                                                   WPS7505


Policy Research Working Paper                     7505




      Direct and Indirect Effects of Malawi’s
     Public Works Program on Food Security
                               Kathleen Beegle
                              Emanuela Galasso
                               Jessica Goldberg




Development Research Group
Poverty and Inequality Team
December 2015
Policy Research Working Paper 7505


  Abstract
  Labor-intensive public works programs are important                               estimate effects on food security and use of fertilizer. There
  social protection tools in low-income settings, intended                          is no evidence that the program improves food security, and
  to supplement income of poor households and improve                               some negative spillovers to untreated households. These
  public infrastructure. In this evaluation of the Malawi                           disappointing results hold even under modifications to the
  Social Action Fund, an at-scale, government-operated pro-                         design of the program to offer work during the lean rather
  gram, across- and within-village randomization is used to                         than harvest season or increase the frequency of payments.




  This paper is a product of the Poverty and Inequality Team, Development Research Group. It is part of a larger effort by
  the World Bank to provide open access to its research and make a contribution to development policy discussions around
  the world. Policy Research Working Papers are also posted on the Web at http://econ.worldbank.org. The authors may be
  contacted at kbeegle@worldbank.org, egalasso@worldbank.org, and goldberg@econ.umd.edu.




         The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development
         issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the
         names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those
         of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and
         its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent.


                                                       Produced by the Research Support Team
    Direct and Indirect Eﬀects of Malawi’s Public Works
                 Program on Food Security
                             ∗                 †
              Kathleen Beegle, Emanuela Galasso, and Jessica Goldberg‡




    JEL Codes: I31, J22, O1. Keywords: public works, food security, Malawi.



  ∗ World  Bank Ghana oﬃce. Email kbeegle@worldbank.org.
  † Development  Economics Research Group, World Bank, 1818 H St. NW, Washington DC 20433. Email
egalasso@worldbank.org.
   ‡ 3115 G Tydings Hall, Department of Economics, University of Maryland, College Park MD 20742. E-mail

goldberg@econ.umd.edu.
1     Introduction
Labor-intensive public works programs (PWPs) are common and costly social protection
tools in low-income settings (Grosh et al. 2008). These programs require beneﬁciaries to
work in order to receive a cash payment or in-kind transfer. They have been widely used
to protect poor households in the face of large macro-economic or agro-climatic shocks, due
to their relatively rapid rollout (Ravallion 1999). In addition, PWPs are widely used as
a building block of national social protection portfolios. Well-known examples include the
Employment Guarantee Scheme in Maharashtra (Ravallion, Datt & Chaudhuri 1993) and
the National Rural Employment Guarantee Act (NREGA) in India (Dutta et al. 2014), and
the Productive Safety Net Project in Ethiopia (Hoddinott et al. 2012). Such programs are
widespread – albeit not on as large a scale – in sub-Saharan Africa, where 39 of 48 sub-
Saharan African countries have government-supported PWPs (World Bank 2015).1
    Much of the literature on cash-for-work programs focuses on estimating the potential
crowding out eﬀect of the program on labor market outcomes or the extent of self-targeting for
a given wage rate or participation requirement (Alatas et al. 2013). Despite the pervasiveness
of cash-for-work programs and their considerable costs, there is limited evidence about their
eﬀectiveness at increasing or smoothing consumption for beneﬁciaries. This paper estimates
the eﬀect of Malawi’s large-scale public works program, which operates under the Malawi
Social Action Fund (MASAF). The stated objectives of the program are to improve lean
season food security and increase use of fertilizer and other agricultural inputs. Though
the PWP relaxed budget constraints by oﬀering beneﬁciaries the opportunity to earn up to
$US 44 in a country with a per capita GNI of only $US 320, we ﬁnd no indication that the
program achieved either objective.
    Malawi’s PWP has been operational since the mid-1990s and provides short-term, labor-
intensive employment opportunities to poor, able-bodied households for the purpose of en-
hancing their food security. The implementation of the program is decentralized, with fund-
ing allocated to each of Malawi’s 31 districts based on population and poverty estimates.
The food security objective is addressed through a combination of support for short-term
consumption as well as investments in fertilizer, which is intended to increase yield in the
subsequent season. Since 2004, the program has been designed to complement Malawi’s
large-scale fertilizer input subsidy program by synchronizing the availability of public works
employment with the availability of fertilizer coupons, at planting season. Malawi’s PWP
ranks fourth in population covered among all such programs in low- and middle-income
   1 Of the 19 sub-Saharan African countries for which the World Bank’s ASPIRE database has cost infor-

mation, spending on PWPs totals more than one-tenth of a percent of GDP in nine, and accounts for more
than 10 percent of the country’s social protection budget in eight.




                                                  2
countries (World Bank 2015).
   We use a randomized controlled trial to evaluate the program as designed and imple-
mented by the Government of Malawi, and to test the hypothesis that changes to the timing
of the program could increase its eﬀect on food security, potentially at the cost of investment
in fertilizer. The design takes advantage of oversubscription for the program: more villages
request PWP activities than can be accommodated, and even in villages that have projects,
not all eligible households can be included. The evaluation includes two levels of random-
ization, across villages and across households in treated villages. Villages that requested
PWP projects were assigned to either a pure control condition (no PWP at all) or one of
four treatment groups that oﬀered the same wages and total number of days of employment,
but diﬀered in the schedule of work and the frequency of payments. Within treated villages,
some households were randomly selected to be included in the program. Village treatment
assignment was stratiﬁed at the regional level to enable estimates of the impact of the pro-
gram across diﬀerent agro-ecological and socio-economic settings within the country. We use
household panel survey data to study the impact of the program (and design variants) on
food security and use of fertilizer, as well as other measures of household resource allocation
and wellbeing in order to understand the mechanisms through which the program aﬀects
these two key outcomes.
   Malawi’s PWP was not eﬀective in achieving its aim of improving food security during
the 2013 lean season. Treated households do not have better food security than households
in control villages, and we can rule out even moderate positive eﬀects on a summary measure
of food security. The program did not increase use of fertilizer or ownership of durable goods.
The lack of impact cannot be explained by increases in prices or crowding out of labor to
the casual wage market. The failure of the PWP to improve nutrition in either the short
run (through consumption support) or longer run (because of increased use of fertilizer) is
especially troubling because the MASAF PWP is the largest social protection scheme in one
of the world’s poorest countries.
   Not only did the program fail to improve food security for treated households, but also,
there is some evidence of negative indirect eﬀects on untreated households in villages with
PWP. In the North, these spillovers are negative and meaningful: a summary measure of food
security falls by one-third of a standard deviation for untreated households in villages with
PWP, relative to households in control villages. In Central Malawi, there is a compelling
pattern of negative but imprecisely estimated indirect eﬀects on untreated households. These
spillovers are not present in the poorest region, the South. However, the combination of no
positive direct eﬀect and some negative indirect eﬀects mean that at best, this extensive
program has no impact on food security.
   For comparison, the two largest public works programs are the NREGA program in India


                                              3
and the PNSP program in Ethiopia. The NREGA program in India appears to have had
somewhat more success in stabilizing consumption. NREGA diﬀers from Malawi’s PWP
in that it functions as a true safety net program that guarantees employment whenever
households need it, rather than oﬀering employment only in speciﬁc seasonal windows. Imbert
& Papp (2015) ﬁnd that the availability of the public works program provides an income ﬂoor
by increasing wages for both participants and non-participants, but unlike the Malawian
program, crowds out private sector work. (Ravi & Engler 2015) report that when jobs were
rationed, households oﬀered work have higher food expenditures and lower food security
than households unable to participate. Zimmermann (2014) ﬁnds that the program helps
households smooth against income shocks.
   Gilligan, Hoddinott & Taﬀesse (2009) ﬁnd modest eﬀects of Ethiopia’s PSNP on food
security and Hoddinott et al. (2012) ﬁnd no eﬀects on use of agricultural inputs, but large
increases in household assets. A more recent study of this program found signiﬁcant im-
provements in food security for households who participated for multiple years (Berhane
et al. 2014). When paired with an explicit program to improve agricultural productivity, the
impacts are larger.
   The negative indirect eﬀects of the Malawi program are also in contrast to evidence of
positive spillover eﬀects of social protection programs in other settings. Imbert & Papp (2015)
and Deiniger & Liu (2013) ﬁnd evidence of a general equilibrium eﬀect of the employment
guarantee scheme in India working through an increase in the casual wage rate, with positive
spillover eﬀects to the income of the poorest quintiles. Angelucci & DeGiorgi (2009) document
positive spillover eﬀects of the Oportunidades program in Mexico to households ineligible for
the program living in the same villages. Their indirect eﬀects operate through risk sharing,
with ineligible households being able to consume more through an increase in transfers and
loans from family and friends in the community. The low levels of risk sharing we detect
in our results are inconsistent with the hypothesis of a crowding-out eﬀect of risk sharing
networks as a response to the program. We do not ﬁnd evidence that the program aﬀected
prices by injecting cash into the economy, or any evidence of labor market tightening induced
by reduced labor supply or increased reservation wages.
   The remainder of the paper documents the details of the program, the experimental
design, and the unexpected impacts. Section 2 describes the program and the design of the
evaluation. Section 3 describes the data and outcomes of interest. In Section 4 we outline
the empirical strategy and the identiﬁcation of the parameters of interest. We explain our
analytic strategy in Section 4, and discuss the results in Section 5. Section 6 discusses
potential mechanisms for direct and indirect eﬀects of the program. Section 7 concludes.




                                              4
2     MASAF program and experimental design
The MASAF PWP has been operational since the mid-1990s and aims to provide short-term
labor-intensive activities to poor, able-bodied households for the purpose of enhancing their
food security, mainly through increased access to farm inputs at the time of the planting
period. The program was designed to be interlinked with Malawi’s large-scale fertilizer
input subsidy program (known as FISP) through the implementation of the PWP in the
planting months of the main agricultural season when the FISP distribution also occurs.
The premise behind this is that the PWP facilitates poor, credit-constrained households
to access subsidized fertilizer. This distinguishes Malawi’s program from traditional PWP
design with implementation during the lean season.
    The MASAF program covers all districts of Malawi with a two-stage targeting approach.
In the ﬁrst stage there is pro-poor geographic targeting, and in the second there is a combi-
nation of community-based targeting and self-selection of beneﬁciaries. The amount of funds
given to a district is proportional to the district’s population and to the poverty rates as well
as other measures of vulnerability. District oﬃcials then target a sub-set of extension plan-
ning areas (EPAs) based on poverty and vulnerability criteria. Traditional Authorities (TAs)
in the EPAs then allocate funds to a subset of selected Group Village Headmen (GVH) who
each oversee between 3-10 villages. The GVH determines how many households will par-
ticipate in each village based on available funding; the GVH then works with the village
committees in each village to select participating households.
    In 2012, as a response to a large currency devaluation, the program was doubled in size
and scaled up to cover about 500,000 households per year. The duration of the project
participation increased from 12 days to 48 days, split in two cycles of 24 days each; the cycles
are further divided into two consecutive 12-day waves and payments are generally made
within one to two weeks of the end of each wave. Projects were mostly road rehabilitation
or construction, with some work on aﬀorestation and irrigation projects. The wage rate was
MK300/day (US$0.92/day) for a total payment of MK 3,600 for a 12 day wave (US$11.01).
    Cycle 1 of PWP is implemented during the planting season (October to December), to
align with the timing of distribution of the fertilizer input subsidy program (FISP). Cycle 2
of PWP takes place post-harvest, in June and July.


2.1    Experimental design
We use a randomized controlled trial to test variants of the PWP that are budget-neutral in
terms of direct costs. Villages were stratiﬁed by geographic region and randomly assigned
(by computer) to one of the four treatment groups or a control condition; households within



                                               5
treatment villages were randomly selected for the program.

2.1.1     Village randomization

The villages in our sampling frame were randomly assigned to one of ﬁve groups (see Figure
1). The ﬁrst of these groups is a control group (Group 0) of villages that do not participate
in the PWP program in the 2012-2013 Season. Groups 1 through 4 participate in the PWP
in the planting season (Cycle 1 of PWP). These four groups vary in terms of the timing of
the second cycle of the program and the schedule of payments in both cycles.
    Comparing the four treatment groups to the control group measures the impact of the
1st cycle of the program (during the planting season) on consumption patterns and farming
investments of households. Among PWP-participating villages, two variants are tested:

    • Timing of the program PWP is designed to take place for two cycles of 24 days, during
        planting and during post-harvest season. In our evaluation design, we maintain the ﬁrst
        cycle at the planting season, and vary the timing of Cycle 2 to take place pre-harvest,
        during the lean season (February-March), instead of during/after the harvest season
        (May onward). Comparing Groups 1-2 and Groups 3-4 measures the consumption
        smoothing or buﬀer role of PWP during the lean season.

    • Schedule of the payments We introduced a variation in the payment approach from
        a lump-sum payment made after 12 days of work to a split-payment variant. Under
        the split-payment alternative, participants are paid three days apart, in ﬁve equal
        installments of MK 720 each.2 The variation of the payment schedule was motivated
        by extensive qualitative work done in preparation of the design of this project showing
        that households treat the lump-sum payments of the PWP diﬀerently from income
        generated through short-term casual labor (day-labor activities referred to as “ganyu”
        in Malawi). Comparing Groups 1 and 3 to Groups 2 and 4 will allow us to compare
        whether lump-sum payments alter the patterns of consumption and investment during
        the planting or lean season.3
   2 The market price of fertilizer in Malawi at the time of this project was approximately MK 5000 for a

50 kg bag. The national fertilizer subsidy program provided roughly half of households in the country with
coupons that allow two bags of fertilizer to be purchased for MK 500 each. Because households face high
transaction costs when redeeming their fertilizer coupons, including transportation costs, long wait times,
and inﬂexibility in the days on which fertilizer can be purchased at the government shops, it is substantially
more eﬃcient to purchase both bags of subsidized fertilizer at once, for MK 1000 plus transportation costs
(which are likely to range between MK 200 and MK 500). While MK 3600 (payment for 12 days of work)
more than covers the cost of purchasing two bags of subsidized fertilizer, a single incremental payment of MK
720 does not.
   3 Note that payments in the study districts were facilitated by the research team for the purposes of the

evaluation, with physical delivery of the cash in conjunction with the district oﬃcials. Although the split-
payment variant slightly increased the cost of implementation, the project variant was introduced to establish



                                                      6
        Payments in the study districts were facilitated by the research team for the purposes
        of the evaluation. This was intended to ensure that payments were made without
        delay, on speciﬁc schedules. Administrative payment records conﬁrm that there are no
        diﬀerences in time lag between work and payment across the districts.

    The cross-randomization is designed to test interaction eﬀects between the two design
variants, since the payment schedule may have a diﬀerential eﬀect depending on the season.
While a lump-sum payment may facilitate investment in a lumpy input in December, split
payments may help smooth consumption during the lean season. A lump sum in February
may be used for staples as well as temptation goods; divided payments can act as a form of
commitment savings that will lead to smoother consumption of staples if people otherwise
have high temptation to spend or high discount rates even over very short periods of time.
    We study a total of 182 villages (EAs) across our ﬁve treatment groups (Figure 1). Twenty
eight districts are included in the PWP program. For the evaluation, we randomly selected 12
districts, stratifying by the country’s three geographic regions. The 12 districts are Blany-
tyre, Chikwawa, Dowa, Karonga, Lilongwe, Mangochi, Mchinji, Mzimba, Nsanje, Ntchisi,
Phalombe, and Zomba. Within selected districts, the list of PWP-eligible villages (pre-
screened) villages from the District Council and Traditional Authorities was compared to
nationally representative survey data from the Integrated Household Survey (IHS3) collected
in 2010 and 2011. We identify the overlap between the sampled enumeration areas (EAs) in
the IHS3 and the list of communities pre-selected for PWP projects in our 12 districts. For
the villages selected for treatment, we randomly chose one project in the in the event that
the village are mapped to two, in order to have unique village-project pairs. The geographi-
cal targeting of the program is reﬂected in the regional breakdown of the sample (see Table
1), with about one-half of the sample drawn from the Southern region, which has a higher
incidence of poverty and food insecurity (Machinjili & Kanyanda 2012). Random assignment
was stratiﬁed by region.

2.1.2     Household randomization

The second level of randomization is at the household level. This level of randomization
improves statistical power and provides a mechanism for testing for the program’s indirect
eﬀects on non-participating households. Under the highly decentralized MASAF program,
the GVH identiﬁes households that are oﬀered PWP employment. The intention of the
program is to target poor households with able-bodied adults. As discussed below, we use
to test the proof of concept of whether the timing of the payment would alter the pattern of consumption and
investments in this context. Under discussion for the scale-up was an e-payments which would have entailed
a small marginal cost of delivery.



                                                     7
the 2010/2011 national household survey as a baseline for this study. By chance, then,
it is likely that one or two of the 16 randomly-surveyed households in our villages will be
among those chosen by the GVH for the PWP. We term these households as “village chosen
beneﬁciaries”.
   For this study, we randomly choose ten households from the 16 survey households in
the village to be oﬀered the program. To ensure that the experimentally-induced program
oﬀer did not aﬀect the village selection process, the list of randomly selected households was
distributed two weeks after lists of village chosen beneﬁciaries were submitted to district
councils. We deﬁne these randomly-chosen households as “top up” households who are
“treated” with the PWP program. Some small share of the “top up” households will also be
village chosen beneﬁciaries. Additionally, given the coverage rate of the status quo program,
one or two of the untreated households in PWP communities were likely to be included in
the program through the village selection process.
    In summary, our study has three groups of households: treated households in PWP
villages (top ups who are randomly oﬀered the program), untreated households in PWP
villages, and households in non-PWP communities. By focusing on the random oﬀer, we
estimate the intent-to-treat eﬀects of the program. By comparing the untreated households
in PWP villages to households in non-PWP villages, we are able to measure the indirect
(spillover) eﬀects of the program.


3     Data
The data for this study come from a panel household survey. The basis for the panel was the
Integrated Household Survey 3 (IHS3) which was ﬁelded in 2010/2011 by Malawi’s National
Statistics Oﬃce. The IHS3 is a cross-section of 12,288 households in 768 enumeration areas
(16 households per community) and has extensive household and agricultural modules.
    The 16 IHS3 households in each EA included in the evaluation were interviewed in four ad-
ditional rounds: before the public works projects started during the planting season (Novem-
ber 2012), after the ﬁrst cycle, pre-harvest (February 2013), after the lean season cycle,
post-harvest (April-May 2013), and ﬁnally after the completion of the 2012/2013 season
(November 2013; see Figure 2). Our ﬁrst survey (before PWP began in all but three villages)
is, in eﬀect, a second baseline to complement the IHS3. However, it could be tainted by an-
ticipation eﬀects if households in PWP villages modiﬁed their behavior before the program
began in expectation of the employment opportunities or other changes it would induce.
Twenty three enumeration areas (approximately 13 percent of the EAs in which the experi-
ment was implemented) were incorrectly classiﬁed as included in the IHS3, and are therefore



                                              8
areas for which IHS3 data were not collected. We will refer to them as the ‘non-IHS3’ sample,
as they no have true baseline data available.4


3.1     Food security measures
We examine food security outcomes with eight indicators and as well as a composite measure.
Our measures include (log) per capita food expenditure and (log) per capita food consumption
in the last week (including home consumption). Total household calories is computed based
on the caloric value of food consumed. A food consumption score is computed following WFP
guidelines and aims to capture both dietary diversity and food frequency; it is the weighted
sum of the number of days the household ate foods from eight food groups in the last week.5
We include a measure of the number of food groups consumed in the last week for seven
main groups.6 We have an indicator for whether the household reported reducing meals in
the last seven days. A food insecurity score is constructed according to WFP guidelines and
takes on a value of 1, 2, 3 or 4 (a higher number indicates greater insecurity).7 A coping
strategy index is a weighted sum of the number of days in the past 7 days that households
had to reduce the quantity and quality of food consumed.8 Finally, as many of these food
security measures are overlapping, we construct a principal components analysis index for
these eight measures as our ninth composite food security measure.
   4 The complexities of partnering with the Malawian government to both implement the intervention and

collect nationally representative data unfortunately preclude the ideal design: true baseline data are not
available in all EAs. Households in these 23 enumeration areas were listed and a sample of 16 households was
randomly drawn during our November 2012 survey. In these communities, we have one round of baseline data,
rather than two. For the IHS3 Households that could not be re-interviewed, the team drew a replacement
household from the original listing. About 9 percent of households are replacements for the original IHS3
household.
   5 The score is calculated based on the sum of weighted number of days in the last week the household ate

food from eight food groups: (2 * number of days of cereals, grains, maize grain/ﬂour, millet, sorghum, ﬂour,
bread and pasta, roots, tubers, and plantains) + (3 * number of days of nuts and pulses) + (number of days
of vegetables) + (4 * number of days of meat, ﬁsh, other meat, and eggs) + (number of days of fruits) + (4 *
number of days of milk products) + (0.5 * number of days of fats and oils) + (0.5 * number of days of sugar,
sugar products, and honey). Spices and condiments are excluded. It has a maximum value of 126.
   6 The seven are described in the previous footnote, with exception of the last group (sugars).
   7 Food insecurity score is 1 if in the past seven days, the household reports not worrying about having

enough food and reports zero days that they: (a) Rely on less preferred and/or less expensive foods, (b) Limit
portion size at meal-times, (c) Reduce number of meals eaten in a day, (d) Restrict consumption by adults in
order for small children to eat, or (e) Borrow food, or rely on help from a friend or relative. Food insecurity
score is 2 if the household reports worrying about having enough food and reports zero days for actions a-e.
Food insecurity score is 3 if the household reports ever relying on less preferred and/or less expensive foods
and b-e are zero. Food security score is 4 if the household reports any days for b-e.
   8 Referring to the ﬁve actions described in the previous footnote, the coping strategy index is the sum of

(a) + (b) + (c) + [3 * (d)] + [2 *(e)]. It has a maximum value of 56.




                                                      9
4     Analysis
We analyze the intent-to-treat eﬀect of Malawi’s public works program using household-
level data from two rounds of post-intervention surveys. Recall that our design includes two
levels of randomization: village-level randomization varied PWP availability and payment
structure, while household-level randomization varies eligibility to participate conditional on
PWP availability. Our main results pool across the four variants of the intervention in order
to estimate the eﬀect of having any public works opportunities in one’s village, and the
additional eﬀect of being a treated household within a PWP village. Therefore, we capture
the direct eﬀect of PWP availability on treated households, and the indirect eﬀect of the
program on untreated households in PWP villages. The indirect eﬀect is important in the
context of rationing.9
    Using data from the lean season (survey round 2), we pool across treatments and estimate
the equation
                    yiv = α + β1 PWPv + β2 PWPv ∗ Topupi + Γd + Θt +               iv                 (1)

where the indicator PWPv is a village-level indicator for the availability of any PWP program
and Topupi is a household-level indicator that equals one if the household was randomly
selected to be oﬀered (“treated”) the program and zero otherwise. The coeﬃcient β1 captures
the indirect eﬀect of the program on untreated households in PWP villages. The coeﬃcient
on the interaction term β2 captures the marginal eﬀect of being randomly selected for PWP
in a village that had a PWP program – that is, the eﬀect of being oﬀered the opportunity to
participate in program. The sum of the two coeﬃcients β1 and β2 captures the total eﬀect
of PWP on treated households compared to households with no PWP in their villages. All
speciﬁcations include district and week-of-interview ﬁxed eﬀects; standard errors are clustered
at the village level. As of survey round two (February 2013), all villages with PWP programs
had completed Cycle 1.
   In half of the PWP villages, Cycle 2 took place in the lean season, after survey round
2 and before survey round 3. Therefore, analyses of the pre-harvest data (survey round 3)
estimate the eﬀect of the lean season PWP compared to the standard harvest season program
   9 One limitation of our design is that we capture the indirect eﬀect only at the implicit coverage rate

as implemented in our study, approximately 18.8%. EAs are deﬁned by NSO to include an average of 200
households. Our randomization ensures that ten of the 16 surveyed households are oﬀered PWP. At the
overall coverage rate of 15%, another 28 households, or 15% of the remaining 184, would be included in the
program. Therefore, in PWP villages, an average of 37.6 out of 200 households have the opportunity to
participate.




                                                   10
for outcomes just before the harvest takes place.10 The regression speciﬁcation becomes

 yiv = α + δ1 Leanv + δ2 Leanv ∗ Topupi + δ3 Harvestv + δ4 Harvestv ∗ Topupi + Γd + Θt +             iv   (2)

The sum δ1 + δ2 is the direct eﬀect of having been oﬀered the lean season program – including
one 24-day work cycle at planting time and a second 24-day cycle during the lean season – on
outcomes just before the harvest. The sum δ3 + δ4 captures the direct eﬀect of the standard
program timing, under which participants have experienced only one 24-day work cycle at
planting season. The diﬀerence between the two sums is the marginal eﬀect on household
outcomes just before harvest of an additional 24 days of potential PWP work during the lean
season.
    The parameters we estimate are intent-to-treat, with identiﬁcation derived from random-
ized village and household treatment status, rather than the endogenous participation status
of households. Intent-to-treat parameters are policy relevant, in that the government can
control the coverage rate and oﬀer of PWP but not households’ decisions to take it up. The
eﬀect of the program on participating households is also of interest, and future research will
use instrumental variables speciﬁcations to estimate treatment-on-the-treated parameters and
will explore heterogeneous eﬀects on village-chosen versus randomly-assigned beneﬁciaries.


4.1       Balance
To explore the balance between treatment and control villages in terms of pre-treatment
covariates and outcomes, we use the IHS3 data from 2010/2011. Although our ﬁrst round
of follow-up survey pre-dates the PWP implementation in all but three villages, the survey
was conducted after the intervention was announced in treatment villages. Anticipation of
the program may have aﬀected household behaviors.
    We ﬁrst examine the subset of our 182 villages which were not included in the IHS3
national household survey in 2010/2011 but were included in our sample due to administrative
errors. They were randomly assigned to our treatment/groups (six, six, three, three, and ﬁve
villages to Groups 0-4 respectively). The diﬀerences between the IHS3 and non-IHS3 villages
reﬂects a composition eﬀect and has a bearing on the external validity of the results, but
it is orthogonal to treatment assignment. Using our ﬁrst round of follow-up data, we ﬁnd
that households in the non-IHS3 sample are better oﬀ than the IHS3 sample, with better
educated household heads, smaller household sizes, and fewer children below the age of 14
  10 Anticipation eﬀects due to the (announced) timing of Cycle 2 could aﬀect survey round 2 outcomes, even

though all villages with PWP programs had the same work opportunities as of that survey. In estimates
available upon request, we show that the eﬀects of the lean and harvest season variants were very similar in
survey round 2, so we prefer the statistical eﬃciency of the pooled speciﬁcation using equation (1).




                                                    11
(see Table 2).
    Unfortunately – and surprisingly, given that randomization was conducted by computer
– there is imbalance in pre-program food security at both the village and household levels.
This is apparent in the estimates of equation (1) for nine food security measures from the
IHS3, reported in Table 3.11 Untreated households in PWP village had worse food security
than households in control villages according to six of eight measures (columns 1-6, but not
7 and 8) and the PCA index; four of the diﬀerences are statistically signiﬁcant at the 95
percent conﬁdence level. Treated households fared better than their untreated neighbors
by all measures, and signiﬁcantly better in terms of the number of food groups consumed
and the PCA index. However, the treated households still have worse food security than
households in control villages according to six of eight measures and the PCA index. Four
of the diﬀerences are statistically signiﬁcant at the 90 percent conﬁdence level; p-values for
the test that treated households have the same food security as households in control villages
and 95 percent conﬁdence intervals are displayed below the point estimates.
    The nature of the imbalance is markedly diﬀerent in the North, where treated and un-
treated households have better baseline food security than households in control villages,
than in the Center or South. We report balancing tests for each of the three regions in
Appendix Table A1.
    As expected, the IHS sample is well-balanced for a range of non food-security outcomes.
Appendix Table A2 reports these tests. Out of 24 coeﬃcients, only one is statistically diﬀerent
from zero. This makes the pre-program diﬀerences in food security all the more diﬃcult
to understand, but suggests that treated households are not systematically better oﬀ than
untreated households along multiple dimensions of wellbeing.
    Because of the imbalance, we present three sets of results for our analysis of food security:
estimates without baseline (IHS3) covariates from the full sample of 182 villages, estimates
without covariates from the subsample of 159 IHS3 villages, and estimates with covariates
from the IHS3 villages. This strategy makes clear the extent to which changes in point
estimates are due to limiting the sample versus controlling for pre-treatment outcomes.
  11 IHS3 surveys were conducted from March 2010 until March 2011. Balancing tests control for month and

year of survey. To control for pre-treatment levels of outcome variables in subsequent regressions given the
strong seasonality in these measures, we use, we use the residual of each measure regressed on month and
year of survey indicators.




                                                    12
5     Results
5.1     Nation-wide eﬀect on lean-season food security
We begin by presenting results from equation (1) which estimates the direct eﬀect of the
program on treated households and the indirect eﬀect on untreated households.12 The sum
of the coeﬃcients β1 + β2 (P W P and P W P ∗ T op − up) compares households in treatment
villages that were themselves included in PWP to households in control villages. This total
eﬀect on top-ups is reported in the tables, along with 95 percent conﬁdence intervals (upper
and lower bounds). Table 4 displays the eﬀect of the program on food security measured
during the lean season (in survey round 2). Higher values indicate better food security for
the ﬁrst ﬁve measures and the PCA index, and lower values indicate better food security for
outcomes in columns 6-8. Panel A contains results from the full sample. The total eﬀect
of the program on top-up households moves in the direction of worse food security for all
measures: negative coeﬃcients for the ﬁrst ﬁve outcomes, positive coeﬃcients for the next
three outcomes, and a negative coeﬃcient in column (9) for the PCA index. The eﬀect on the
PCA index is equivalent to one-tenth of a standard deviation of that outcome for households
in control villages, but the coeﬃcient – like those for all but one of the individual outcomes
– is not statistically diﬀerent from zero.
    However, these diﬀerences are at least partially driven by pre-existing diﬀerences in food
security. To account for this imbalance, we ﬁrst limit the sample to the 159 villages covered
by the IHS3 survey (Panel B) and and then include pre-treatment values of the outcomes
(Panel C). In subsequent analyses, we will focus on estimates that correspond to Panel C,
but we show all three speciﬁcations here for transparency.13
   The pattern and magnitude of the direct eﬀect in the restricted sample in Panel B is
very similar to the eﬀects reported in Panel A. Including pre-treatment outcomes (purged
of seasonal variation) from the IHS3 reduces the absolute value of the estimated eﬀects by
between one quarter and one third (Table 4, Panel C). Treated households have worse food
security than households in villages without any public works programs for six of eight mea-
sures, but none of the diﬀerences are statistically diﬀerent from zero. The point estimate
of the eﬀect on the summary index is is very close to zero, and we can rule out any pos-
  12 Our ITT estimates are of the eﬀect of being randomly chosen for inclusion in the PWP program, not of

participating in the program. A small number of households were not randomly chosen to be included in the
program but were eligible through village selection procedures. In our speciﬁcation, these households pooled
with the untreated households. One might expect village chosen beneﬁciary households to have either better
outcomes than untreated households, because of the eﬀect of the program, or worse outcomes, if the poorest
households were indeed chosen and the program was not enough to oﬀset their relative disadvantage.
  13 An alternative to the ANCOVA-style speciﬁcations in Panel C is to include the PCA index of food

security from the IHS in all of the food security regressions. Results are very similar when controlling for
baseline variation in this way; results are available upon request.



                                                    13
itive eﬀect of more than 0.1 standard deviations. Overall, a program designed to improve
food security did not: households oﬀered the opportunity to participate in public works in
November/December 2012 and January 2013 did not have better food security during the
lean season than households in villages without a public works program.
  The indirect eﬀect of the program on untreated households in villages with PWP is
measured by the village-level indicator for public works, P W P . This coeﬃcient captures
the diﬀerence between mean outcomes for households in control villages and households that
were not randomly chosen for the PWP program in villages that had the program. Village-
chosen beneﬁciaries who were selected by their communities rather than the randomization
procedure are included among these six untreated households in our estimates. We interpret
this as a spillover eﬀect of the PWP.
    In the full sample (Panel A), this spillover eﬀect reduces food security. The diﬀerences
are statistically signiﬁcant for the food consumption score, food insecurity measure, and
the PCA index. The magnitude of the eﬀect on the latter is equivalent to 0.15 standard
deviations. Results are similar in the restricted sample (Panel B). Including pre-treatment
values of the outcomes (Panel C) reduces the estimated spillovers by about one half. Indirect
eﬀects are negative for ﬁve of eight outcomes and statistically signiﬁcant for two. Not only
does PWP fail to improve the food security of households randomly oﬀered the program, but
also, there are some indications that it may reduce food security for their neighbors.14 To
further understand the pattern of indirect and direct eﬀects, we turn to separate estimation
for each of the country’s three regions.


5.2     Heterogeneous eﬀects by geographic region
We estimate equation (1) separately for the North, Central, and South regions of the country
to understand whether the program was more eﬀective in some places than others, and to
consider whether regional variation provides any insight about the mechanisms for the eﬀects
we ﬁnd. Because of the baseline imbalance, our preferred speciﬁcations include pre-treatment
outcomes; results for the full sample and the baseline subsample are in the Appendix. As
reﬂected in the sample means reported at the bottom of each panel, the Northern region is the
country’s wealthiest and most food secure, and the Southern region is both more populous
and less food secure than other parts of the country.
  14 Jensen & Miller (2011) is an example of another study that attempts to improve nutrition by relaxing

the budget constraint, but ﬁnds zero or negative eﬀects. In that paper, price subsidies for staple goods led
some households to substitute towards more expensive calories from meat or fats, so total caloric intake fell
and nutrient intake did not rise. We do not observe an increase in dietary diversity, but note that ﬁnding a
nutritional elasticity close to zero is not without precedent.




                                                     14
5.2.1    North

The program has the biggest impact on food security in the North. The direct eﬀect of the
program reduces log per capita food expenditures and consumption; the food consumption
score; household caloric intake; and the number of food groups consumed in the previous
week. The eﬀects on expenditures and calories consumed by treated households relative
to controls are signiﬁcantly diﬀerent from zero. The program also appears to reduce food
insecurity by three subjective measures: reductions in meals consumed, the food insecurity
indicator, and the coping strategy index. For all three outcomes, the direct eﬀect of PWP is
statistically signiﬁcant. The total eﬀect on the PCA index is negative but not statistically
diﬀerent from zero.
   The indirect eﬀect of the program in the North is quite pronounced. Untreated house-
holds in PWP villages have worse food security than households in control villages for six
of eight measures (the P W P coeﬃcient in Table 5, Panel A) , with reductions in per capita
food consumption and the food consumption score signiﬁcant at the 90 percent level and
increases in the need to reduce meals and the WFP food insecurity indicator signiﬁcant at
the 95 percent level. The PCA index falls by 0.580, or one third of a standard deviation
for households in control villages in the region. That decline is signiﬁcant at the 90 percent
level.

5.2.2    Central

In the Central region, the direct eﬀect of the program is not statistically diﬀerent from zero
for any of the food security outcomes. The pattern and absolute value of point estimates are
close to those for the full sample: food security falls by six of eight measures, and the eﬀect
on the PCA index is small and negative.
    The indirect eﬀects are also imprecisely estimated, though greater in absolute value than
the direct eﬀects. For seven of the eight outcomes, untreated households in PWP villages
have worse food security than households in control villages. The impact on the PCA index
is -0.189, or almost one-tenth of a standard deviation of the PCA index for households in
control villages, but not statistically signiﬁcant at conventional levels. The overall pattern
of results is suggestive and troubling in this region: there is no evidence that the program
improves lean season food security for treated households and a pattern of coeﬃcients that
while not statistically signiﬁcant, consistently points towards negative eﬀects on untreated
households.




                                              15
5.2.3   South

PWP has very little impact on food security for treated or untreated households in the
South. The point estimate of the eﬀect is small and positive for four outcomes and small and
negative for the other four; the eﬀect on the PCA index is small and positive. The pattern
of indirect eﬀects is similarly unclear, with ﬁve outcomes showing small improvements three
showing small reductions in food security. The indirect eﬀect on the PCA index is 0.083,
which is both small relative to the standard deviation in the control group and relative to
point estimates in other regions.


5.3     Eﬀects of program variants in the pre-harvest period
As designed by the government, the second PWP cycle takes place during the harvest period,
beginning in May. This timing is suboptimal if marginal utility of consumption is higher and
opportunity cost of working lower during the lean season. To determine whether changing the
seasonality of the program could improve its eﬀectiveness, our evaluation randomly assigned
half of treated villages to move the second cycle, making public works start sooner, in March
2013. Survey round 3 took place after the March-April Cycle 2 and just before harvest.
In this round, villages assigned to the lean season treatment had completed work cycles in
November-January and March, but those assigned to the standard harvest season treatment
had not yet begun their second work cycle. The survey captures food security and other
outcomes at the very end of the lean season, before the new harvest has begun. Estimating
equation (2) allows us to test the marginal eﬀect of rescheduling the second cycle of PWP
for the lean season instead of the harvest season.
   We report national results in Table 6. As before, Panel A includes the full sample and no
controls for pre-treatment outcomes; Panel B limits the sample to observations for which pre-
treatment outcomes are available; and Panel C incorporates those pre-treatment outcomes in
the regressions. In addition to the coeﬃcients from equation (2), we report the total direct
eﬀect of the lean and harvest season programs (δ1 + δ2 and δ3 + δ4 , respectively); the 95%
conﬁdence interval for those eﬀects; and p-values for the tests that the direct eﬀect of the
lean season treatment equals the direct eﬀect of the harvest season treatment and that the
corresponding indirect eﬀects are equal.
   The direct eﬀect of the lean season program on treated households is small and generally
not signiﬁcantly diﬀerent from zero. We focus on Panel C to account for baseline imbalance.
The absolute value of the eﬀect on treated households is very small for all outcomes, and
there is no consistent pattern to the direction of the eﬀect. Similarly, spillover eﬀects on
untreated households in lean season PWP villages are close to zero (the point estimates in
the row labeled “lean”).


                                             16
   The eﬀect on treated households in villages whose second PWP cycle is scheduled for the
harvest season (soon after survey round 3) is also small for most outcomes, but the pattern
of coeﬃcients is in the direction of worse food security than households in control villages:
food security falls for six of eight measures and the PCA index, which falls by about 0.05
standard deviations. The indirect eﬀect (coeﬃcients in the the “harvest” row) is negative for
ﬁve of the eight outcomes, and the magnitude of the (negative) indirect eﬀect on the PCA
index is half that of the direct eﬀect.
   We cannot reject that the eﬀect of PWP on households treated with the lean season pro-
gram equals that of households treated with the harvest season for any individual outcomes.
This is despite the additional 24 days of work oﬀered to the former group of households. The
gap in eﬀects on the PCA index is 0.08 standard deviations – with the lean season PWP
(which had oﬀered two cycles of PWP) slightly improving and the harvest season PWP (only
one cycle oﬀered as of survey round 3) slightly worsening food security compared to villages
with no PWP. The p-value for the test that the direct eﬀects of those two variants on the
PCA index are equal is 0.22. Diﬀerences in indirect eﬀects of the two program variants are
even less precisely estimated and show no consistent pattern. In short, there is little evidence
that changing the timing of Cycle 2 is eﬀective in improving food security, even after treated
households in the lean season program have been oﬀered 48 days of work compared to 24
days for households in the harvest season treatment.

5.3.1   North

As with the eﬀects on lean-season food security, estimating equation (2) on the full sample
obscures important regional heterogeneity. In the North, the direct eﬀect of the lean season
program is small, with point estimates that are positive for six of eight outcomes and the PCA
index (Table 7, Panel A). The indirect eﬀect of the lean season program, though, is negative
for all outcomes and signiﬁcantly so for dietary diversity and the PCA index, which falls by
0.45 SD relative to households in the control group. The harvest season program performs
even worse. The direct eﬀect is small and the pattern is inconsistent, but the point estimate
for the PCA index is negative. The indirect eﬀect is negative for all outcomes, including
signiﬁcant reductions in dietary diversity and signiﬁcant increases in all three measures of
food insecurity. The negative eﬀect on the PCA index is equivalent to 0.48 SDs of the control
group.

5.3.2   Central

In the Central region, the lean season program improves food security for both treated and
untreated households in PWP villages (Table 7, Panel B). The direct eﬀect is positive for


                                              17
seven of eight outcomes and the PCA index, which increases by 0.17 SDs though none of
the diﬀerences relative to households in the control villages are statistically signiﬁcant at
conventional levels. The pattern and magnitude of indirect eﬀects are very similar: food
security improves by all eight measures, including an imprecisely-estimated 0.17 SDs of the
PCA index.
  Impacts of the harvest season program in this region are puzzling. The direct eﬀects are, if
anything, negative. Food security falls along six of eight measures, though none signiﬁcantly.
The magnitude of the eﬀect on the PCA index is close to zero. The indirect eﬀects, however,
are positive. Food security improves by seven of eight outcomes, including a signiﬁcant
decline in the number of days in which meals were reduced. The PCA index is 0.22 SD
higher for untreated households in villages with PWP than for households in control villages,
though the diﬀerence is not statistically signiﬁcant. Since even the treated households did
not work in the period leading up to survey round 3, it is conceivable that the indirect eﬀects
are due to anticipatory behavior by untreated households who are not on the cusp of another
24 day work cycle, relative to treated households whose next work opportunity is pending.

5.3.3   South

Finally, the lack of eﬀects of PWP in the South we observed in survey round 2 continues
in survey round 3. For the lean season program, the pattern of direct eﬀects is weak, with
food security falling for ﬁve of eight outcomes and the PCA index. Indirect eﬀects are no
more conclusive, and the point estimate of the indirect eﬀect on the PCA index is very close
to zero. The harvest season program has negative direct eﬀects on food security for six of
eight outcomes and reduces the PCA index by 0.10 SDs; none of the changes are signiﬁcant.
Indirect eﬀects are negative for ﬁve of eight outcomes, but the reductions in per capita
consumption and per capita caloric intake are signiﬁcant. The decline in the PCA index is
-0.1 SDs (as the negative impacts on the ﬁrst ﬁve food security measures are somewhat oﬀset
by imprecise positive eﬀects on the three measures of food insecurity).
   Each panel of Table 7 includes tests that the direct eﬀect of the lean season program
(where treated households were oﬀered 48 days of work as of survey round 3) equals the
direct eﬀect of the harvest season program (which oﬀered 24 days of work before survey
round 3), and that the corresponding indirect eﬀects are equal. We reject equality of the two
program variants for very few outcomes, but the overall pattern of results for food security
in all three regions is much more favorable – or at least less damaging – for the lean season
program (which oﬀered 48 days of PWP) than the harvest season program (as yet oﬀered 24
ays of PWP).




                                              18
6       Mechanisms
Malawi’s PWP shifts household budget constraints out by providing paid work that is oﬀset
by reductions in other labor supply. Despite this, households oﬀered PWP do not have better
food security or use more agricultural inputs as a result of the program, and food security
worsens among untreated households in PWP villages compared to their counterparts in
villages with no program. We discuss, and when possible test, potential mechanisms for the
lack of positive direct eﬀects in any region, and for the negative spillovers in the Northern
and Central regions.


6.1     Study design
6.1.1    Low power

Lack of statistical power is one possible explanation for a null main eﬀect. It would be a
more plausible explanation had we found positive but imprecisely estimated point estimates,
however. Instead, we ﬁnd predominantly negative eﬀects that are not statistically diﬀerent
from zero. This is true at not only the national level but also within each region.
    We calculate conﬁdence intervals for the total eﬀect on treated households using the delta
method. In most cases, we can rule out meaningful positive eﬀects of the program: nationally,
the upper bound (at the 95 percent conﬁdence level) of the improvement in food security
for any of the eight individual indicators is always less than 0.2 standard deviations of the
outcome in the control group. In the full sample, the standard deviation of the PCA index for
food security is 2.018, and the conﬁdence interval for the eﬀect of the public works program
on treated households is [-0.476, 0.087]. In the IHS3 subsample, the standard deviation in
the control group is 2.051, and the conﬁdence interval for the eﬀect is [-0.337, 0.204].
    By region, we can rule out eﬀects greater than 0.16 SDs in the North, 0.21 SDs in the
Center, and 0.24 SDs in the South. Thus, even moderate direct eﬀects are outside the
conﬁdence intervals in each region. It does not appear that lack of statistical power explains
failure to identify positive eﬀects of the program.

6.1.2    Low take-up

The second possibility is that household-level intervention, which chose households for inclu-
sion in PWP at random, resulted in low take-up and therefore small intent-to-treat (ITT)
estimates. The ITT estimates are not biased, but they are – by construction – smaller in
absolute value than the treatment on the treated (TOT) eﬀects. Since assignment to the
treatment group increases MASAF participation, the TOT and ITT eﬀects have the same



                                              19
sign. Therefore, discussion of the TOT does not oﬀer an explanation for coeﬃcients with
unanticipated direction of impact.
   As designed and implemented by the government, the program is targeted to vulnera-
ble households, who might participate at higher rates than randomly selected households.
Across the full study, 57 percent of treated (i.e. randomly-selected) households in our study
participated in MASAF (sum of coeﬃcients in column 1 of Table 8). Since there are consider-
able within-village spillovers, using household treatment status as an instrument violates the
Stable Unit Treatment Value Assumption (SUTVA) and is not a valid speciﬁcation. Instead,
we can instrument for PWP participation using village randomization, where the ﬁrst stage
equation is
                 Any PWP participationiv = δ0 + δ1 PWPv + Γd + Θt + νiv                     (3)

In this speciﬁcation, the treatment eﬀect incorporates direct and within-village spillover
eﬀects; the assumption is that there are no across-village spillovers within our sample and is
justiﬁed by the distance between study villages. Participation is less than 100 % by design:
δ1 = 0.145, and the ﬁrst stage F-statistic 182.44. The national treatment-on-the-treated
(TOT) parameters reported in Table 9 are larger in magnitude than the weighted average of
the direct and indirect eﬀects reported in Table 4, but imprecisely estimated and always in
the direction of reducing food security.
   Participation varies somewhat by region at 58 percent of treated households in the North,
42 percent in the Center, and 65 percent in the South. Regional TOT estimates are reported
in Table 10, and indicate worse food security for PWP participants in all regions. Low
take-up does not explain the lack of impact of PWP on food security.


6.2     Design of PWP
6.2.1   Value of transfer

A key design feature that may contribute to the lack of direct eﬀect is low total value of PWP
income. The wage rate for the program is set by the government at MK 300 ($0.92) per day,
with total possible earnings of MK 14,400 ($44.16). The wage rate is low by international
standards, but in a country with gross national income per capita of $320 ($730, adjusted
for purchasing power parity), $44 is non trivial. Local political constraints made it infeasible
to vary the wage rate for this study, so our experimental design does not allow us to speak
to the eﬀect of PWP with higher wages. Despite extensive expenditure data in our surveys,
we are not able to detect increased spending on any category: food, agricultural inputs or
business investments, non-food consumption, or durables. This limitation is shared with
many studies on micro-ﬁnance, which similarly fail to detect the eﬀect of increased cash on


                                              20
household consumption or expenditures.
   The lack of beneﬁcial eﬀects of the lean season program variant on survey round 3 out-
comes also undermines the idea that a more generous program would transform the eﬀects.
Treated households in villages selected for the lean season treatment were eligible for 24
additional days of work in March (for a total of 48 days from November to March). Yet in
surveys conducted the following month, their food security is no better than either house-
holds in control villages or treated households in the harvest season villages (oﬀered 24 days
from November to January).

6.2.2   Timing

A second hypothesis related to the design of the PWP program concerns timing of the
program. First, the program covers periods where the opportunity cost of time is potentially
high. Perhaps work on PWP activities crowded out labor supplied to the household’s own
farm or to the wage labor market. We do not have data about hours of work in household
agriculture, but note that since survey round 2 is conducted before the harvest, any reduction
in food security due to reduced harvests from households’ own plots cannot explain the results
and would instead exacerbate the negative eﬀect on food security measured later in the year.
   We report the eﬀect of the public works program on labor supply in daily wage mar-
kets(“ganyu”) in Table 8 (columns 3 and 4). Both the direct and indirect eﬀects of the
program are small and not statistically signiﬁcant at the extensive margin of participation;
at the intensive margin, PWP if anything crowds in wage labor, though standard errors
are large. It appears that households have excess supply of labor, a ﬁnding consistent with
Goldberg (Forthcoming).
   The planting season begins earlier in the South than in the North, and the Government
activated PWP activities earlier in the South. PWP and the fertilizer subsidy program are
administered separately and are not perfectly synchronized. In three study districts, fertilizer
subsidy coupon distribution took place between the ﬁrst and second 12-day waves of PWP
activities and in the remaining nine districts, fertilizer coupon distribution overlapped with
PWP work and payment. The three districts without overlap were Blantyre (South); Dowa
(Center); and Karonga (North). The North accounts for a smaller fraction of total population
and therefore of our sample, so discordant timing in one study district represents a larger
share of beneﬁciaries in that region than in the Center or South.

6.2.3   Targeting

Regressive or ineﬀective targeting potentially explains both lack of direct eﬀects and nega-
tive indirect eﬀects. PWP is intended for the able-bodied poor, and uses a combination of


                                              21
community wealth ranking exercises and low wages to target the program. In practice, the
characteristics of participants may diﬀer from the eligibility criteria because of diﬀerences in
how local oﬃcials select beneﬁciaries and in the opportunity cost of participation. There are
two types of targeting that may modulate impacts. The ﬁrst is the selection of village-chosen
beneﬁciaries. As noted above, some untreated households in our study were village-chosen
beneﬁciaries. We examine the correlation between baseline per capita food consumption and
participation of these households as an indication of whether the village selection procedures
targeted poorer households. Our preferred measure of baseline food security comes from the
IHS3, because unlike round 1 of the evaluation data, the IHS3 data were collected before
the intervention was announced. We cannot assess targeting on short-term food security,
because survey round 1 data were collected after the program was announced and may be
aﬀected by anticipated PWP earnings. However, if PWP is used locally in response to short-
term shocks, the lag between the IHS3 and program implementation may explain lack of
correlation between time-varying characteristics such as food security and participation.
    Nationally, Figure 3 illustrates that there is very little correlation between food security
in 2010/2011 and participation through the village selection process. This may suggest that
the village selection process either responds to short-term food security or relies on criteria
that are orthogonal to long term consumption.15
    As shown in Figure 4, the relationship between food security and PWP participation
varies by region. In the North and Central regions, participation of untreated households
was uniform across the distribution of baseline food consumption.16 In the South, households
with lower baseline food consumption were if anything less likely to be chosen by GVH and
participate in PWP.
    The second type of targeting is self-targeting, captured by participation by randomly-
selected households from diﬀerent points of the distribution of baseline food consumption.
This mimics an unrationed PWP like NREGA. Among treated households, the correlation
between accepting PWP work and IHS3 food consumption is negative in the North (indicating
pro-poor targeting and self-selection of the poorest households), but not in the Central or
Southern regions. Though self-targeting seems more prevalent in the North, there is no
evidence of displacement of work in the market for casual wage labor (“ganyu”) in any
region.
    Overall, Malawi’s PWP is rationed and not targeted towards the food-insecure. Mistar-
geting could explain the lack of improvement in food security if the program employed people
  15 The  autocorrelation between log per capita food consumption in the IHS3 and survey round 1 in control
villages 0.30. Over shorter horizons, between any two adjacent survey rounds, the autocorrelation in control
villages is close to 0.5.
  16 Food consumption data from the IHS3, accounting for seasonality by detrending by week-of-interview.




                                                    22
who had lower marginal propensity to consume food, but the geographic heterogeneity in tar-
geting does not seem to explain the regional heterogeneity in results. PWP was, if anything,
slightly better targeted in the region where it led to the most pronounced negative spillovers.

6.2.4   Project type

PWP activities included road building and tree-planting, which conceptually could have
required diﬀerent eﬀort levels or induced diﬀerential selection by participants. While project
type is unlikely to explain the lack of direct eﬀect, diﬀerent project types across the three
regions could lead to heterogeneous eﬀects. In fact, the mix of projects was similar across the
three regions. We have limited anecdotal information about the day-to-day work activities of
beneﬁciaries, with no evidence of systematic diﬀerences by region. As discussed previously,
PWP activities do not displace wage work in any region. We conclude that any diﬀerences
in work activities are unlikely to explain regional diﬀerences in program impacts.


6.3     Equilibrium eﬀects
6.3.1   Prices

Spillovers could operate through goods markets. If increased demand by treated households
drove up the price of food or other goods, the program may have reduced the purchasing
power of both treated households and their untreated neighbors. A change in price level has
the potential to explain both the lack of positive eﬀects for treated households and negative
eﬀects on their untreated neighbors, though diﬀerences in market conditions across the three
regions would be necessary to explain why the negative spillovers were concentrated in the
North and Central regions but not the South.
   We test whether village-level prices were diﬀerent in treatment and control villages, us-
ing a price index constructed from households’ reported prices for the ﬁve most commonly
purchased goods. The speciﬁcation for village level diﬀerences is

                              yiv = α + β PWPv + Γd + Θt +      iv                          (4)

Note that this speciﬁcation estimates the eﬀect of PWP on prices at the coverage rate in the
experiment. As reported in Table 16, the diﬀerence in the price index between treatment
and control villages is neither economically nor statistically signiﬁcant. Nationally, prices
in treatment villages are nearly identical – within one percent – to those in control villages.
Neither is there any evidence of price eﬀects at the regional level. In the North, the magnitude
of the coeﬃcient on the P W P indicator is near zero; prices are about seven percent lower for



                                              23
treatment villages than controls in the Center and seven percent higher in the South. None
of the diﬀerences are statistically signiﬁcant.

6.3.2   Labor markets

Imbert & Papp (2015) ﬁnd that the large Indian PWP increases wages in the low-skill market.
A similar eﬀect on equilibrium wages in Malawi could explain negative externalities by driving
up labor costs for untreated households that hire workers, especially since planting season
PWP activities coincide with peak labor demand periods. To reduce survey fatigue, we did
not repeat the full IHS-3 agricultural modules about hired labor, and therefore do not have
direct evidence about wages. To test this hypothesis indirectly, we interact the treatment
variables with an indicator for whether the household reported hiring any agricultural labor
in the IHS-3; about 20 percent of households previously hired labor.
   We estimate

        yiv =α + ψ1 PWP, no hired labori v + ψ2 PWP, hired labori v
             + ψ3 PWP, no hired labori v ∗ Topupi + ψ4 PWP, hired labori v ∗ Topupi        (5)
             + ψ5 Any previous hired labori + Γd + Θt +    iv


ψ1 is the indirect eﬀect of PWP on households that did not hire any labor during the
2010/2011 growing season and ψ2 is the indirect eﬀect on households that hired in the past.
If labor market tightening was responsible for negative spillover eﬀects, we would expect ψ2
to be negative (and more negative than ψ1 ).
   Instead, as reported in Table 17, the indirect eﬀect of PWP is negative for households
that do not hire, a group who might beneﬁt and should not be harmed by higher equilibrium
wages. In column (9) of Panel C, the regression for the PCA index that includes the baseline
value, the point estimate of ψ1 is -0.21, not statistically diﬀerent from zero at conventional
levels but substantial relative to the mean in the control group (0.51). The indirect eﬀect
on households that previously hired labor and would face higher costs due to an increase in
                        ˆ2 = 0.05. Thus, there is no evidence to support the labor market
wages is close to zero: ψ
tightening channel for the negative spillovers.


6.4     Change in transfers from institutions or individuals
Other institutions – government or NGO programs – or individuals could have changed their
behavior towards untreated households or otherwise compensated control villages with other
programs. In this case, the untreated households in PWP villages, which beneﬁted from
neither PWP nor these alternative programs, may be worse in comparison. We have data


                                                  24
about receipt of ﬁve types of beneﬁts: food assistance or maize distribution, school-based
nutrition programs, scholarships or other help with school fees, cash transfers, and fertilizer
coupons. As shown in Table 14, there is no evidence that untreated households diﬀer in
receipt of these beneﬁts from either treated households or households in control villages.
(Indeed, the only economically or statistically meaningful diﬀerence is in fertilizer coupons,
which are more likely to be available to treated households in accordance with the designed
linkage between PWP and the national fertilizer subsidy scheme).
   A related hypothesis is that treated households were not the ﬁnal beneﬁciaries of the
program, but instead shared their income with other households or used it to replace or
repay loans. We examine four outcomes in Table 15: whether the household received any
transfers, whether it made any transfers to others, whether it received any new loans, and
whether it made any loan payments in the previous month. PWP does not have a signiﬁcant
direct eﬀect any of these indicators, and the magnitudes of the direct eﬀects are small as well
as imprecisely estimated.
   It is also possible that contacts reduced transfers to – or increased demands for transfers
from – households in villages with PWP. Even untreated households in PWP villages could
be aﬀected through this channel. For example, family members living in other locations may
know that PWP was active in a village, but have inaccurate information about who actually
beneﬁted from the program. They may falsely believe that untreated households received
income from PWP and therefore do not need support through social networks or even have
funds they could share with others. Results reported in Table 15 indicate that untreated
households are signiﬁcantly less likely than households in control villages to make transfers
to others, ruling out the possibility of misguided pressure to share income. Untreated house-
holds do not signiﬁcantly diﬀer from controls (or treated households) in their probabilities
of receiving transfers or accessing loans, and eﬀects are not diﬀerent across the three regions
(results available upon request).


6.5    Behavioral responses by treated and untreated households
Finally, households oﬀered PWP may make choices that oﬀset or limit its eﬀect on food
security. If households behave according to the predictions of the permanent income hypoth-
esis (PIH), then the temporary increase in income will be allocated to consumption in all
future periods, and the change in consumption in any one period would be diminishingly
small, as in Chen, Mu & Ravallion (2009). This could account for the lack of positive eﬀects
on food security of treated households, but cannot explain negative spillovers to untreated
households. Perhaps more importantly, the myriad credit market imperfections in Malawi
that lead to well-documented seasonal variation in consumption make it unlikely that the


                                              25
PIH holds in this context.
   Another possibility is that treated households responded to the program by voluntarily
reducing their food consumption in order to pool existing resources and wages from the
program in order to make indivisible purchases of durables or investments in fertilizer or
other agricultural inputs. We test this hypothesis by estimating the eﬀect of PWP on durable
ownership (Table 11) and use of agricultural inputs (Table 12). We consider three measures
of durable ownership: the probability of owning at least one of 32 goods covered in the
survey, the total number of durable goods owned, and a PCA index of durable ownership.
The total eﬀect of PWP on treated households is not signiﬁcantly diﬀerent from zero for any
of the three outcomes, and the imprecise estimates for the number of goods and PCA index
are negative. There is no evidence that households that are covered by PWP use program
earnings or other resources to purchase durable goods, and these eﬀects do not diﬀer by
region (results available upon request).
   Use of agricultural inputs is important as an outcome in its own right – since comple-
mentarities with the fertilizer subsidy scheme drive the design of the program and increased
fertilizer use is a major stated goal – as well as a potential explanation for failure to improve
food security. In Malawi, fertilizer is applied twice to both maize (the staple crop) and to-
bacco (the main cash crop). We examine the probability that a household uses any chemical
fertilizer during the 2012/2013 season; the log of expenditure on fertilizer for each of the ﬁrst
and second applications; and the log of quantity of fertilizer used in each of the ﬁrst and sec-
ond applications. For the expenditure and quantity variables, we use the inverse hyperbolic
sine transformation in order to be able to take logs of variables where some observations are
zeros. The national point estimates of both the direct and indirect eﬀects on any of these
outcomes are close to zero, and none of the coeﬃcients are statistically signiﬁcant.
   However, results reported in Table 13 indicate important heterogeneity by region: while
fertilizer use is unchanged in the Central and Southern regions, it increases for both treated
and untreated households in PWP villages compared to non-PWP villages in the North.
Recall that negative direct and indirect eﬀects on food security are also most pronounced in
the North. Households in that region may have devoted PWP income and other resources to
purchase of fertilizer rather than purchase of food. We highlight this result, because it is the
only evidence we have of behavior that is consistent with the unexpected negative spillovers
on food security and matches the geographic pattern of direct and indirect results.


6.6    Remaining possibilities
One potential explanation for the observed negative spillovers to untreated households is
strategic self reporting by untreated households who do not experience real reductions in


                                               26
food security but choose to under-report consumption in order to signal their potential el-
igibility and increase their chances of being oﬀered PWP or other beneﬁts in the future.
We argue that two features of our research design make it less likely in this study than in
many other randomized controlled trials that rely on similar data. First, data collection
and program implementation are conducted by diﬀerent institutions. The National Statis-
tics Oﬃce is responsible for survey work, and the Local Development Fund with assistance
from Innovations for Poverty Action manages PWP activities and payments. Respondents
do not report their food consumption or other outcomes to people who are employed by or
connected to the administration of the public works program. Second, households in our
study17 were included in the IHS3, a nationally-representative survey that is not connected
to the provision of any public or private beneﬁts. These households had been administered
the same survey instrument by the same institution (NSO) before the study began, and had
not received any beneﬁts that depended on their answers. These features distinguish our
evaluation from RCTs that rely on data collected by employees of the same organization
that administers the intervention, and from evaluations of subjects who have no reason not
to infer that surveys are collected for the purpose of allocating beneﬁts.
    We are not able to assess the reliability of our various outcome measures in order to
determine whether perceived or self-reported food security may have fallen absent a change
in objective consumption. While the eight diﬀerent measures of nutrition we examine are
all related to food security, they measure diﬀerent aspects of the concept. Therefore, we
cannot use the correlations between items in the IHS3 and subsequent survey rounds as an
indication of item reliability, because the program itself may have aﬀected some measures
of food security diﬀerently than others. If strategic misreporting is a concern, it will not
be identiﬁed by comparing so-called “subjective” measures to “objective” measures (such as
expenditure).
    While we cannot directly rule out strategic misreporting, we think it less likely to inﬂuence
results in our evaluation than in many other studies relying on self-reported data. The
remainder of our discussion focuses on explanations for real negative spillovers. We cannot
rule out that PWP changed the behavior of untreated households in some way that worsened
food security. Such a mechanism would have to take a very speciﬁc form to explain both
the result and be consistent with the behaviors we observe. One possibility is as follows:
PWP eligibility causes a change in the behavior of treated households that is not captured
by our survey (either because it falls outside the recall period, or because it is too small to
be measured precisely). Untreated households respond to the behavior or consequences of
  17 Because of the error in NSO’s original village listing, about 20 percent of villages we study were not part

of the IHS3. As discussed previously, we do not observe diﬀerences in the impact of PWP across IHS3 and
non-IHS3 villages.



                                                      27
the behavior by the treated households. For example, treated households might spend their
earnings by increasing consumption immediately after payment, which would not be captured
by a the seven-day recall period of the survey which takes place a few weeks after the payment.
Untreated households respond by similarly increasing very short run consumption – to “keep
up” up with their neighbors, or because their reference point was aﬀected – but then have to
reduce consumption in subsequent periods because their short-term splurge was not ﬁnanced
by PWP earnings. This explanation requires peculiar behavior by the untreated households,
however – either their marginal utility of “keeping up” with their neighbors is high enough to
oﬀset reduced consumption later on, or their preferences are time-inconsistent, or they have
erroneous expectations about future income.
   An alternative explanation requiring less complicated behavior by the untreated house-
holds is that immediately after they are paid, treated households purchase food in bulk
quantities for storage and later consumption. These purchases are not captured in the survey
recall period and therefore do not translated into increased food security from the program,
but they do reduce the availability of commodities at local markets for untreated households.
The equilibrium eﬀect is on the quantity of goods, not prices, perhaps because of frictions
in the wholesale market. Food security for untreated households falls because of supply-side
factors.
   We see some evidence of this behavior in the fertilizer purchases of untreated households
in the North. Treated and untreated households increase their expenditure on and use of
fertilizer by about the same amount – nearly double – relative to controls. For treated
households, the MK 6,000 increase in spending across the basal and top dressing fertilizer
applications could have come entirely from PWP earnings. Untreated households, though,
did not have extra income to use for the fertilizer. They did not have the opportunity to
earn wages from public works activities and did not work more in the casual wage labor
market. If the increase in use of fertilizer had not been accompanied by a reduction in food
security, we would suspect that treated households had shared the spoils of the program
with their untreated neighbors. We see no evidence of this in our direct measures of cash
and in-kind transfers. The pattern of food security and fertilizer use results makes it more
likely that instead, untreated households in the North modeled their behavior on that of
their treated neighbors. We cannot distinguish between an adjustment of the reference point
about appropriate fertilizer use or incorrect expectations that PWP or other opportunities
would soon be available to untreated households.
   Relatedly, Bazzi, Sumatro & Suryahadi (2015) ﬁnd that a large-scale CCT in Indonesia
does not aﬀect consumption growth for households that receive transfers on schedule, but
reduces consumption growth for households whose second transfer is delayed. In our context,
untreated households in PWP villages may behave like Indonesian households whose transfers


                                              28
were delayed: their expectations about future income changed, but were unfulﬁlled (at least
within the survey recall period).


7     Conclusion
Public works programs can stabilize income and improve food security of beneﬁciaries by
providing earnings opportunities, and can achieve targeting through low wages and work
requirements that promote self-selection. While Malawi’s PWP oﬀers households the op-
portunity to earn approximately $22 at planting season and an additional $22 later in the
year, it does not have a measurable short-term eﬀect on lean season food security for treated
households. Even improving the structure of the program by rescheduling the second work
cycle from the harvest season to the lean season does not generate measurable improvements
in the food security of treated households.
    While the maximum possible income from PWP is substantial in light the country’s per-
capita GNI of $320, the magnitude of the transfer is low compared to other social protection
tools such as cash transfers being studied in Malawi,18 or other PWP programs in Sub-Saharan
Africa.19
    Households may have spread consumption across the four to eight month (depending on
treatment group) PWP period or saved for even longer durations. Then, changes in weekly
spending (the interval captured in our survey period) may be too small to detect, especially
since extra spending may have been spread across many diﬀerent categories of goods. We can
rule out signiﬁcant improvements in the two outcomes speciﬁcally targeted by the program
– food security and use of fertilizer – but there may have been small, diﬀuse increases in
these or other outcomes that we do not detect. This interpretation allows for the possibility
that the PWP was welfare-improving for households that chose to participate and thus is
consistent with their revealed preferences, while still ineﬀective in achieving its main policy
objectives.
    In Malawi, the public works program is designed with an additional goal: it is timed
to coincide with the planting season to promote take-up of the country’s fertilizer subsidy
scheme. However, our results do not support the hypothesis that the two programs are
complementary. While households included in PWP are more likely to receive fertilizer
  18 In their study of the cash transfer project in one district of Malawi in 2007, Miller, Tsoka & Reichert

(2011) ﬁnd large, positive eﬀects on beneﬁciary households in program villages compared to households in
control villages screened as eligible but not given the program. In this program the size of the beneﬁt is
signiﬁcantly larger, with transfers totaling $168 per household over the course of a year (equivalent to about
$250 in 2012 price levels), an amount more than ﬁve times what households received from PWP 2012/2013.
  19 Comparable PWP programs in the region, such as the ones in Ethiopia, Sierra Leone and Ghana, have

maximum earnings of between $60 and $190.




                                                     29
coupons (consistent of the policy of interlinkage with the fertilizer subsidy), and hence pay
less for the fertilizer they use, they do not use more fertilizer. In the Central and Southern
regions, treated households do not increase fertilizer use relative to households in control
villages, and in the North, treated and untreated households in PWP villages both increase
their fertilizer use relative to households in control village by the same amount. This may
suggest that other constraints, including lack of knowledge of the correct amount of fertilizer
to use (Duﬂo, Kremer & Robinson 2008), are more important than credit constraints in
limiting use of fertilizer.
   The program is rationed nationally, with funding available to cover only 15 percent of
households satisfying eligibility criteria. When coverage is extended to additional households
chosen randomly rather than through the community wealth ranking exercises, take-up is
about 50 percent. However, participation is higher among less food-secure households only
in the North, suggesting that the program is not achieving pro-poor targeting (as measured
by these indicators) through self-selection at the current wage rate. Despite this, the program
does not displace labor supplied to households’ own farms or to the market for casual day
labor, likely because of slack labor markets even during peak agricultural periods.
   The indirect eﬀects of the public works program are small or, surprisingly, negative. In
Northern and Central Malawi, food security for untreated households in villages with PWP
programs is not only lower than food security for their treated neighbors, but also lower than
food security in control villages without PWP activities. This is in contrast to expectations
and to the eﬀects of other large-scale transfer programs. For example, Oportunidades, the
conditional cash transfer program in Mexico, generated positive eﬀects on the consumption
of treated households and positive externalities to non-beneﬁciary households (Angelucci &
DeGiorgi 2009).
   An explanation for this unexpected ﬁnding has proven elusive. Cash transfers in Mexico
generated equilibrium eﬀects on commodity prices (Cunha, DeGiorgi & Jayachandran 2014),
but we ﬁnd no evidence of price increases in Malawian villages with PWP. The large public
works program in India raised equilibrium wages, but we reject the labor market tightening
explanation for negative spillovers in Malawi. Pressure to share money could have explained
the negative eﬀect on untreated households if relatives mistakenly believed that because PWP
was present in the village, even the untreated households had beneﬁted and could contribute
to the social network. However, there is no evidence of increased income sharing from PWP
villages and if anything, untreated households made fewer contributions to their networks
than households in control villages.
   A possibility we cannot exclude is that untreated households reduce food consumption
in reaction to an unobserved change in the behavior of treated households or to erroneous
expectations of their own future income. We discuss two examples of this: untreated house-


                                              30
holds over-spend to match a short term and thus unobserved increase in consumption by their
treated neighbors, and compensate by reducing consumption during the period we do observe,
or a combination of stock-outs and sticky prices reduces food availability. The matching and
substantial increases in fertilizer use by both treated and untreated households in the North,
the region with the strongest negative spillovers, are consistent with this hypothesis. We
lack direct evidence to test related behaviors, but present them as examples of the types of
mechanisms that could explain the unexpected ﬁnding that Malawi’s public works program
reduces the food security of untreated households in villages with PWP activities. Identifying
the mechanism remains a priority for both understanding household spending patterns and
for informing policy.




                                             31
Figures

                                       Figure 1: Experimental design

                                                                Treatment at the EA level
                                          Cycle 1: Planting season PWP           Cycle 1: Planting season PWP
                                          Cycle 2: Harvest season PWP             Cycle 2: Lean season PWP
 No PWP                                                                 Group 0
                                                                        [38 EAs]
 PWP with lump sum payment                    Group 1 (status quo)                          Group 3
                                                     [40 EAs]                               [35 EAs]
 PWP with split payment                              Group 2                                Group 4
                                                     [34 EAs]                               [35 EAs]



                                               Figure 2: Timeline
    Year-round         Pre-planting        Planting             Lean              Lean          Early harvest        Harvest          Pre-planting

                                       Wave 1: Nov.-Dec.                    Wave 1: Feb.-Mar.                    Wave 1: May-Jun.
Mar. 2010-Mar. 2011   Oct.-Nov. 2012                         Feb. 2013                          Apr.-May 2013                          Oct. 2013
                                       Wave 2: Jan. 2013                    Wave 2: Mar.-Apr.                    Wave 2: Jul.-Aug.

                      Survey round 1    PWP   Group   1                                                           PWP Group 1
                            (post       PWP   Group   2                                                           PWP Group 2
       IHS3                                                Survey round 2                       Survey round 3                       Survey round 4
                      announcement      PWP   Group   3                       PWP Group 3
                       of treatment)    PWP   Group   4                       PWP Group 4




                                                           32
                   Figure 3: National targeting based on IHS3 food security




MASAF participation is an indicator for whether any member of the household has participated in PWP
in the month preceding survey round 2 (dashed line is for treated households, and solid line for untreated
households. Log per capita food consumption from IHS3.


                   Figure 4: Regional targeting based on IHS3 food security




MASAF participation is an indicator for whether any member of the household has participated in PWP
in the month preceding survey round 2 (dashed line is for treated households, and solid line for untreated
households. Log per capita food consumption from IHS3.



                                                   33
Tables

         Table 1: Regional breakdown of treatment assignment



                Control EA             PWP EA             Total
     North          4                     18                22
                  10.5%                 12.5%             12.1%

     Central           14                  50               64
                     36.8%               34.7%            35.2%

     South             20                  76               96
                     52.6%               52.8%            52.7%

     Total            38                   144             182
                     100%                 100%            100%




                                 34
    Table 2: Descriptive statistics of households (survey round 1) by IHS3 status


                                      Non IHS3 EAs       IHS3 EAs      p-value:
                                                                       non IHS3 = IHS3

Female headed household               0.304              0.260         0.800
                                      (0.461)            (0.439)
Highest level of education, HoH       6.337              6.191         0.066
                                      (3.831)            (3.334)
Head attended secondary school        0.230              0.186         0.003
                                      (0.422)            (0.389)
Household size                        4.380              4.969         0.000
                                      (1.998)            (2.291)
Number of children under 14           2.149              2.388         0.005
                                      (1.568)            (1.718)
Number of EA                          [23]               [159]

Standard deviations in parentheses.




                                         35
                                             Table 3: Food security: balancing tests (IHS3)


     Dependent variable:                     (1)          (2)           (3)        (4)          (5)          (6)          (7)          (8)        (9)
                                             ln (p.c.     ln (p.c.      food       per adult    # food       any day      food         coping     PCA
                                             food exp.)   food cons.)   cons.      equivalent   groups       w/reduced    insecurity   strategy   index
                                             last week    last week     score      calories     consumed     meals        score        index      cols. 1-8

     PWP                                     -0.193**     -0.128**      -3.864**   -1762.095    -0.380***    0.016        -0.041       -0.712     -0.348**
                                             (0.089)      (0.051)       (1.279)    (1841.631)   (0.098)      (0.030)      (0.097)      (0.586)    (0.148)
     PWP * Top-up                            0.066        0.047         1.345*     405.184      0.170**      -0.012       -0.118*      -0.112     0.183**
                                             (0.053)      (0.034)       (0.780)    (1382.837)   (0.062)      (0.017)      (0.062)      (0.288)    (0.081)




36
     R-squared                               0.09         0.15          0.12       0.17         0.13         0.19         0.23         0.21       0.19
     Total eﬀect on top-ups                  -0.127       -0.082        -2.519     -1356.911    -0.210       0.005        -0.158       -0.823     -0.165
     P-value for non-zero eﬀect on top-ups   0.13         0.08          0.04       0.43         0.02         0.88         0.09         0.17       0.25
     Upper bound (total eﬀect)               0.037        0.009         -0.169     2022.784     -0.031       0.064        0.025        0.336      0.118
     Lower bound (total eﬀect)               -0.291       -0.173        -4.870     -4736.605    -0.389       -0.055       -0.342       -1.982     -0.448

     Observations                             2260        2274          2270        2274         2270        2274         2274          2274      2256
     Mean of dep. var.                        5.27        6.16          45.33       64728.56     5.08        0.22         2.34          4.57      0.05
     in control group
     OLS estimates. Standard errors clustered at the EA level. ∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.001
     Higher values indicate worse food security of columns 6, 7, and 8. Total calories are Winsorized at the 10th and 90th percentiles.
     Estimates include district and week-of-interview ﬁxed eﬀects.
                                                Table 4: Food security (survey round 2)


     Dependent variable:                     (1)          (2)           (3)       (4)          (5)        (6)           (7)          (8)        (9)
                                             ln (p.c.     ln (p.c.      food      per adult    # food     any day       food         coping     PCA
                                             food exp.)   food cons.)   cons.     equivalent   groups     w/reduced     insecurity   strategy   index
                                             last week    last week     score     calories     consumed   meals         score        index      cols. 1-8

     Panel A: full sample, no covariates
     PWP                                     -0.040       -0.071        -2.026*   -96.593      -0.116     0.065*        0.181**      0.566      -0.293**
                                             (0.080)      (0.044)       (1.072)   (61.277)     (0.106)    (0.034)       (0.080)      (0.635)    (0.148)
     PWP * Top-up                            -0.002       0.017         0.151     8.060        0.025      -0.048*       -0.129**     -0.156     0.099
                                             (0.050)      (0.031)       (0.665)   (50.639)     (0.068)    (0.025)       (0.061)      (0.489)    (0.096)
     R-squared                               0.15         0.17          0.14      0.13         0.18       0.08          0.08         0.09       0.16
     Total eﬀect on top-ups                  -0.041       -0.054        -1.874    -88.533      -0.091     0.017         0.052        0.409      -0.195
     P-value for non-zero eﬀect on top-ups   0.59         0.20          0.07      0.13         0.38       0.62          0.52         0.51       0.18
     Upper bound (total eﬀect)               0.108        0.028         0.163     24.807       0.113      0.083         0.210        1.631      0.087
     Lower bound (total eﬀect)               -0.191       -0.135        -3.911    -201.874     -0.295     -0.049        -0.106       -0.812     -0.476

     Observations                            2792         2837          2825      2835         2825       2808          2800         2802       2738
     Mean of dep. var.                       5.83         6.58          38.30     2341.49      4.32       0.49          3.12         9.09       0.13
     in control group

     Panel B: baseline subsample, no covariates
     PWP                                     0.004        -0.069        -2.149*   -64.478      -0.077     0.073**       0.216**      0.221      -0.258
                                             (0.092)      (0.050)       (1.183)   (68.238)     (0.115)    (0.035)       (0.084)      (0.671)    (0.160)




37
     PWP * Top-up                            0.034        0.026         0.676     -12.607      0.091      -0.051*       -0.162**     0.115      0.135
                                             (0.057)      (0.034)       (0.753)   (57.816)     (0.080)    (0.030)       (0.070)      (0.573)    (0.108)
     R-squared                               0.16         0.19          0.16      0.15         0.21       0.10          0.10         0.10       0.18
     Total eﬀect on top-ups                  0.038        -0.043        -1.473    -77.086      0.014      0.022         0.055        0.337      -0.124
     P-value for non-zero eﬀect on top-ups 0.66           0.34          0.20      0.21         0.90       0.53          0.50         0.61       0.42
     Upper bound (total eﬀect)               0.208        0.046         0.759     42.916       0.230      0.090         0.215        1.612      0.174
     Lower bound (total eﬀect)               -0.132       -0.132        -3.706    -197.088     -0.201     -0.046        -0.105       -0.939     -0.421

     Panel C: baseline sub-sample, including pre-treatment outcomes
     PWP                                     0.043       -0.042         -1.427    -50.716      0.005      0.070**       0.209**      0.242      -0.148
                                             (0.085)     (0.048)        (1.099)   (68.695)     (0.107)    (0.035)       (0.082)      (0.656)    (0.148)
     PWP * Top-up                            0.018       0.014          0.349     -19.091      0.046      -0.050*       -0.152**     0.127      0.082
                                             (0.056)     (0.034)        (0.732)   (57.822)     (0.080)    (0.030)       (0.070)      (0.574)    (0.106)
     R-squared                               0.21        0.24           0.21      0.17         0.25       0.10          0.10         0.11       0.26
     Total eﬀect on top-ups                  0.061       -0.028         -1.079    -69.807      0.051      0.020         0.056        0.370      -0.067
     P-value for non-zero eﬀect on top-ups 0.44          0.52           0.31      0.26         0.62       0.55          0.48         0.56       0.63
     Upper bound (total eﬀect)               0.215       0.058          0.992     51.181       0.250      0.088         0.214        1.607      0.204
     Lower bound (total eﬀect)               -0.092      -0.114         -3.149    -190.796     -0.149     -0.047        -0.101       -0.868     -0.337

     Observations                             2165        2216          2201      2215          2201        2201        2193         2195       2123
     Mean of dep. var.                        5.78        6.60          38.82     2329.06       4.32        0.48        3.12         9.32       0.15
     in control group
     OLS estimates. Standard errors clustered at the EA level. ∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.001
     Higher values indicate worse food security of columns 6, 7, and 8. Total calories are Winsorized at the 10th and 90th percentiles.
     Estimates include district and week-of-interview ﬁxed eﬀects.
                                               Table 5: Regional food security (survey round 2)

     Dependent variable:                     (1)          (2)           (3)       (4)          (5)          (6)          (7)          (8)        (9)
                                             ln (p.c.     ln (p.c.      food      per adult    # food       any day      food         coping     PCA
                                             food exp.)   food cons.)   cons.     equivalent   groups       w/reduced    insecurity   strategy   index
                                             last week    last week     score     calories     consumed     meals        score        index      cols. 1-8

     Panel A: North
     PWP                                     0.038        -0.208*       -3.987*   66.063       -0.172       0.203**      0.511**      1.406      -0.580*
                                             (0.122)      (0.121)       (2.148)   (224.061)    (0.202)      (0.061)      (0.179)      (0.995)    (0.302)
     PWP * Top-up                            -0.224       0.051         -0.615    -447.895**   0.024        -0.309***    -0.713**     -2.126**   0.305
                                             (0.136)      (0.091)       (1.793)   (140.374)    (0.200)      (0.079)      (0.187)      (0.905)    (0.267)
     R-squared                               0.36         0.20          0.16      0.15         0.19         0.18         0.09         0.11       0.26
     Total eﬀect on top-ups                  -0.185       -0.157        -4.602    -381.832     -0.148       -0.106       -0.202       -0.720     -0.275
     P-value for non-zero eﬀect on top-ups   0.07         0.50          0.15      0.00         0.73         0.00         0.01         0.06       0.95
     Upper bound (total eﬀect)               0.034        0.048         -0.155    -6.362       0.254        0.006        0.121        0.903      0.261
     Lower bound (total eﬀect)               -0.405       -0.363        -9.050    -757.302     -0.551       -0.217       -0.524       -2.343     -0.810

     Observations                            307          310           307       310          307          308          307          307        301
     Mean of dep. var.                       6.87         7.38          52.97     3230.30      5.95         0.43         2.66         5.43       2.24
     in control group

     Panel B: Central
     PWP                                     -0.068       -0.055        -1.267    -176.239     -0.074       0.029        0.199        -0.427     -0.189
                                             (0.154)      (0.084)       (1.884)   (107.781)    (0.205)      (0.052)      (0.145)      (0.917)    (0.283)
     PWP * Top-up                            0.188        0.037         0.437     142.813*     0.160        0.018        -0.082       0.745      0.123
                                             (0.117)      (0.055)       (1.414)   (81.181)     (0.154)      (0.047)      (0.133)      (0.843)    (0.199)




38
     R-squared                               0.09         0.09          0.07      0.04         0.08         0.09         0.06         0.05       0.10
     Total eﬀect on top-ups                  0.120        -0.018        -0.831    -33.426      0.087        0.047        0.118        0.318      -0.066
     P-value for non-zero eﬀect on top-ups   0.13         0.86          0.87      0.45         0.37         0.41         0.88         0.47       0.87
     Upper bound (total eﬀect)               0.386        0.135         2.448     157.891      0.452        0.141        0.389        2.066      0.435
     Lower bound (total eﬀect)               -0.146       -0.172        -4.109    -224.742     -0.279       -0.046       -0.154       -1.430     -0.567

     Observations                            787          822           819       821          819          809          808          808        771
     Mean of dep. var.                       5.38         6.47          34.33     1997.38      3.95         0.36         2.87         7.75       -0.17
     in control group

     Panel C: South
     PWP                                     0.179        0.046         0.019     -13.063      0.162        0.048        0.082        -0.331     0.083
                                             (0.123)      (0.067)       (1.545)   (99.449)     (0.135)      (0.056)      (0.109)      (1.100)    (0.189)
     PWP * Top-up                            -0.035       -0.016        0.570     -16.477      -0.036       -0.029       -0.044       0.333      -0.012
                                             (0.059)      (0.048)       (0.955)   (86.951)     (0.099)      (0.039)      (0.078)      (0.942)    (0.135)
     R-squared                               0.08         0.14          0.16      0.10         0.17         0.03         0.05         0.04       0.16
     Total eﬀect on top-ups                  0.144        0.030         0.589     -29.541      0.126        0.018        0.038        0.002      0.072
     P-value for non-zero eﬀect on top-ups   0.45         0.87          0.57      0.76         0.64         0.89         0.97         0.84       0.82
     Upper bound (total eﬀect)               0.381        0.148         3.611     150.841      0.391        0.129        0.255        2.046      0.438
     Lower bound (total eﬀect)               -0.092       -0.088        -2.433    -209.922     -0.138       -0.093       -0.178       -2.042     -0.294

     Observations                             1071        1084          1075       1084          1075       1084         1078         1080       1051
     Mean of dep. var.                        5.78        6.49          38.41      2342.21       4.16       0.59         3.44         11.56      -0.19
     in control group
     OLS estimates. All speciﬁcations include pre-treatment outcomes from the IHS3, residualized to allow for week-of-survey eﬀects
     in the IHS3 data. Standard errors clustered at the EA level. ∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.001
     Higher values indicate worse food security of columns 6, 7, and 8. Total calories are Winsorized at the 10th and 90th percentiles.
     Estimates include district and week-of-interview ﬁxed eﬀects.
                                                     Table 6: Food security (survey round 3)

     Dependent variable:                             (1)          (2)           (3)       (4)          (5)        (6)          (7)          (8)        (9)
                                                     ln (p.c.     ln (p.c.      food      per adult    # food     any day      food         coping     PCA
                                                     food exp.)   food cons.)   cons.     equivalent   groups     w/reduced    insecurity   strategy   index
                                                     last week    last week     score     calories     consumed   meals        score        index      cols. 1-8

     Panel A: full sample, no covariates
     Lean                                            -0.036       -0.073        -0.715    -22.351      -0.052     0.011        0.032        0.441      -0.129
                                                     (0.121)      (0.051)       (1.202)   (21.604)     (0.094)    (0.028)      (0.093)      (0.404)    (0.156)
     Lean season PWP * topup                         -0.039       0.058         -0.453    8.070        0.007      -0.061**     -0.161*      -0.353     0.126
                                                     (0.098)      (0.049)       (1.047)   (17.825)     (0.074)    (0.027)      (0.086)      (0.472)    (0.133)
     Harvest                                         -0.227**     -0.105**      -2.010*   -36.834      -0.127     -0.012       -0.032       -0.189     -0.218
                                                     (0.113)      (0.041)       (1.061)   (22.333)     (0.091)    (0.027)      (0.085)      (0.366)    (0.137)
     Harvest season PWP * topup                      -0.042       0.029         -0.513    -13.509      -0.116     0.001        0.030        0.641      -0.089
                                                     (0.095)      (0.041)       (0.881)   (18.024)     (0.083)    (0.032)      (0.097)      (0.444)    (0.130)
     R-squared                                       0.10         0.16          0.13      0.09         0.18       0.13         0.18         0.21       0.21
     Total eﬀect (lean PWP)                          -0.076       -0.015        -1.168    -14.281      -0.045     -0.050       -0.129       0.088      -0.002
     Upper bound (total eﬀect, lean PWP)             0.143        0.072         1.073     24.259       0.132      0.000        0.031        0.797      0.274
     Lower bound (total eﬀect, lean PWP)             -0.294       -0.102        -3.410    -52.822      -0.222     -0.101       -0.290       -0.621     -0.279
     Total eﬀect (harvest PWP)                       -0.269       -0.076        -2.523    -50.343      -0.244     -0.011       -0.002       0.452      -0.307
     Upper bound (total eﬀect, harvest PWP)          -0.060       0.008         -0.482    -10.519      -0.072     0.038        0.150        1.126      -0.051
     Lower bound (total eﬀect, harvest PWP)          -0.477       -0.161        -4.565    -90.167      -0.415     -0.060       -0.153       -0.222     -0.562
     p-value: direct (lean) = direct (harvest)       0.05         0.16          0.15      0.05         0.02       0.15         0.12         0.37       0.02
     p-value: indirect (lean) = indirect (harvest)   0.10         0.51          0.21      0.52         0.40       0.46         0.51         0.17       0.55

     Observations                                    2783         2804          2797      2804         2797       2779         2774         2776       2746
     Mean of dep. var.                               5.37         6.76          44.30     730.26       4.78       0.29         2.63         5.15       0.16
     in control group

     Panel B: baseline subsample, no covariates
     Lean                                            0.019        -0.014        -0.207    -5.417       -0.057     0.005        0.014        0.268      -0.057
                                                     (0.140)      (0.063)       (1.342)   (24.936)     (0.109)    (0.033)      (0.099)      (0.472)    (0.176)
     Lean season PWP * topup                         -0.061       0.019         -0.070    -5.162       0.012      -0.078**     -0.191**     -0.317     0.114
                                                     (0.111)      (0.056)       (1.190)   (19.086)     (0.086)    (0.033)      (0.094)      (0.540)    (0.148)
     Harvest                                         -0.214       -0.089*       -1.575    -33.882      -0.098     -0.007       -0.015       -0.430     -0.200
                                                     (0.135)      (0.052)       (1.274)   (26.173)     (0.111)    (0.035)      (0.101)      (0.466)    (0.168)




39
     Harvest season PWP * topup                      0.027        0.042         -0.087    -9.093       -0.102     -0.020       -0.053       0.626      -0.004
                                                     (0.115)      (0.050)       (1.100)   (20.479)     (0.098)    (0.041)      (0.113)      (0.527)    (0.158)
     R-squared                                       0.11         0.16          0.13      0.08         0.18       0.12         0.17         0.19       0.20
     Total eﬀect (lean PWP)                          -0.042       0.004         -0.276    -10.579      -0.044     -0.072       -0.177       -0.049     0.057
     Upper bound (total eﬀect, lean PWP)             0.199        0.102         2.099     33.403       0.161      -0.017       -0.008       0.768      0.368
     Lower bound (total eﬀect, lean PWP)             -0.283       -0.094        -2.652    -54.560      -0.249     -0.127       -0.346       -0.866     -0.253
     Total eﬀect (harvest PWP)                       -0.187       -0.047        -1.662    -42.976      -0.200     -0.028       -0.068       0.196      -0.204
     Upper bound (total eﬀect, harvest PWP)          0.052        0.051         0.583     3.120        -0.000     0.029        0.098        1.004      0.095
     Lower bound (total eﬀect, harvest PWP)          -0.427       -0.146        -3.907    -89.072      -0.400     -0.085       -0.233       -0.612     -0.502
     p-value: direct (lean) = direct (harvest)       0.19         0.28          0.17      0.09         0.09       0.13         0.22         0.58       0.06
     p-value: indirect (lean) = indirect (harvest)   0.10         0.23          0.29      0.27         0.69       0.75         0.80         0.20       0.41

     Panel C: baseline sub-sample, including pre-treatment outcomes
     Lean                                          0.020       0.002            0.041     -6.366       0.005      0.005        0.018        0.313      -0.018
                                                   (0.125)     (0.059)          (1.311)   (23.770)     (0.109)    (0.033)      (0.099)      (0.477)    (0.172)
     Lean season PWP * topup                       -0.085      0.008            -0.449    -8.811       -0.031     -0.077**     -0.178*      -0.291     0.046
                                                   (0.112)     (0.055)          (1.223)   (19.892)     (0.092)    (0.033)      (0.092)      (0.536)    (0.150)
     Harvest                                       -0.136      -0.051           -0.871    -28.704      -0.021     -0.008       -0.005       -0.363     -0.110
                                                   (0.117)     (0.046)          (1.208)   (24.837)     (0.107)    (0.035)      (0.101)      (0.466)    (0.158)
     Harvest season PWP * topup                    -0.006      0.022            -0.395    -7.981       -0.131     -0.020       -0.047       0.630      -0.056
                                                   (0.110)     (0.050)          (1.072)   (19.509)     (0.096)    (0.041)      (0.115)      (0.525)    (0.154)
     R-squared                                     0.19        0.22             0.17      0.15         0.21       0.13         0.18         0.20       0.26
     Total eﬀect (lean PWP)                        -0.065      0.010            -0.408    -15.178      -0.026     -0.071       -0.160       0.023      0.027
     Upper bound (total eﬀect, lean PWP)           0.139       0.095            1.826     27.517       0.172      -0.016       0.007        0.849      0.317
     Lower bound (total eﬀect, lean PWP)           -0.270      -0.075           -2.643    -57.873      -0.223     -0.126       -0.328       -0.803     -0.262
     Total eﬀect (harvest PWP)                     -0.143      -0.029           -1.266    -36.684      -0.152     -0.028       -0.053       0.267      -0.167
     Upper bound (total eﬀect, harvest PWP)        0.064       0.060            0.883     6.515        0.045      0.029        0.112        1.082      0.113
     Lower bound (total eﬀect, harvest PWP)        -0.349      -0.118           -3.415    -79.883      -0.348     -0.084       -0.217       -0.547     -0.446
     p-value: direct (lean) = direct (harvest)     0.44        0.39             0.39      0.28         0.17       0.14         0.23         0.58       0.15
     p-value: indirect (lean) = indirect (harvest) 0.24        0.38             0.48      0.38         0.80       0.73         0.84         0.21       0.59

     Observations                                   2167         2197         2190       2197         2190        2179        2175          2177       2138
     Mean of dep. var.                              5.37         6.77         44.48      732.59       4.82        0.32        2.71          5.69       0.15
     in control group
     OLS estimates. Standard errors clustered at the EA level. ∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.001
     Higher values indicate worse food security of columns 6, 7, and 8. Total calories are Winsorized at the 10th and 90th percentiles.
     Estimates include district and week-of-interview ﬁxed eﬀects.
                                           Table 7: Regional food security (survey round 3)

     Dependent variable:                             (1)          (2)           (3)       (4)          (5)        (6)          (7)          (8)        (9)
                                                     ln (p.c.     ln (p.c.      food      per adult    # food     any day      food         coping     PCA
                                                     food exp.)   food cons.)   cons.     equivalent   groups     w/reduced    insecurity   strategy   index
                                                     last week    last week     score     calories     consumed   meals        score        index      cols. 1-8

     Panel A: North
     Lean                                            -0.074       -0.244        -5.326    -138.945     -0.611**   0.082        0.360        1.542      -0.858*
                                                     (0.306)      (0.151)       (4.440)   (291.417)    (0.232)    (0.053)      (0.247)      (1.189)    (0.452)
     Lean season PWP * topup                         0.224        0.281         7.132**   281.108      0.502**    -0.122       -0.549**     -0.938     0.957**
                                                     (0.363)      (0.189)       (2.538)   (300.595)    (0.165)    (0.087)      (0.248)      (0.949)    (0.448)
     Harvest                                         -0.327       -0.142        -5.158    63.711       -0.564**   0.244**      0.641**      2.668*     -0.914**
                                                     (0.246)      (0.091)       (3.856)   (149.937)    (0.241)    (0.101)      (0.265)      (1.371)    (0.404)
     Harvest season PWP * topup                      0.418        0.159         2.534     -58.121      0.175      -0.330**     -0.777**     -2.283     0.762*
                                                     (0.272)      (0.099)       (1.570)   (147.801)    (0.123)    (0.125)      (0.333)      (1.397)    (0.383)
      R-squared                                      0.28         0.24          0.18      0.14         0.29       0.09         0.07         0.04       0.25
     Total eﬀect (lean PWP)                          0.149        0.036         1.806     142.162      -0.109     -0.041       -0.190       0.604      0.098
     Upper bound (total eﬀect, lean PWP)             0.590        0.271         9.348     409.881      0.394      0.083        0.160        2.512      0.888
     Lower bound (total eﬀect, lean PWP)             -0.292       -0.198        -5.735    -125.556     -0.613     -0.165       -0.539       -1.304     -0.691
     Total eﬀect (harvest PWP)                       0.091        0.016         -2.624    5.590        -0.390     -0.086       -0.136       0.384      -0.151
     Upper bound (total eﬀect, harvest PWP)          0.485        0.248         4.914     386.242      0.034      0.011        0.244        2.444      0.595
     Lower bound (total eﬀect, harvest PWP)          -0.303       -0.215        -10.161   -375.063     -0.813     -0.183       -0.516       -1.675     -0.898
     p-value: direct (lean) = direct (harvest)       0.83         0.89          0.08      0.53         0.18       0.52         0.83         0.81       0.58
     p-value: indirect (lean) = indirect (harvest)   0.44         0.44          0.96      0.47         0.78       0.16         0.38         0.45       0.90
     Observations                                    303          306           303       306          303        306          305          306        299
     Mean of dep. var.                               6.33         7.35          54.20     3109.72      6.26       0.21         2.29         3.55       1.80
     in control group

     Panel B: Center
     Lean                                            0.027        0.076         1.252     22.502       0.190      -0.085**     -0.205       -0.561     0.311
                                                     (0.269)      (0.106)       (2.464)   (149.156)    (0.194)    (0.041)      (0.173)      (0.514)    (0.315)
     Lean season PWP * topup                         0.058        -0.039        0.503     -46.508      -0.012     0.023        -0.093       0.110      -0.014
                                                     (0.173)      (0.075)       (2.045)   (110.372)    (0.156)    (0.038)      (0.154)      (0.654)    (0.229)
     Harvest                                         -0.030       0.098         1.959     -9.626       0.325      -0.081**     -0.167       -0.718     0.400
                                                     (0.234)      (0.087)       (2.292)   (133.070)    (0.201)    (0.036)      (0.146)      (0.538)    (0.277)
     Harvest season PWP * topup                      -0.081       -0.122*       -2.587    -113.317     -0.327*    0.069        0.164        1.216*     -0.463*
                                                     (0.189)      (0.068)       (1.849)   (121.448)    (0.185)    (0.050)      (0.176)      (0.668)    (0.251)




40
     R-squared                                       0.11         0.09          0.12      0.06         0.05       0.06         0.16         0.07       0.12
     Total eﬀect (lean PWP)                          0.085        0.037         1.755     -24.006      0.178      -0.062       -0.298       -0.451     0.297
     Upper bound (total eﬀect, lean PWP)             0.496        0.201         5.767     233.485      0.513      0.007        -0.019       0.545      0.803
     Lower bound (total eﬀect, lean PWP)             -0.325       -0.127        -2.257    -281.498     -0.157     -0.130       -0.577       -1.448     -0.208
     Total eﬀect (harvest PWP)                       -0.111       -0.024        -0.627    -122.943     -0.002     -0.012       -0.003       0.499      -0.063
     Upper bound (total eﬀect, harvest PWP)          0.326        0.137         3.527     103.946      0.359      0.060        0.268        1.461      0.463
     Lower bound (total eﬀect, harvest PWP)          -0.547       -0.185        -4.782    -349.833     -0.363     -0.083       -0.274       -0.463     -0.590
     p-value: direct (lean) = direct (harvest)       0.20         0.34          0.16      0.36         0.26       0.22         0.04         0.11       0.10
     p-value: indirect (lean) = indirect (harvest)   0.81         0.81          0.76      0.82         0.48       0.93         0.83         0.80       0.75
     Observations                                    803          817           816       817          816        815          814          814        799
     Mean of dep. var.                               5.15         6.72          43.69     2440.25      4.48       0.20         2.33         3.11       0.13
     in control group

     Panel C: South
     Lean                                            0.106        0.039         1.004     -21.687      0.031      0.062        0.104        0.643      0.008
                                                     (0.143)      (0.074)       (1.486)   (113.464)    (0.144)    (0.057)      (0.136)      (0.837)    (0.223)
     Lean season PWP * topup                         -0.263*      -0.027        -3.051*   -13.066      -0.193     -0.144**     -0.150       -0.375     -0.114
                                                     (0.151)      (0.075)       (1.548)   (112.308)    (0.120)    (0.053)      (0.125)      (0.944)    (0.203)
     Harvest                                         -0.109       -0.134**      -1.632    -157.059*    -0.161     -0.026       -0.059       -0.945     -0.191
                                                     (0.161)      (0.062)       (1.339)   (91.914)     (0.130)    (0.052)      (0.142)      (0.746)    (0.194)
     Harvest season PWP * topup                      -0.089       0.082         0.060     134.343      -0.091     0.013        0.023        1.108      -0.012
                                                     (0.149)      (0.075)       (1.550)   (105.327)    (0.130)    (0.055)      (0.146)      (0.792)    (0.196)
     R-squared                                       0.13         0.17          0.14      0.12         0.13       0.08         0.08         0.13       0.21
     Total eﬀect (lean PWP)                          -0.157       0.012         -2.047    -34.753      -0.161     -0.081       -0.046       0.268      -0.107
     Upper bound (total eﬀect, lean PWP)             0.112        0.124         0.407     151.498      0.097      0.011        0.187        1.668      0.267
     Lower bound (total eﬀect, lean PWP)             -0.426       -0.100        -4.502    -221.004     -0.420     -0.174       -0.279       -1.132     -0.481
     Total eﬀect (harvest PWP)                       -0.198       -0.052        -1.572    -22.716      -0.252     -0.013       -0.036       0.163      -0.203
     Upper bound (total eﬀect, harvest PWP)          0.063        0.063         0.516     158.527      0.006      0.085        0.198        1.568      0.140
     Lower bound (total eﬀect, harvest PWP)          -0.460       -0.168        -3.661    -203.959     -0.509     -0.111       -0.269       -1.242     -0.546
     p-value: direct (lean) = direct (harvest)       0.78         0.33          0.70      0.91         0.48       0.15         0.93         0.89       0.58
     p-value: indirect (lean) = indirect (harvest)   0.24         0.04          0.12      0.27         0.17       0.14         0.30         0.08       0.38
     Observations                                    1061         1074          1071      1074         1071       1058         1056         1057       1040
     Mean of dep. var.                               5.28         6.66          42.46     2650.42      4.69       0.44         3.12         8.24       -0.31
     in control group
     OLS estimates. All speciﬁcations include pre-treatment outcomes from the IHS3, residualized to allow for week-of-survey eﬀects
     in the IHS3 data. Standard errors clustered at the EA level. ∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.001
     Higher values indicate worse food security of columns 6, 7, and 8. Total calories are Winsorized at the 10th and 90th percentiles.
     Estimates include district and week-of-interview ﬁxed eﬀects.
                       Table 8: Labor supply (survey round 2)


Dependent variable:                      (1)          (2)           (3)          (4)
                                         Indicator:   Total         Indicator:   Total HH
                                         any HH       person-days   any HH       person-days
                                         MASAF        of MASAF      ganyu        of ganyu

PWP                                      0.118***     1.531***      0.032        0.329
                                         (0.026)      (0.348)       (0.032)      (0.632)
PWP * Top-up                             0.374***     4.663***      -0.016       0.160
                                         (0.025)      (0.339)       (0.019)      (0.421)
R-squared                                0.26         0.21          0.03         0.04
P-value for non-zero eﬀect on top-ups    0.00         0.00          0.62         0.43
Upper bound (total eﬀect)                0.547        6.979         0.076        1.713
Lower bound (total eﬀect)                0.437        5.410         -0.045       -0.735

Observations                            2820         2836             2818        2836
Mean of dep. var.                       0.04         0.51             0.41        4.43
in control group
OLS estimates. Standard errors clustered at the EA level. All estimates include
district and week-of-interview ﬁxed eﬀects. ∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.001




                                          41
                                             Table 9: Food security, TOT (round 2)


     Dependent variable:           (1)          (2)           (3)       (4)          (5)         (6)           (7)          (8)        (9)
                                   ln (p.c.     ln (p.c.      food      per adult    # food      any day       food         coping     PCA
                                   food exp.)   food cons.)   cons.     equivalent   groups      w/reduced     insecurity   strategy   index
                                   last week    last week     score     calories     consumed    meals         score        index      cols. 1-8

     HH participated in MASAF      -0.124       -0.171        -5.580*   -82.769*     -0.299      0.110         0.319        1.283      -0.682




42
     in past 30 days               (0.225)      (0.122)       (3.089)   (50.278)     (0.302)     (0.093)       (0.228)      (1.748)    (0.423)

     R-squared                      0.15         0.16           0.12      0.07          0.17      0.07         0.07        0.09        0.13
     Observations                   2776         2820           2805      2820          2805      2789         2781        2783        2724
     2SLS estimates. HH MASAF participation instrumented with PWP.
     Standard errors clustered at the EA level. ∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.001
     Higher values indicate worse food security of columns 6, 7, and 8. Total calories are Winsorized at the 10th and 90th percentiles.
     Estimates include district and week-of-interview ﬁxed eﬀects.
                                       Table 10: Regional food security, TOT (survey round 2)


     Dependent variable:        (1)            (2)           (3)         (4)          (5)        (6)          (7)          (8)        (9)
                                ln (p.c.       ln (p.c.      food        per adult    # food     any day      food         coping     PCA
                                food exp.)     food cons.)   cons.       equivalent   groups     w/reduced    insecurity   strategy   index
                                last week      last week     score       calories     consumed   meals        score        index      cols. 1-8

     Panel A: North
     HH participated in PWP     -0.509         -0.447*       -12.501**   -549.642     -0.361     -0.098       0.004        -1.639     -0.811
     in past 30 days            (0.375)        (0.236)       (5.060)     (531.664)    (0.444)    (0.185)      (0.409)      (2.550)    (0.728)

     R-squared                  0.24           0.00          0.00        0.09         0.09       0.06         0.03         0.05       0.12
     Observations               336            338           337         338          337        336          335          335        332

     Panel B: Center




43
     HH participated in PWP     -0.349         -0.257        -7.461      -351.598     -0.375     0.253        0.670        2.507      -1.189
     in past 30 days            (0.527)        (0.275)       (5.551)     (277.377)    (0.636)    (0.183)      (0.532)      (3.223)    (0.900)

     R-squared                  0.03           0.03          0.00        0.00         0.05       0.03         0.00         0.02       0.00
     Observations               945            983           981         983          981        960          959          959        921

     Panel C: South
     HH participated in PWP     0.149          -0.050        -1.268      -257.588     -0.162     0.062        0.118        0.380      -0.270
     in past 30 days            (0.262)        (0.148)       (3.773)     (218.456)    (0.375)    (0.116)      (0.237)      (2.324)    (0.498)

     R-squared                   0.01         0.05          0.08          0.05         0.07      0.03          0.04       0.05        0.04
     Observations                1495         1499          1487          1499         1487      1493          1487       1489        1471
     2SLS estimates. HH PWP participation instrumented with PWP and PWP * Top-up.
     Standard errors clustered at the EA level. ∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.001
     Higher values indicate worse food security of columns 6, 7, and 8. Total calories are Winsorized at the 10th and 90th percentiles.
     Estimates include district and week-of-interview ﬁxed eﬀects.
                            Table 11: Durable goods (survey round 2)


Dependent variable:                               (1)                                   (2)                    (3)
                                                  Indicator:                            Number of              PCA index
                                                  HH owns at least 1                    durable goods          for
                                                  of the 32 durable goods               owned                  durable goods

PWP                                               -0.034                                -0.256                 -0.084
                                                  (0.022)                               (0.209)                (0.191)
PWP * Top-up                                      0.009                                 -0.033                 -0.095
                                                  (0.019)                               (0.120)                (0.101)
R-squared                                         0.04                                  0.06                   0.03
P-value for non-zero eﬀect on top-ups             0.24                                  0.15                   0.34
Upper bound (total eﬀect)                         0.017                                 0.107                  0.190
Lower bound (total eﬀect)                         -0.067                                -0.685                 -0.548

Observations                             2831                 2831             2253
Mean of dep. var.                        0.82                 2.89             -0.03
in control group
OLS estimates. Standard errors clustered at the EA level.
The PCA index for durable goods is only deﬁned for HHs that own at least one good.
Estimates include district and week-of-interview ﬁxed eﬀects.
∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.001


                         Table 12: Agricultural inputs (survey round 2)

Dependent variable:                     (1)                 (2)               (3)                 (4)               (5)
                                        Indicator: HH       ln (kg            ln (kg              ln (exp.          ln (exp.
                                        used any            fertilizer, 1st   fertilizer, 2nd     fertilizer, 1st   fertilizer, 2nd
                                        fertilizer          application)      application)        application)      application)

PWP                                     0.010               0.020             0.051               -0.084            0.009
                                        (0.032)             (0.148)           (0.165)             (0.263)           (0.283)
PWP * Top-up                            0.003               0.040             0.082               0.038             0.135
                                        (0.018)             (0.083)           (0.106)             (0.152)           (0.178)
R-squared                               0.24                0.24              0.11                0.23              0.11
Total eﬀect on top-ups                  0.013               0.060             0.133               -0.046            0.145
P-value for non-zero eﬀect on top-ups   0.68                0.68              0.40                0.86              0.58
Upper bound (total eﬀect)               0.074               0.343             0.438               0.453             0.660
Lower bound (total eﬀect)               -0.048              -0.223            -0.173              -0.545            -0.371

Observations                             2764            2764                 2764                2764              2764
Mean of dep. var.                        0.68            3.07                 1.98                5.26              3.41
in control group
OLS estimates. Standard errors clustered at the EA level.
Estimates include district and week-of-interview ﬁxed eﬀects.
∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.001




                                                      44
              Table 13: Regional agricultural inputs (survey round 2)


Dependent variable:                      (1)              (2)               (3)
                                         Indicator: HH    ln (kg            ln (kg
                                         used any         fertilizer, 1st   fertilizer, 2nd
                                         fertilizer       application)      application)

Panel A: North
PWP                                      0.236**          0.964*            1.083**
                                         (0.111)          (0.490)           (0.470)
PWP * Top-up                             -0.064           -0.294            -0.274
                                         (0.058)          (0.244)           (0.329)
R-squared                                0.16             0.22              0.16
Total eﬀect on top-ups                   0.171            0.670             0.809
P-value for non-zero eﬀect on top-ups    0.13             0.18              0.04
Upper bound (total eﬀect)                0.385            1.616             1.521
Lower bound (total eﬀect)                -0.042           -0.276            0.097

Observations                             278              278               278
Mean of dep. var.                        0.42             1.90              1.44
in control group

Panel B: Center
PWP                                      -0.005           0.009             0.175
                                         (0.035)          (0.195)           (0.227)
PWP * Top-up                             0.042            0.202             0.282
                                         (0.033)          (0.159)           (0.222)
R-squared                                0.11             0.15              0.12
Total eﬀect on top-ups                   0.037            0.211             0.458
P-value for non-zero eﬀect on top-ups    0.25             0.27              0.03
Upper bound (total eﬀect)                0.100            0.583             0.869
Lower bound (total eﬀect)                -0.026           -0.161            0.046

Observations                             779              779               779
Mean of dep. var.                        0.78             3.67              2.00
in control group

Panel C: South
PWP                                      0.009            0.103             -0.238
                                         (0.060)          (0.268)           (0.289)
PWP * Top-up                             0.001            0.017             0.138
                                         (0.026)          (0.120)           (0.163)
R-squared                                0.37             0.36              0.17
Total eﬀect on top-ups                   0.010            0.120             -0.100
P-value for non-zero eﬀect on top-ups    0.87             0.64              0.70
Upper bound (total eﬀect)                0.123            0.622             0.416
Lower bound (total eﬀect)                -0.103           -0.381            -0.616

Observations                             1006              1007            1007
Mean of dep. var.                        0.61              2.68            1.84
in control group                           45
OLS estimates. All speciﬁcations include pre-treatment outcomes
from the IHS3, residualized to allow for week-of-survey eﬀects in the IHS3 data.
Standard errors clustered at the EA level. ∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.001
Estimates include district and week-of-interview ﬁxed eﬀects.
                         Table 14: Other programs (survey round 2)


Dependent variable:                     (1)             (2)           (3)           (4)          (5)
                                        Indicator:      Indicator:    Indicator:    Indicator:   Indicator
                                        any free food   any school    any           any cash     any fertilizer
                                        or maize        feeding       scholarship   transfer     coupon

PWP                                     0.002           0.004         -0.002        -0.000       0.047
                                        (0.025)         (0.030)       (0.004)       (0.005)      (0.032)
PWP * Top-up                            -0.013          -0.009        -0.002        0.004        0.018
                                        (0.013)         (0.017)       (0.003)       (0.005)      (0.021)
R-squared                               0.24            0.24          0.01          0.02         0.15
P-value for non-zero eﬀect on top-ups   0.67            0.84          0.19          0.51         0.04
Upper bound (total eﬀect)               0.037           0.053         0.002         0.015        0.124
Lower bound (total eﬀect)               -0.058          -0.065        -0.010        -0.007       0.005

Observations                             2840           2840          2840          2840         2840
Mean of dep. var.                        0.14           0.22          0.01          0.01         0.50
in control group
OLS estimates. Standard errors clustered at the EA level.
Estimates include district and week-of-interview ﬁxed eﬀects.
∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.001


                       Table 15: Transfers and loans (survey round 2)


Dependent variable:                           (1)                 (2)               (3)           (4)
                                              Indicator:          Indicator:        Indicator:    Indicator:
                                              any transfers       any transfers     any loans     any loan
                                              received            given             received      payments

PWP                                           0.005               -0.053**          -0.032        0.003
                                              (0.025)             (0.020)           (0.024)       (0.013)
PWP * Top-up                                  -0.028              0.007             -0.002        0.001
                                              (0.020)             (0.016)           (0.016)       (0.011)
R-squared                                     0.11                0.18              0.04          0.03
P-value for non-zero eﬀect on top-ups         0.33                0.02              0.13          0.74
Upper bound (total eﬀect)                     0.023               -0.009            0.010         0.027
Lower bound (total eﬀect)                     -0.070              -0.084            -0.079        -0.019

Observations                             2840           2840                        2840          2840
Mean of dep. var.                        0.24           0.22                        0.20          0.06
in control group
OLS estimates. Standard errors clustered at the EA level.
Estimates include district and week-of-interview ﬁxed eﬀects.
∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.001



                                                  46
               Table 16: Price index (survey round 2)



                     (1)              (2)               (3)        (4)
                     National         North             Center     South

PWP                  -3.335           -2.404            -36.293    30.750
                     (16.613)         (17.735)          (23.572)   (27.488)
R-squared            0.27             0.63              0.10       0.26

Observations          2793        339      959       1495
Mean of dep. var. 485.25          556.2    512.531 452.08
in control group
OLS estimates. Standard errors clustered at the EA level.
Estimates include district and week-of-interview ﬁxed eﬀects.
∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.001




                                47
                                  Table 17: Test of labor market tightening (food security, survey round 2)

     Dependent variable:                                          (1)          (2)           (3)       (4)          (5)        (6)         (7)          (8)        (9)
                                                                  ln (p.c.     ln (p.c.      food      per adult    # food     any day     food         coping     PCA
                                                                  food exp.)   food cons.)   cons.     equivalent   groups     w/reduced   insecurity   strategy   index
                                                                  last week    last week     score     calories     consumed   meals       score        index      cols. 1-8

     Panel B: baseline subsample, no covariates
     PWP, no hired labor                                          -0.036       -0.060        -2.299*   -62.565      -0.086     0.087**     0.160*       -0.028     -0.257*
                                                                  (0.099)      (0.051)       (1.168)   (78.012)     (0.112)    (0.036)     (0.083)      (0.691)    (0.154)
     PWP, hired labor                                             0.176        -0.032        -0.721    17.164       0.072      0.007       0.283        0.250      -0.068
                                                                  (0.147)      (0.086)       (1.983)   (116.669)    (0.193)    (0.065)     (0.175)      (1.079)    (0.282)
     PWP * Top-up, no hired labor                                 0.039        0.003         0.144     -27.947      0.070      -0.046      -0.094       0.339      0.065
                                                                  (0.066)      (0.039)       (0.796)   (67.599)     (0.087)    (0.031)     (0.069)      (0.653)    (0.114)
     PWP * Top-up, hired labor                                    -0.022       0.061         1.550     -7.838       0.091      -0.050      -0.316**     -0.270     0.219
                                                                  (0.106)      (0.065)       (1.629)   (107.456)    (0.147)    (0.051)     (0.139)      (0.901)    (0.210)
     Any hired labor in previous season                           0.184        0.191**       2.933*    267.373**    0.359**    -0.090*     -0.316*      -2.260**   0.650**
                                                                  (0.122)      (0.074)       (1.504)   (101.700)    (0.151)    (0.054)     (0.161)      (0.890)    (0.244)
     R-squared                                                    0.18         0.21          0.18      0.17         0.23       0.11        0.11         0.11       0.22
     Total eﬀect on top-ups, no labor                             0.00         -0.06         -2.15     -90.51       -0.02      0.04        0.07         0.31       -0.19
     Total eﬀect on top-ups, hired labor                          0.15         0.03          0.83      9.33         0.16       -0.04       -0.03        -0.02      0.15
     P-value for non-zero eﬀect on top-ups, no labor              0.975        0.220         0.052     0.206        0.881      0.233       0.416        0.635      0.186
     P-value for non-zero eﬀect on top-ups, hired labor           0.260        0.723         0.658     0.930        0.346      0.472       0.837        0.985      0.565
     P-value for equal indirect eﬀect, no labor and hired labor   0.179        0.767         0.433     0.555        0.424      0.251       0.529        0.808      0.528
     P-value for equal direct eﬀect, no labor and hired labor     0.298        0.323         0.107     0.422        0.315      0.175       0.584        0.764      0.214




48
     Panel C: baseline sub-sample, including pre-treatment outcomes
     PWP, no hired labor                                        -0.019         -0.040        -1.868*   -56.275      -0.018     0.087**     0.167**      0.030      -0.207
                                                                (0.096)        (0.048)       (1.089)   (76.501)     (0.104)    (0.035)     (0.084)      (0.672)    (0.139)
     PWP, hired labor                                           0.225*         -0.025        -0.292    8.325        0.165      -0.002      0.265        0.255      0.048
                                                                (0.129)        (0.079)       (1.742)   (115.953)    (0.181)    (0.064)     (0.174)      (1.063)    (0.259)
     PWP * Top-up, no hired labor                               0.035          -0.006        -0.029    -33.400      0.039      -0.045      -0.089       0.340      0.038
                                                                (0.066)        (0.039)       (0.779)   (68.285)     (0.086)    (0.031)     (0.070)      (0.655)    (0.115)
     PWP * Top-up, hired labor                                  -0.051         0.048         1.039     -14.707      0.025      -0.047      -0.296**     -0.214     0.123
                                                                (0.098)        (0.062)       (1.555)   (105.278)    (0.145)    (0.051)     (0.138)      (0.900)    (0.197)
     Any hired labor in previous season                         0.036          0.113*        1.376     231.969**    0.225      -0.078      -0.278*      -2.088**   0.311
                                                                (0.118)        (0.065)       (1.319)   (100.340)    (0.143)    (0.053)     (0.160)      (0.869)    (0.222)
     R-squared                                                  0.22           0.26          0.23      0.19         0.26       0.12        0.12         0.12       0.29
     Total eﬀect on top-ups, no labor                           0.02           -0.05         -1.90     -89.68       0.02       0.04        0.08         0.37       -0.17
     Total eﬀect on top-ups, hired labor                        0.17           0.02          0.75      -6.38        0.19       -0.05       -0.03        0.04       0.17
     P-value for non-zero eﬀect on top-ups, no labor            0.860          0.291         0.067     0.200        0.843      0.218       0.338        0.557      0.190
     P-value for non-zero eﬀect on top-ups, hired labor         0.146          0.758         0.648     0.952        0.231      0.404       0.845        0.969      0.469
     P-value for equal indirect eﬀect, no labor and hired labor 0.100          0.864         0.381     0.626        0.325      0.201       0.613        0.842      0.354
     P-value for equal direct eﬀect, no labor and hired labor   0.254          0.383         0.111     0.495        0.310      0.136       0.544        0.759      0.173

     Observations                                                 2165        2216           2201      2215        2201        2201       2193       2195          2123
     Mean of dep. var.                                            5.96        6.70           40.37     2456.92     4.57        0.42       3.00       8.19          0.51
     in control group
     OLS estimates. Standard errors clustered at the EA level. ∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.001
     Higher values indicate worse food security of columns 6, 7, and 8. Total calories are Winsorized at the 10th and 90th percentiles.
     Estimates include district and week-of-interview ﬁxed eﬀects. Panel A is omitted because the measure of previous hired labor comes from the IHS-3.
References
Alatas, Vivi, Abhijit Banerjee, Rema Hanna, Benjamin Olken, Ririn Purna-
    masari, and Matthew Wai-Poi. 2013. “Ordeal Mechanisms in Targeting: Theory
    and Evidence from a Field Experiment in Indonesia.” National Bureau of Economic
    Research Working Paper 19127.

Angelucci, Manuela, and Giacomo DeGiorgi. 2009. “Indirect Eﬀects of an Aid Pro-
    gram: How do Cash Transfers Aﬀect Ineligibles’ Consumption?” American Economic
    Review, 99(1): 486–508.

Bazzi, Samuel, Sudarno Sumatro, and Asep Suryahadi. 2015. “It’s All in the Timing:
    Cash Transfers and Consumption Smoothing in a Developing Country.” Working Paper.

Berhane, Guush, Daniel Gilligan, John Hoddinott, Neha Kumar, and Ale-
    mayehu Seyoum Taﬀesse. 2014. “Can Social Protection Work in Africa? The Impact
    of Ethiopia’s Productive Safety Net Programme.” Economic Development and Cultural
    Change, 63(1): 1–26.

Chen, Shaohua, Ren Mu, and Martin Ravallion. 2009. “Are there lasting impacts of
    aid to poor areas?” Journal of Public Economics, 93: 512–528.

Cunha, Jesse, Giacomo DeGiorgi, and Seema Jayachandran. 2014. “The Price Ef-
    fects of In-Kind Transfers: Evidence from a Mexican Food Assistance Program.” Work-
    ing Paper.

Deiniger, Klaus, and Yanyan Liu. 2013. “Welfare and Poverty Impacts of India’s National
    Rural Employment Guarantee Scheme.” World Bank Policy Research Paper no. 6543.

Duﬂo, Esther, Michael Kremer, and Jonathan Robinson. 2008. “How High Are
    Rates of Return to Fertilizer? Evidence from Field Experiments in Kenya.” American
    Economic Review, 98(2): 482–488.

Dutta, Puja, Rinku Murgai, Martin Ravallion, and Dominique vandeWalle. 2014.
    “Right to Work? Assessing India’s Employment Guarantee Scheme in Bihar.” World
    Bank Working Paper, https://openknowledge.worldbank.org/handle/10986/17195.

Gilligan, Daniel, John Hoddinott, and Alemayehu Seyoum Taﬀesse. 2009. “The
    Impact of Ethiopia’s Productive Safety Net Programme and its Linkages.” Journal of
    Development Studies, 45(10): 1684–1706.




                                          49
Goldberg, Jessica. Forthcoming. “Kwacha Gonna Do? Experimental Evidence about La-
    bor Supply in Rural Malawi.” American Economic Journal: Applied Economics.

Grosh, Margaret, Carlo del Ninno, Emil Tesliuc, and Azedine Ouerghi. 2008. For
    Protection and Promotion: The Design and Implementation of Eﬀective Safety Nets.
    Washington DC:The World Bank.

Hoddinott, John, Guush Berhane, Daniel Gilligan, Neha Kumar, and Ale-
    mayehu Seyoum Taﬀesse. 2012. “The Impact of Ethiopia’s Productive Safety Net
    Programme and Related Transfers on Agricultural Productivity.” Journal of African
    Economies.

Imbert, Clement, and John Papp. 2015. “Labor Market Eﬀects of Social Programs:
    Evidence from India’s Employment Guarantee.” American Economic Journal: Applied
    Economics, 7(2): 233–263.

Jensen, Robert T., and Nolan H. Miller. 2011. “Do Consumer Price Subsidies Really
    Improve Nutrition?” The Review of Economics and Statistics, 93(4): 1205–1223.

Machinjili, Charles, and Shelton Kanyanda. 2012. “Integrated Household Survey 2010-
    2011: Household Socio-Economic Characteristics Report.” Malawi National Statistical
    Oﬃce, Zomba, Malawi.

Miller, C. M., M. Tsoka, and K. Reichert. 2011. “The Impact of the Social Cash
    Transfer Scheme on Food Security in Malawi.” Food Policy, 36(2): 230–238.

Ravallion, Martin. 1999. “Appraising Workfare.” The World Bank Research Observer,
    14(1): 31–48.

Ravallion, Martin, Gaurav Datt, and Shubham Chaudhuri. 1993. “Does Maharash-
    tra’s Employment Guarantee Scheme Guarantee Employment? Eﬀects of the 1988 Wage
    Increase.” Economic Development and Cultural Change, 41(2): 251–275.

Ravi, Shamika, and Monika Engler. 2015. “Workfare as an Eﬀective Way to Fight
    Poverty: The Case of India’s NREGS.” World Development, 67: 57–71.

World Bank. 2015. “The State of the Social Safety Nets 2015.” Washington DC:World
    Bank.

Zimmermann, Laura. 2014. “Why Guarantee Employment? Evidence from a Large Indian
    Public-Works Program.” University of Georgia Working Paper.



                                          50
Online Appendix (not for publication)




                          51
                                             Table A1: Regional food security: balancing tests (IHS3)

     Dependent variable:                      (1)          (2)           (3)       (4)          (5)        (6)          (7)          (8)        (9)
                                              ln (p.c.     ln (p.c.      food      per adult    # food     any day      food         coping     PCA
                                              food exp.)   food cons.)   cons.     equivalent   groups     w/reduced    insecurity   strategy   index
                                              last week    last week     score     calories     consumed   meals        score        index      cols. 1-8

     Panel A: North
     PWP                                      -0.217       -0.062        -1.430    -380.847     -0.098     -0.226*      -0.712       -5.217*    0.294
                                              (0.236)      (0.109)       (2.142)   (244.381)    (0.183)    (0.130)      (0.420)      (2.736)    (0.379)
     PWP * Top-up                             0.097        0.051         2.258     164.287      0.370**    0.033        0.147        0.941      0.174
                                              (0.132)      (0.080)       (2.141)   (117.519)    (0.153)    (0.046)      (0.159)      (0.662)    (0.152)
     R-squared                                0.10         0.03          0.09      0.06         0.12       0.11         0.09         0.13       0.06
     Total eﬀect on top-ups                   -0.120       -0.011        0.828     -216.559     0.272      -0.193       -0.565       -4.276     0.468
     P-value for non-zero eﬀect on top-ups    0.59         0.91          0.72      0.38         0.10       0.14         0.19         0.13       0.24
     Upper bound (total eﬀect)                0.315        0.173         5.313     255.711      0.584      0.055        0.253        1.071      1.221
     Lower bound (total eﬀect)                -0.555       -0.195        -3.658    -688.829     -0.039     -0.441       -1.383       -9.623     -0.285

     Observations                             317          318           316       318          316        318          318          318        315
     Mean of dep. var.                        5.61         6.34          45.15     2874.98      5.18       0.32         2.35         6.58       0.22
     in control group

     Panel B: Central
     PWP                                      -0.171       -0.027        -1.630    122.477      -0.360**   0.061**      0.271*       0.122      -0.307
                                              (0.186)      (0.089)       (2.411)   (117.671)    (0.172)    (0.028)      (0.148)      (0.790)    (0.283)
     PWP * Top-up                             0.068        0.041         2.227*    -31.790      0.141      -0.016       -0.191*      -0.357     0.206
                                              (0.088)      (0.057)       (1.322)   (90.828)     (0.107)    (0.026)      (0.105)      (0.407)    (0.145)




52
     R-squared                                0.04         0.09          0.12      0.19         0.08       0.03         0.04         0.05       0.08
     Total eﬀect on top-ups                   -0.103       0.014         0.597     90.687       -0.218     0.045        0.080        -0.235     -0.101
     P-value for non-zero eﬀect on top-ups    0.55         0.86          0.81      0.41         0.18       0.12         0.54         0.76       0.70
     Upper bound (total eﬀect)                0.233        0.172         5.328     306.205      0.098      0.101        0.332        1.268      0.417
     Lower bound (total eﬀect)                -0.440       -0.143        -4.135    -124.831     -0.535     -0.011       -0.172       -1.739     -0.618

     Observations                             846          850           849       850          849        850          850          850        845
     Mean of dep. var.                        5.11         6.20          45.08     2512.70      4.98       0.10         1.89         3.26       0.22
     in control group

     Panel C: South
     PWP                                      -0.060       -0.148*       -2.043    -133.039*    -0.327**   0.035        -0.094       0.262      -0.317
                                              (0.133)      (0.075)       (1.880)   (74.876)     (0.133)    (0.035)      (0.108)      (0.744)    (0.206)
     PWP * Top-up                             0.053        0.055         0.576     62.504       0.152*     -0.028       -0.150*      -0.391     0.207*
                                              (0.075)      (0.050)       (1.040)   (77.293)     (0.083)    (0.027)      (0.084)      (0.461)    (0.124)
     R-squared                                0.05         0.14          0.09      0.09         0.13       0.21         0.32         0.23       0.21
     Total eﬀect on top-ups                   -0.008       -0.093        -1.467    -70.535      -0.175     0.007        -0.245       -0.129     -0.110
     P-value for non-zero eﬀect on top-ups    0.95         0.20          0.42      0.30         0.17       0.84         0.02         0.86       0.59
     Upper bound (total eﬀect)                0.241        0.048         2.069     62.959       0.070      0.074        -0.040       1.254      0.284
     Lower bound (total eﬀect)                -0.256       -0.233        -5.004    -204.029     -0.419     -0.061       -0.449       -1.512     -0.503

     Observations                             1097        1106          1105      1106          1105        1106        1106         1106       1096
     Mean of dep. var.                        5.29        6.07          45.57     2482.46       5.12        0.28        2.68         5.03       -0.11
     in control group
     OLS estimates. Standard errors clustered at the EA level. ∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.001
     Higher values indicate worse food security of columns 6, 7, and 8. Total calories are Winsorized at the 10th and 90th percentiles.
     Estimates include district and week-of-interview ﬁxed eﬀects.
                                                                       Table A2: Balance (IHS survey)

                                            (1)          (2)           (3)              (4)            (5)            (6)          (7)          (8)           (9)          (10)            (11)        (12)
                                            Indicator:   Total HH      Indicator: HH    ln (kg         ln (kg         Indicator:   Indicator:   Indicator:    Indicator:   Indicator: HH   Number of   PCA
                                            any HH       person-days   used any         fertilizer     fertilizer,    any          any school   any           any cash     owns at least   durable     index for
                                            ganyu        ganyu         fertilizer       1st            2nd            food         feeding      scholarship   transfer     1 of the 32     goods       durable
                                                                                        application)   application)   or maize                                             durable goods   owned       goods

     PWP                                    0.055        8.202         -0.043           -0.265         0.050          0.032**      -0.016       -0.000        0.004        -0.026          -0.120      -0.012
                                            (0.035)      (5.543)       (0.034)          (0.165)        (0.194)        (0.016)      (0.039)      (0.003)       (0.005)      (0.027)         (0.239)     (0.151)
     PWP * Top-up                           -0.039       -2.819        0.010            0.109          0.031          -0.016       0.006        -0.001        -0.006       0.024           -0.036      -0.012




53
                                            (0.024)      (4.068)       (0.016)          (0.084)        (0.112)        (0.011)      (0.012)      (0.002)       (0.004)      (0.019)         (0.114)     (0.063)
     R-squared                              0.07         0.14          0.38             0.37           0.13           0.08         0.19         0.01          0.03         0.03            0.06        0.06
     Total eﬀect on top-ups                 0.016        5.383         -0.033           -0.156         0.081          0.016        -0.010       -0.001        -0.002       -0.002          -0.155      -0.025
     P-value for non-zero eﬀect on top-ups  0.63         0.31          0.30             0.32           0.66           0.27         0.79         0.65          0.65         0.94            0.50        0.87
     Upper bound (total eﬀect)              0.081        15.731        0.030            0.152          0.441          0.045        0.066        0.004         0.006        0.048           0.293       0.265
     Lower bound (total eﬀect)              -0.049       -4.966        -0.096           -0.464         -0.279         -0.012       -0.086       -0.006        -0.009       -0.052          -0.604      -0.315
     Observations                           2274         2274          2125             2126           2126           2274         2274         2274          2274         2274            2274        1850
     Mean of dep. var.                      0.46         37.73         0.70             3.43           1.49           0.04         0.15         0.00          0.00         0.82            2.78        -0.68
     in control group
     OLS estimates. Standard errors clustered at the EA level. Estimates include district and week-of-interview ﬁxed eﬀects.
     The PCA index for durable goods is only deﬁned for HHs that own at least one good. ∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.001
                                      Table A3: Food security (Northern region, survey round 2)


     Dependent variable:                     (1)          (2)           (3)        (4)           (5)         (6)           (7)          (8)        (9)
                                             ln (p.c.     ln (p.c.      food       per adult     # food      any day       food         coping     PCA
                                             food exp.)   food cons.)   cons.      equivalent    groups      w/reduced     insecurity   strategy   index
                                             last week    last week     score      calories      consumed    meals         score        index      cols. 1-8

     Panel A: full sample, no covariates
     PWP                                     -0.085       -0.177        -3.189     -4.790        -0.161      0.086         0.271        0.191      -0.369
                                             (0.160)      (0.116)       (2.113)    (214.986)     (0.182)     (0.068)       (0.168)      (0.915)    (0.318)
     PWP * Top-up                            -0.149       0.047         -1.626     -357.906**    0.052       -0.233**      -0.534**     -1.426     0.197
                                             (0.140)      (0.083)       (2.091)    (102.649)     (0.205)     (0.080)       (0.188)      (0.913)    (0.287)
     R-squared                               0.26         0.08          0.06       0.14          0.10        0.10          0.06         0.07       0.15
     P-value for non-zero eﬀect on top-ups   0.18         0.24          0.03       0.08          0.60        0.04          0.12         0.16       0.60
     Upper bound (total eﬀect)               0.093        0.081         -0.665     30.276        0.291       -0.013        0.056        0.420      0.459
     Lower bound (total eﬀect)               -0.559       -0.340        -8.964     -755.669      -0.507      -0.281        -0.582       -2.890     -0.802

     Observations                            337          339           338        339           338         337           336          336        333
     Mean of dep. var.                       6.84         7.35          52.39      3210.67       5.92        0.45          2.70         5.63       2.16
     in control group

     Panel B: baseline subsample, no covariates
     PWP                                     -0.052       -0.226*       -5.210**   48.901        -0.241      0.163**       0.468**      0.869      -0.573
                                             (0.147)      (0.128)       (2.101)    (225.883)     (0.203)     (0.073)       (0.189)      (1.063)    (0.347)




54
     PWP * Top-up                            -0.194       0.068         0.227      -428.807**    0.131       -0.297**      -0.694**     -1.955**   0.356
                                             (0.146)      (0.098)       (1.759)    (142.158)     (0.193)     (0.079)       (0.192)      (0.903)    (0.286)
     R-squared                               0.26         0.08          0.07       0.14          0.12        0.14          0.08         0.08       0.16
     Total eﬀect on top-ups                  -0.246       -0.158        -4.983     -379.906      -0.110      -0.134        -0.226       -1.086     -0.217
     P-value for non-zero eﬀect on top-ups 0.13           0.17          0.04       0.07          0.62        0.08          0.22         0.26       0.53
     Upper bound (total eﬀect)               0.059        0.058         -0.457     3.088         0.315       0.007         0.123        0.750      0.453
     Lower bound (total eﬀect)               -0.550       -0.375        -9.509     -762.899      -0.535      -0.275        -0.575       -2.923     -0.887

     Panel C: baseline sub-sample, including pre-treatment outcomes
     PWP                                     0.038       -0.208*        -3.987*    66.063        -0.172      0.203**       0.511**      1.406      -0.580*
                                             (0.122)     (0.121)        (2.148)    (224.061)     (0.202)     (0.061)       (0.179)      (0.995)    (0.302)
     PWP * Top-up                            -0.224      0.051          -0.615     -447.895**    0.024       -0.309***     -0.713**     -2.126**   0.305
                                             (0.136)     (0.091)        (1.793)    (140.374)     (0.200)     (0.079)       (0.187)      (0.905)    (0.267)
     R-squared                               0.36        0.20           0.16       0.15          0.19        0.18          0.09         0.11       0.26
     Total eﬀect on top-ups                  -0.185      -0.157         -4.602     -381.832      -0.148      -0.106        -0.202       -0.720     -0.275
     P-value for non-zero eﬀect on top-ups 0.07          0.50           0.15       0.00          0.73        0.00          0.01         0.06       0.95
     Upper bound (total eﬀect)               0.034       0.048          -0.155     -6.362        0.254       0.006         0.121        0.903      0.261
     Lower bound (total eﬀect)               -0.405      -0.363         -9.050     -757.302      -0.551      -0.217        -0.524       -2.343     -0.810

     Observations                             307         310           307         310           307         308         307           307        301
     Mean of dep. var.                        6.87        7.38          52.97       3230.30       5.95        0.43        2.66          5.43       2.24
     in control group
     OLS estimates. Standard errors clustered at the EA level. ∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.001
     Higher values indicate worse food security of columns 6, 7, and 8. Total calories are Winsorized at the 10th and 90th percentiles.
     Estimates include district and week-of-interview ﬁxed eﬀects.
                                       Table A4: Food security (Central region, survey round 2)


     Dependent variable:                     (1)          (2)           (3)       (4)          (5)         (6)          (7)          (8)        (9)
                                             ln (p.c.     ln (p.c.      food      per adult    # food      any day      food         coping     PCA
                                             food exp.)   food cons.)   cons.     equivalent   groups      w/reduced    insecurity   strategy   index
                                             last week    last week     score     calories     consumed    meals        score        index      cols. 1-8

     Panel A: full sample, no covariates
     PWP                                     -0.161       -0.078        -2.287    -146.233*    -0.184      0.088        0.265*       0.830      -0.422*
                                             (0.145)      (0.072)       (1.497)   (85.239)     (0.169)     (0.053)      (0.144)      (0.891)    (0.243)
     PWP * Top-up                            0.133        0.011         0.679     82.720       0.155       -0.033       -0.152       -0.092     0.168
                                             (0.110)      (0.049)       (1.189)   (78.535)     (0.119)     (0.043)      (0.120)      (0.777)    (0.176)
     R-squared                               0.05         0.05          0.04      0.02         0.06        0.07         0.05         0.04       0.04
     P-value for non-zero eﬀect on top-ups   0.84         0.35          0.26      0.41         0.86        0.27         0.44         0.41       0.28
     Upper bound (total eﬀect)               0.239        0.072         1.169     86.055       0.293       0.152        0.398        2.500      0.206
     Lower bound (total eﬀect)               -0.296       -0.206        -4.386    -213.081     -0.351      -0.042       -0.173       -1.024     -0.713

     Observations                            952          990           988       988          988         967          966          966        926
     Mean of dep. var.                       5.44         6.48          34.41     2037.26      3.98        0.36         2.88         7.39       -0.11
     in control group

     Panel B: baseline subsample, no covariates
     PWP                                     -0.088       -0.063        -1.755    -179.367*    -0.131      0.041        0.225        -0.318     -0.291
                                             (0.160)      (0.083)       (1.878)   (105.339)    (0.211)     (0.052)      (0.146)      (0.943)    (0.278)




55
     PWP * Top-up                            0.202*       0.044         0.788     139.589*     0.188       0.015        -0.094       0.682      0.170
                                             (0.115)      (0.055)       (1.451)   (82.978)     (0.152)     (0.047)      (0.132)      (0.835)    (0.197)
     R-squared                               0.06         0.05          0.04      0.02         0.06        0.08         0.05         0.04       0.04
     Total eﬀect on top-ups                  0.115        -0.019        -0.967    -39.778      0.057       0.056        0.131        0.364      -0.120
     P-value for non-zero eﬀect on top-ups 0.43           0.81          0.56      0.67         0.77        0.24         0.34         0.70       0.64
     Upper bound (total eﬀect)               0.398        0.134         2.299     143.285      0.432       0.148        0.400        2.180      0.375
     Lower bound (total eﬀect)               -0.169       -0.172        -4.233    -222.841     -0.317      -0.036       -0.138       -1.452     -0.615

     Panel C: baseline sub-sample, including pre-treatment outcomes
     PWP                                     -0.068      -0.055         -1.267    -176.239     -0.074      0.029        0.199        -0.427     -0.189
                                             (0.154)     (0.084)        (1.884)   (107.781)    (0.205)     (0.052)      (0.145)      (0.917)    (0.283)
     PWP * Top-up                            0.188       0.037          0.437     142.813*     0.160       0.018        -0.082       0.745      0.123
                                             (0.117)     (0.055)        (1.414)   (81.181)     (0.154)     (0.047)      (0.133)      (0.843)    (0.199)
     R-squared                               0.09        0.09           0.07      0.04         0.08        0.09         0.06         0.05       0.10
     Total eﬀect on top-ups                  0.120       -0.018         -0.831    -33.426      0.087       0.047        0.118        0.318      -0.066
     P-value for non-zero eﬀect on top-ups 0.13          0.86           0.87      0.45         0.37        0.41         0.88         0.47       0.87
     Upper bound (total eﬀect)               0.386       0.135          2.448     157.891      0.452       0.141        0.389        2.066      0.435
     Lower bound (total eﬀect)               -0.146      -0.172         -4.109    -224.742     -0.279      -0.046       -0.154       -1.430     -0.567

     Observations                             787         822           819       821           819         809         808          808        771
     Mean of dep. var.                        5.38        6.47          34.33     1997.38       3.95        0.36        2.87         7.75       -0.17
     in control group
     OLS estimates. Standard errors clustered at the EA level. ∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.001
     Higher values indicate worse food security of columns 6, 7, and 8. Total calories are Winsorized at the 10th and 90th percentiles.
     Estimates include district and week-of-interview ﬁxed eﬀects.
                                      Table A5: Food security (Southern region, survey round 2)


     Dependent variable:                     (1)          (2)           (3)       (4)        (5)         (6)          (7)          (8)        (9)
                                             ln (p.c.     ln (p.c.      food      total      # food      any day      food         coping     PCA
                                             food exp.)   food cons.)   cons.     HH         groups      w/reduced    insecurity   strategy   index
                                             last week    last week     score     calories   consumed    meals        score        index      cols. 1-8

     Panel A: full sample, no covariates
     PWP                                     0.090        -0.030        -0.760    -129.604   -0.027      0.035        0.064        0.109      -0.134
                                             (0.108)      (0.066)       (1.591)   (95.622)   (0.159)     (0.050)      (0.100)      (1.036)    (0.215)
     PWP * Top-up                            -0.051       0.014         0.262     49.405     -0.060      -0.018       -0.023       0.055      0.040
                                             (0.051)      (0.045)       (0.855)   (74.091)   (0.089)     (0.032)      (0.068)      (0.739)    (0.125)
     R-squared                               0.01         0.06          0.09      0.06       0.09        0.03         0.04         0.05       0.04
     P-value for non-zero eﬀect on top-ups   0.71         0.80          0.75      0.39       0.58        0.73         0.68         0.87       0.65
     Upper bound (total eﬀect)               0.249        0.104         2.562     102.727    0.220       0.115        0.238        2.095      0.311
     Lower bound (total eﬀect)               -0.169       -0.136        -3.558    -263.126   -0.393      -0.081       -0.156       -1.767     -0.499

     Observations                            1503         1508          1499      1508       1499        1504         1498         1500       1479
     Mean of dep. var.                       5.89         6.49          38.22     2382.57    4.24        0.59         3.36         10.94      -0.10
     in control group

     Panel B: baseline subsample, no covariates
     PWP                                     0.161        0.001         -0.378    -28.741    0.095       0.049        0.076        -0.343     -0.010
                                             (0.131)      (0.074)       (1.705)   (98.328)   (0.151)     (0.056)      (0.114)      (1.117)    (0.221)




56
     PWP * Top-up                            -0.023       -0.002        0.704     -4.712     0.005       -0.031       -0.057       0.312      0.041
                                             (0.064)      (0.050)       (0.970)   (85.672)   (0.103)     (0.039)      (0.078)      (0.945)    (0.142)
     R-squared                               0.02         0.07          0.09      0.07       0.10        0.03         0.04         0.03       0.05
     Total eﬀect on top-ups                  0.138        -0.001        0.326     -33.453    0.100       0.018        0.018        -0.031     0.032
     P-value for non-zero eﬀect on top-ups 0.28           0.99          0.85      0.72       0.51        0.75         0.87         0.98       0.89
     Upper bound (total eﬀect)               0.387        0.130         3.673     147.250    0.396       0.130        0.240        2.066      0.460
     Lower bound (total eﬀect)               -0.110       -0.131        -3.021    -214.155   -0.196      -0.093       -0.204       -2.128     -0.397

     Panel C: baseline sub-sample, including pre-treatment outcomes
     PWP                                     0.179       0.046          0.019     -13.063    0.162       0.048        0.082        -0.331     0.083
                                             (0.123)     (0.067)        (1.545)   (99.449)   (0.135)     (0.056)      (0.109)      (1.100)    (0.189)
     PWP * Top-up                            -0.035      -0.016         0.570     -16.477    -0.036      -0.029       -0.044       0.333      -0.012
                                             (0.059)     (0.048)        (0.955)   (86.951)   (0.099)     (0.039)      (0.078)      (0.942)    (0.135)
     R-squared                               0.08        0.14           0.16      0.10       0.17        0.03         0.05         0.04       0.16
     Total eﬀect on top-ups                  0.144       0.030          0.589     -29.541    0.126       0.018        0.038        0.002      0.072
     P-value for non-zero eﬀect on top-ups 0.45          0.87           0.57      0.76       0.64        0.89         0.97         0.84       0.82
     Upper bound (total eﬀect)               0.381       0.148          3.611     150.841    0.391       0.129        0.255        2.046      0.438
     Lower bound (total eﬀect)               -0.092      -0.088         -2.433    -209.922   -0.138      -0.093       -0.178       -2.042     -0.294

     Observations                             1071        1084          1075      1084        1075        1084         1078        1080       1051
     Mean of dep. var.                        5.78        6.49          38.41     2342.21     4.16        0.59         3.44        11.56      -0.19
     in control group
     OLS estimates. Standard errors clustered at the EA level. ∗p < 0.10, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.001
     Higher values indicate worse food security of columns 6, 7, and 8. Total calories are Winsorized at the 10th and 90th percentiles.
     Estimates include district and week-of-interview ﬁxed eﬀects.