandemic is far from being over. Recognizing this, the
government is already preparing an assistance program to help low-income families who are
adversely affected by the health impacts of the pandemic as well as the economic impacts due to
lockdown restrictions put in place.
The experience over the past year shows that the hopes of a V-shaped recovery is fading away. It
will take a while for the key sources of jobs and growth in Cambodia to return to its pre pandemic
levels. Sectors such as tourism and garment exports have continued to suffer due to lack of demand.
Migration flows and remittances, an important source of livelihood for many, may take some time
to bounce back. In this situation, the government should continue to support the poor and
vulnerable families throughout the (somewhat long and painful) recovery process. Continued
support will prevent families from falling deeper into poverty and helps them recover quickly when
the economy bounces back.
Expand the coverage of the support to include the near poor and the newly poor
The CTP-COVID program was limited in scope by the reach and coverage of the IDPoor program
and database, which covered fewer than one in five households. As seen above, countries
worldwide have done a much better job of reaching a larger share of the population during the
pandemic through their social assistance program. The coverage of social assistance during the
42
pandemic in Cambodia lags that affects global as well as regional peers. This has become painfully
clear in the recent months when the disease caseload increased, and severe lockdown restrictions
were put in place to contain the pandemic. A lot more people where affected adversely by the
pandemic including the near-poor and the new-poor as well as many low-income families that lost
the key sources of income and livelihoods. The coverage of social assistance should extend do
these families as well.
Recognizing this, the government has recently announced a support program to low-income
families residing in areas placed in lockdown to contain the pandemic as well as low-income
families that were adversely affected by the health impact of the pandemic. The identification
process for the IDPoor program will be adapted to identify more beneficiaries for social assistance.
This is a commendable, albeit somewhat late, step in the right direction.
Explore using mobile technology to quickly identify and register potential beneficiaries
Many countries that have massively scaled up social assistance during the pandemic have often
utilized innovative technologies to identify and register potential beneficiaries. Use of technologies
is particularly relevant in this context due to its ability to be implemented at scale and limit human
to human contact and disease spread. Such avenues could be explored for Cambodia as well. Data
suggest that mobile penetration is quite high in Cambodia with over 95 percent ownership in urban
areas. The government could partner with telecommunications providers to conduct en-masse
outreach, screening, and registration of potential beneficiaries through mobile phones in specific
areas of the country.
Countries with similar or lower level of mobile penetration and related infrastructure have
managed to quickly and successfully implement such a technology aided registration process. In
Togo, for example, individuals from certain targeted areas of the country were sent an SMS and
asked to respond to a few basic questions. An automated system then determined if the responses
met the program eligibility criteria. Some countries also used cell phone usage information to filter
for higher income individuals. In Pakistan, for example, applicants with very low average monthly
phone bills over the previous six months were considered eligible.
The government is currently considering such a mobile based system for monitoring of the recently
social assistance package. But such a system can also be used to quickly identify and register
potential beneficiaries for support. Such system should be designed to complement the current and
proposed IDPoor identification processes and reduce the implementation burden at the local level
and/or speed up the registration process.
b) MEDIUM-TERM POLICIES
Develop the IDPoor program to serve as a social registry
An expansion in coverage of the IDPoor program will need to be accompanied by additional
reforms to transform the database into a social registry for all current and future social programs.
This will involve a series of reforms to make the database richer and more integrated with other
information systems.
43
First, the coverage of the IDPoor program we need to expand significantly from its current
coverage. Currently, after an enthusiastic implementation of the OD-IDPoor process over the last
year, the coverage stands at about 20% of the households in the country. Though this is a
significant increase from the coverage at the beginning of the pandemic, it is still much lower than
that of many comparable countries. For instance, the social registries in the Malaysia, Pakistan,
Philippines, and Thailand cover over 80 percent of the population; Chile, Colombia, and Indonesia
cover over 65 percent of the population. Many of these countries already had a high coverage of
their social registries, but most of them significantly expanded coverage during the pandemic.
Second, the scoring system used by the IDPoor program needs to be modernized. The current
system is based on a scorecard approach that can only classify household into IDPoor 1 and IDPoor
2. Previous World Bank study has shown that switching to a data-based scoring system such as
the proxy-means-test regression can improve the accuracy of the targeting system in identifying
the needy as well as allow the system to classify households into granular levels of poverty and
vulnerability. Such granular targeting system is particularly useful and allows the social assistance
programs to target various levels of poverty and vulnerability. It also allows for social assistance
program to quickly expand coverage in response to crises such as the current pandemic. The
government is already moving towards a streamlined version of the regression-based scoring
system for the social assistance package being prepared for lockdown areas.
Third, the IDPoor program should be linked to the national identification system which allows for
better data cross-checks an interoperability with other data systems. In many countries, social
registries regularly crosscheck information across available administrative databases such as the
civil registry, vehicle registry, property registry, or tax registry. Advanced systems directly use the
information from administrative databases to complement or to cross-validate the information
collected through surveys. While Cambodia may not have appropriate infrastructure yet to link the
IDPoor database with other administrative databases, it can still be linked to the national
identification system which uniquely identifies individuals in the country. Initiatives are currently
underway to collect the national ID cards accurately from the household heads. The collection of
national ID cards should be extended to all adult members of the household. This also allows for
better data sharing with beneficiary databases which allows for better policy planning and
budgeting.
Introduce strong and adequate social assistance programs post-pandemic.
While the CTP-COVID program helped poor and vulnerable households in the short term, more
assistance is likely to be needed post-pandemic as the adverse effect of the pandemic are likely to
linger as the economy gradually recovers. This is particularly important for Cambodia as the pre-
pandemic level of social assistance was extremely low both in terms of coverage and adequacy of
benefits. The post-pandemic level of social assistance will need to be much higher than the pre-
pandemic level both in terms of coverage and adequacy of social assistance.
Social assistance can continue to be targeted using the expanded and reformed IDPoor program.
Such assistance can begin with a high coverage which can be lowered as the economy recovers.
In addition to targeting by IDPoor program, social assistance can be disproportionately provided
to groups that the government feels are more vulnerable (for example, the elderly, children, or the
disabled). The schematic of such a design is shown in Figure 25. A simplified version of this design
44
was implemented in the CTP-COVID as well, with levels of benefits higher for IDPoor 1 compared
to IDPoor 2 and several vulnerable groups such as household with children and elderly receiving
higher amounts. With greater coverage and more granular classification of poverty and
vulnerability, social assistance can be designed to provide the level of benefits in accordance to
their need and vulnerabilities.
Figure 25: Schematic of varying benefit levels by poverty and vulnerability
Source: World Bank staff
While receiving social assistance is important, the adequacy of the benefits is also important for
social assistance to be effective and impactful. As seen above, the impact of social assistance
increases with the amount of the transfer and low transfer amounts may not be effective. While
the benefit amounts under the CTP-COVID program was a significant share of the consumption
of the poor, the same cannot be said about pre pandemic programs. For instance, the CTP-PWYC
provides a total USD 190 to women for seeking pre-natal and ante-natal care over the course of
three years. While the benefits may elicit health seeking behavior, they are inadequate to improve
maternal and child nutrition as the benefit only amounts to less than 2% of the average
consumption of the households.
From poverty targeted social assistance programs, the benefit amounts should be benchmarked to
the consumption levels and/or the consumption gap of the poor. This way of benchmarking
assistance to the actual consumption levels of the poor would help better understand the potential
impact of the assistance programs in making a tangible difference in the lives of the beneficiaries.
Larger program benefits would be much for effective at supporting the poor and vulnerable to get
out of poverty.
45
Introduce shock responsive elements to social assistance programs
The CTP-COVID program was an example of how countries can set up additional social assistance
in response to a crisis. While COVID-19 may have been a global crisis that simultaneously affected
families across the world at a scale not seen in recent history, households experience idiosyncratic
and covariate shocks regularly. As COVID-19 and response to it through greater social assistance
demonstrated, social assistance at the times of crises can really be effective in softening the blow
and help people recover from the adverse impact of the shock. Cambodia faces shocks due to
health and natural disasters (e.g. flooding) in a regular basis plunging many households into
poverty, vulnerability, and financial hardships. Cambodia should learn from the CTP-COVID
experience to extend social assistance programs to those affected by such shocks.
Together with expansion of the IDPoor program as a social registry, it could collect data on a wide
array of exposures to shocks and relevant vulnerabilities. In addition, the data in the registry could
be linked to data from local administrations, geographic information systems, or other information
systems on disasters so that when a shocks affects a particular part of the country, households can
be immediately pre-identified based on the information available in the registry so that assistance
can start flowing immediately. In essence, this recommendation is to do similar things (collect
data, identify affected beneficiaries, extend social assistance) the government did to respond to the
COVID-19 crisis through social assistance more proactively so that beneficiaries can start
receiving support sooner when future shocks, small or large scale, hit the country.
Post-pandemic social assistance programs, for instance the ‘family package’ of social assistance
that the government is currently preparing, should have an in-built shock-responsive modality
incorporated in its design. Such modality would allow for quick horizontal and vertical expansion
of eligibility as well as benefit amounts at the times of shocks and crises. This will, of course, need
to be complemented with shock-responsive financing which can be drawn-upon immediately at
times of shocks and crises.
46
6. ANNEX: ADDITIONAL RESULTS
Appendix Table 1: Determinants of social assistance receipt and benefit amounts
(1) (2) (3) (4)
Received any Received any Amount of Average monthly
SA in past 2 months relief payments relief payments amt of payments
Household size (base = 3 or less)
4 0.174*** 0.161*** 6.725 1.723
(0.0453) (0.0458) (17.65) (1.764)
5 0.182*** 0.193*** 37.13** 5.841***
(0.0460) (0.0466) (17.89) (1.788)
6 0.230*** 0.244*** 61.47*** 8.830***
(0.0497) (0.0503) (18.76) (1.876)
7 0.242*** 0.264*** 78.12*** 10.70***
(0.0554) (0.0561) (20.30) (2.030)
8 or more 0.256*** 0.290*** 116.1*** 19.23***
(0.0569) (0.0576) (20.61) (2.060)
Number of children under 5
1 0.0286 0.0170 4.973 -0.629
(0.0523) (0.0529) (18.55) (1.854)
2 0.0347 0.0144 17.07 2.422
(0.0547) (0.0553) (19.37) (1.937)
3 or more 0.0691 0.0325 49.18* 4.397
(0.0799) (0.0809) (26.95) (2.694)
Number of elderly in household
1 0.0750** 0.0831** -9.650 -1.953*
(0.0339) (0.0343) (11.05) (1.105)
2 or more 0.0498 -0.00105 49.71** 3.987*
(0.0625) (0.0633) (20.87) (2.087)
Socio-economic status in Jan 2020
(Base = 1, Poorest)
2 -0.0835** -0.0878** -13.18 -1.569
(0.0403) (0.0408) (13.22) (1.322)
3 -0.105** -0.121*** 8.908 -1.152
(0.0446) (0.0452) (14.93) (1.493)
4 -0.173*** -0.163*** -4.909 -0.290
(0.0479) (0.0485) (16.40) (1.640)
5 (average) or higher -0.132*** -0.146*** -14.19 -2.365*
(0.0365) (0.0369) (12.07) (1.207)
Constant 0.588*** 0.588*** 315.7*** 39.54***
(0.0638) (0.0645) (22.36) (2.235)
Observations 1,309 1,309 912 912
R-squared 0.048 0.053 0.110 0.209
Standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Appendix Table 2: Impact of CTP-COVID on measures of household welfare
47
(1) (2) (3)
A. HH wellbeing improved relative to Jan 2020
Received SA in past 2 months 0.0465***
(0.0171)
Received relief transfer 0.0470*** -0.186***
(0.0171) (0.0374)
Amount of transfer (in $100) 0.0701***
(0.0101)
B. HH wellbeing improved since June 2020
Received SA in past 2 months 0.0217
(0.0132)
Received relief transfer 0.0225* -0.0960***
(0.0132) (0.0292)
Amount of transfer (in $100) 0.0357***
(0.00785)
C. Index of food insecurity
Received SA in past 2 months 0.0720***
(0.0127)
Received relief transfer 0.0564*** 0.143***
(0.0128) (0.0283)
Amount of transfer (in $100) -0.0262***
(0.00760)
D. Optimistic about positive prospects in the near future
Received SA in past 2 months -0.0299
(0.0300)
Received relief transfer -0.0328 -0.218***
(0.0300) (0.0664)
Amount of transfer (in $100) 0.0559***
(0.0178)
48