Published by:

Ministry of Foreign Affairs of Denmark

Evaluation Department
Asiatisk plads 2
1448 Copenhagen K
Denmark
E-mail: eval@um.dk
The publication can be downloaded from:
www.evaluation.dk
ISBN: 978-87-7087-586-8
ISBN: 978-87-7087-587-5 (internet version)
Other Evaluation Studies
2007/1: Assistance to Development Research
2007/2: Synthesis of Evaluations on Technical Assistance
2007/3: Evaluating the Impact of Rural Roads in Nicaragua
2008/1: Synthesis of Evaluations of HIV/AIDS Assistance
2008/2: Synergies between Bilateral and Multilateral Activities
2008/3: Assessing Multilateral Organisation Effectiveness
2008/4: Economic Empowerment of Women
2008/5: Public Private Partnership Programme
2009/1: Evaluating Aid Effectiveness in the Aggregate:
Methodological Issues
2009/2: Addressing the Underlying and Basic Causes of
Child Undernutrition in Developing Countries:
What Works and Why?
2009/3: Experiences with Conducting Evaluations Jointly with
Partner Countries
2009/4: Synthesis of Impact Evaluations of Microcredit
2009/5: Synthesis of Evaluations on Support to Business
Development
2010/1: Evaluating Aid Effectiveness in the Aggregate:
A critical assessment of the evidence
2010/2: Gender and Value Chain Development
2011/1: CCPAs Open Fun Football Schools Programme
EVALUATION STUDY
2011/2
Agricultural input subsidies
in Sub-Saharan Africa

!"#$%#&'() +&%,-
./0'1%$&%0#$ ')2%& 3%43','53 ') +%46+#7#0#) .80'1#
















Kenneth Baltzer, Assistant Proessor, PhD.
lenrik lansen, Proessor, PhD.
Institute o lood and Resource Lconomics, Uniersity o Copenhagen
Rolighedsej 25, 1958 lrederiksberg C.
_________________________________________
Useul comments and suggestions rom Senior Adisor Jens Kosted, Institute o lood and Resource
Lconomics, are grateully acknowledged.
1he iews expressed are those o the authors and do not necessarily represent the iews o the Ministry
o loreign Aairs o Denmark. Lrrors and omissions are the responsibility o the authors.

9#4$5 (8 :()&5)&3
1. Introduction ................................................................................................................................................... 1
2. Agricultural input subsidies.......................................................................................................................... 1
2.1. Agricultural inputs.................................................................................................................................. 1
2.2. Uniersal` s. smart` input subsidy programmes......................................................................... 2
3. Methodology .................................................................................................................................................. 3
3.1. Assessment criteria................................................................................................................................. 3
3.1.1. Liciency ........................................................................................................................................ 3
3.1.2. Lquity .............................................................................................................................................. 6
3.1.3. Sustainability................................................................................................................................... 6
3.2. 1he political economy o input subsidies...........................................................................................
3.3. Structure o the study ............................................................................................................................
3.4. Selection o cases and literature ........................................................................................................... 8
4. Selected input subsidy programmes in Sub-Saharan Arica .................................................................... 8
4.1. Malawi, Agricultural Input Support Programme,larm Input Support Programme ,AISP,.... 10
4.1.1. Background................................................................................................................................... 10
4.1.2. Outcome ....................................................................................................................................... 10
4.1.3. Scope ............................................................................................................................................. 12
4.1.4. 1argeting ....................................................................................................................................... 12
4.1.5. Deliery ......................................................................................................................................... 13
4.1.6. Lxit strategy.................................................................................................................................. 14
4.1.. Assessment ................................................................................................................................... 14
4.2. Zambia, lertilizer Support Programme ,ZlSP,.............................................................................. 15
4.2.1. Background................................................................................................................................... 15
4.2.2. Outcome ....................................................................................................................................... 16
4.2.3. Scope ............................................................................................................................................. 1
4.2.4. 1argeting ....................................................................................................................................... 1
4.2.5. Deliery ......................................................................................................................................... 18
4.2.6. Lxit strategy.................................................................................................................................. 19
4.2.. Assessment ................................................................................................................................... 19
4.3. Ghana, lertilizer Subsidy Programme ,GlSP,................................................................................ 20
4.3.1. Background................................................................................................................................... 20
4.3.2. Outcome ....................................................................................................................................... 21
4.3.3. Scope ............................................................................................................................................. 21
4.3.4. 1argeting ....................................................................................................................................... 22
4.3.5. Deliery ......................................................................................................................................... 22
4.3.6. Lxit strategy.................................................................................................................................. 23
4.3.. Assessment ................................................................................................................................... 23
4.4. 1anzania, National Agricultural Input Voucher Scheme ,NAIVS,.............................................. 24
4.4.1. Background................................................................................................................................... 24
4.4.2. Outcome ....................................................................................................................................... 24
4.4.3. Scope ............................................................................................................................................. 24
4.4.4. 1argeting ....................................................................................................................................... 25
4.4.5. Deliery ......................................................................................................................................... 26
4.4.6. Lxit strategy.................................................................................................................................. 26
4.4.. Assessment ................................................................................................................................... 26
5. Lessons Learned .......................................................................................................................................... 2
5.1. Background........................................................................................................................................... 2
5.2. Outcome................................................................................................................................................ 28
5.3. Scope...................................................................................................................................................... 28
5.4. 1argeting................................................................................................................................................ 29
5.5. Deliery ................................................................................................................................................. 29
5.6. Lxit Strategy.......................................................................................................................................... 30
6. Conclusion.................................................................................................................................................... 30
. Reerences..................................................................................................................................................... 32



;'3& (8 .4405"'#&'()3

AISP Agricultural Input Support Programme, Malawi
AU Arican Union
CAN Calcium Ammonium Nitrate
CNlA Citizens Network or loreign Aairs
GDP Gross Domestic Product
GlSP lertilizer Subsidy Programme, Ghana
GlS Ghanaian cedis
NAIVS National Agricultural Input Voucher Scheme, 1anzania
NDP National Democratic Congress
NLPAD New Partnership or Arican Deelopment
NGO Non-Goernmental Organisation
NMB National Microinance Bank
NPK Nitrogen Phosphorous Kalium ,Potassium,
NPP New Patriotic Party
MK Malawian Kwacha
OPV Open Pollinated Variety
SSA Sub-Saharan Arica
USD US Dollar
VDC Village Deelopment Committee
VVC Village Voucher Committee
ZlSP lertilizer Support Programme, Zambia
ZMK Zambian Kwacha
1


<= >)&0(,%1&'()
1he purpose o this study is to ealuate the experiences o recent applications o agricultural input
subsidies in Sub-Saharan Arica ,SSA,. Subsidising agricultural inputs is controersial. On the one hand,
agricultural input use in SSA is ery low by international standards, and the hope is that subsidies may
induce armers to adopt the use o inputs and thereby increase agricultural productiity. On the other
hand, many economists argue that agricultural subsidies o all kinds are expensie, mainly beneit the
wrong people, and distort agricultural markets by encouraging armers to oerusing whateer is
subsidised. 1he recent input subsidy programmes reiewed in this study attempt to meet these
challenges by introducing so-called smart` subsidies, which are speciically designed to maximise
eects at the lowest cost ,we will discuss smart subsidies in more detail below,.
1he oerall objectie o the ealuation study is to ...proide an assessment o smart subsidy
programmes in SSA, ocusing on obtained results as well as the way results hae ,or hae not, been
achieed, hindering and enabling actors, pre-conditions, etc.` ,1erms o Reerence, included in the
Appendix,. \e ealuate the oerall perormance o selected input subsidy programmes, identiy the
most important actors aecting programme perormance, and outline areas where knowledge is scarce
or non-existent.
\e hae selected agricultural input subsidy programmes in our SSA countries, Malawi, Zambia, Ghana
and 1anzania to orm the basis or the ealuation study. 1he need to assess the subsidy programmes in
some detail dictates our ocus on relatiely ew cases. 1he Malawi programme has receied a great deal
o attention as an example o a successul input subsidy programme. It is massie in scale, targeting is
based on a oucher system, and deliery o inputs to armers is largely state-managed. It is also
reasonably well documented in the literature. 1he large scale programme in Zambia proides an
example o an alternatie non-oucher targeting system and eatures state-drien deliery. 1he Ghana
case demonstrates a relatiely small programme with a ery market oriented deliery system, and the
1anzania case proides insights into the targeting perormance o a oucher scheme. 1ogether, the
selected cases proide considerable breadth in terms o dierences in the scope, design,
implementation and outcome o the programmes, as well as surrounding conditions.
1he ealuation study is organised as ollows: In the next section we discuss what is meant by smart`
subsidies. 1he economic philosophy which is the oundation o the concept o smart subsidies also
orms the basis or our methodology as outlined in section 3. Section 4 ealuates each o the our cases
and section 5 summarises the lessons learned and section 6 concludes.
?= ./0'1%$&%0#$ ')2%& 3%43','53
?=<= ./0'1%$&%0#$ ')2%&3
\e use agricultural inputs as a common term or a range o materials, which may be used to enhance
agricultural productiity. Most important among these are ertilizers and improed seeds. All the
programmes reiewed subsidise ertilizers, and most o them combine ertilizers with improed seeds
in small packages.
2

1he use o agricultural inputs is undamental in modern agriculture in deeloped countries, and they
were a primary ingredient in the green reolution that swept through Asia and Latin America during the
60s and 0s. loweer, the green reolution largely by-passed SSA, and the use o agricultural inputs
remains ery low. In 2002-2003 Sub-Saharan Arican armers used on aerage 9 kg o ertilizers per ha
o arable land compared to 100 kg per ha in South Asia, 135 in Southeast Asia and 3 in Latin America
,Craword et al, 2006,. \hile agricultural production and productiity soared in Asia and Latin America
during the last our decades, they hae largely stagnated in Arica, resulting in a rising dependency on
imported grains and an increase in the number o undernourished people ,\iggins and Brooks, 2010,
luture Agricultures, 2010,.
?=?= @A)'"503#$B "3= @3C#0&B ')2%& 3%43',- 20(/0#CC53
Many Arican countries, including Kenya, 1anzania, Malawi, Zimbabwe and Zambia pursued large
scale uniersal` subsidy programmes rom the 1960`s up through the 1980`s ,Dorward, 2009,. 1hese
programmes were characterised by a goernment-controlled input ,and output, marketing system, in
which armers were supplied with agricultural inputs at controlled and subsidised prices, and oten on
heaily subsidised credit. 1he experiences under these programmes were mixed. 1he programmes
succeeded in raising input use by armers and increasing agricultural productiity in many cases.
loweer, they were extremely expensie, most subsidies tended to beneit relatiely well-o and better
connected armers, and the adances in agricultural productiity were dependent on continued
goernment support. lurther, the ertilizer subsidy programmes were prone to ineiciencies arising
rom high administratie costs, goernment monopolies and political manipulation ,Banul, 2010b,. As
the subsidy programmes were dismantled and input markets liberalised as a part o the structural
adjustment process in the 1980`s and 1990`s, input use and agricultural productiity declined ,Craword
et al, 2006,.
Ater a period o liberalised input markets by the end o the last century, new subsidy programmes
began to emerge in seeral Arican countries. 1he Malawian goernment pioneered the return to large-
scale subsidies in 1998, when it began distributing ree ertilizer to armers ,Banul, 2010b,. Other
countries, such as Nigeria, Zambia, 1anzania, Kenya, Ghana soon ollowed Malawi`s example. In 2006,
Abuja, Nigeria, hosted the Arica lertilizer Summit under the auspices o the Arican Union ,AU,, the
New Partnership or Arican Deelopment ,NLPAD, and the Goernment o Nigeria ,\awson, 2010,.
An important output o that summit was the Abuja Declaration on lertilizer or Arican Green
Reolution, in which AU member states set out to increase ertilizer intensity to an aerage o 50 kg,ha
by 2015. One o the instruments in a ie point action plan was to implement smart subsidy
programmes to improe access to ertilizers or small-holder armers.
Smart subsidy programmes are meant to address the shortcomings o the uniersal subsidies. 1o be
smart`, subsidy programmes should adhere to a number o design principles, which can be
summarised under the ollowing headlines ,Minde et al, 2008, 1iba, 2009,:
! 9#0/5&')/ 3251'8'1 8#0C503. Smart subsidies should be targeted speciically at armers, who do
not already apply agricultural inputs, as well as the poorest and most ulnerable households.
1his reduces the risks o displacing commercial ,non-subsidised, input sales and promotes pro-
poor growth.
! D#0E5&64#35, 3($%&'()3. Smart subsidy programmes should utilise and support the urther
deelopment o existing priate input supply networks, rather than supplant them with state-
controlled distribution systems. 1his enhances the eiciency o input deliery as well as
increases the likelihood that the programme has a sustained impact ater its termination.
3

! !F'& 3&0#&5/-. Smart subsidy programmes should deise credible exit strategies to put a time
limit on the support. 1his is primarily to reduce the risks that the programme becomes
hijacked` by political interests ,Dorward, 2009, and to acilitate long term sustainability. I
stakeholders expect the support to continue indeinitely they are less likely to prepare or sel-
sustained use o inputs on market terms. Also, a irm exit strategy helps control the costs o the
programme.
1he three characteristics are largely complementary. I subsidies are well targeted, the greater demand
or inputs is likely to encourage potential entrepreneurs to establish new businesses, which promotes
the deelopment o a competitie input market. loweer, i the subsidised inputs primarily displace
commercial input sales, priate dealers are hurt by the unair` state-supported competition and may
choose to exit the market, thereby reducing competition. Similarly, the more eicient is the targeting
and input deliery system, the more eectie and credible the exit strategy will be.
As we will see, none o the subsidy programmes considered in this study can be characterised as smart
subsidies in the pure sense based on these criteria. 1hey are based on the oerall idea, and they all
contain some o the elements mentioned aboe, but none o them go all the way.
G= D5&7(,($(/-
G=<= .33533C5)& 10'&50'#
1he characterisation o smart subsidies gien aboe suggests that the concept is based on the economic
principles o eiciency, equity and sustainability. \e will thereore apply these principles as our
assessment criteria. In the ollowing, we briely discuss each principle in relation to smart input
subsidies.
G=<=<= !88'1'5)1-
1here is strong eidence to suggest that agricultural inputs raise productiity substantially, and that they
are essential or sustaining intensie agriculture in the long term without depleting soil ertility
,Craword et al, 2006,. 1he obious question is thereore why so ew armers in SSA hae adopted the
use o agricultural inputs to capture some o these potential beneits. 1here are two possible answers to
this question: 1, the economic costs o deliering agricultural inputs to the armers are too high and the
beneits in terms higher production too low or adoption o agricultural inputs to be a proitable
inestment, or 2, certain barriers, what economists call market ailures, preent armers rom realising
the economic potential o agricultural inputs. I the irst answer is correct, agricultural input subsidies
are ineicient. Subsidies merely encourage the adoption o inputs, which are more costly to procure
than the beneits they proide. I the second answer is correct, subsidies may be eicient as they help
armers oercome the market distortions generated by the market ailures. 1he discussion below will
elaborate on this iew.
1he irst answer may be correct in some geographical areas and,or some periods o time. Due to
poorly deeloped inrastructure, the costs o transporting inputs to remote areas, particularly in land-
locked countries, are ery high. Banul ,2010b, suggests that around 50 o market ertilizer prices
across SSA can be attributed to transaction costs compared with e.g. 20 in 1hailand. I armer density
is also low, the potential demand or expensie agricultural inputs may be so low that agro-dealers will
ind it hard to coer the costs o setting up a shop. Coupled with relatiely low agricultural productiity,
the inestment could simply be unproitable, demand or inputs may not exist, and suppliers will be
4

unwilling to oer access to inputs. In such a case, input subsidies could boost demand and encourage
input suppliers to expand their presence to remote areas. loweer, the subsidies would be ineicient.
Some o the costs o supply would shit rom armers to the state, but the costs would still outweigh
the economic beneits. lunding or subsidies could be better spent on policies aimed at lowering the
transaction costs, such as inrastructure inestments and market deregulation.
1he proitability o agricultural inputs also aries signiicantly oer time. ligure 1 shows the world
market ertilizer and maize price indices rom 2000-2010. lrom 2005-2008 the world market price o
maize, one o the most important staple crops in SSA, almost doubled, which alone would make maize
production more proitable. loweer, in the same period ertilizer prices rose much aster than output
prices and reached record high leels in 2008. So i an inestment in ertilizers more or less broke een
in 2006, it would hae become ery unproitable in 2008. Again, in this case an input subsidy would be
ineicient as it would encourage unproitable use o inputs.
H'/%05 <I H50&'$'J50 #), D#'J5 20'153 ?KKK 6 ?K<K

Source: \orld Bank, Global Lconomic Monitor ,GLM, Commodities database.
Notes: Prices are real USD indices o world market prices.
1he second possible answer to why agricultural input adoption in SSA is so low suggests that market
ailures exist to distort input markets and discourage armers rom using agricultural inputs. Lxamples
o market ailures most requently cited in the literature are credit constraints, imperect competition
and risk o crop ailure ,Dorward, 2009,:
! :05,'& 1()3&0#')&3: I armers are unable to obtain the necessary unding ,or i credit costs are
too high,, they may not be able to make an otherwise proitable inestment in agricultural
inputs. 1his is what Dorward ,2009, reers to as the aordability problem. A subsidy reduces
the unding needs, but may not necessarily resole the distortion completely, as armers still
hae to coer the subsidised prices.
! >C250851& 1(C25&'&'()I I agricultural input markets are imperectly competitie, input
suppliers tend to charge higher prices in order to capture greater proits or to coer more
ineicient business practices. 1his may result in armers not being able to aord inestments,
which would be proitable with a more competitie market. In this case, an input subsidy can
hae both positie and negatie consequences. It may increase aggregate demand, attract new
entrants to the market and increase competition. loweer, i this does not happen, or instance
5

i the demand impact is too weak or i the subsidies are implemented in a way that aours
incumbents, the subsidy may largely beneit the imperectly competitie irms.
! L'3E (8 10(2 8#'$%05: Inesting in agricultural inputs is a risky business, particularly since many
hybrid seeds and ertilizers require a reasonably well timed application and stable water supply.
A season o prolonged drought can largely wipe out the entire inestment and generate
signiicant losses. Particularly the poorest smallholders are ery ulnerable to poor harests and
may not be able to absorb the costs o a ailed inestment. Rather than risk losing eerything,
they may choose not to apply agricultural inputs, settling or a smaller but more stable surplus.
Agricultural input subsidies increase the expected beneits o the inestment and reduce the
costs o a ailed inestment.
It ollows rom this discussion that input subsidies may be eicient i they counteract distortions
generated by market ailures and ineicient i they do not. loweer, market ailures are hard to
measure, and estimates o how subsidies aect their distortions are usually not aailable. In practise, it
will be diicult to clearly distinguish between unproitable input use and market ailures. lor instance,
the time dimension complicates matters greatly. Looking only at 2008 when ertilizer prices were ery
high, we may come to the conclusion that ertilizer use was ineicient or many armers and should be
discontinued. loweer, a sudden drop in ertilizer demand could hae detrimental eects on the input
market by driing ulnerable suppliers out o business, which could oer time exacerbate problems o
imperectly competitie markets and limited access to inputs in some areas. \ould it in such cases be
better to intensiy subsidisation o inputs to prop up demand, thereby sacriicing short term eiciency
or long term gains
Another dilemma relates to the poor state o deelopment o the priate input markets in many SSA
countries. Proponents o smart input subsidies ,e.g. Minde et al 2008, 1iba, 2009, emphasise taking a
market-oriented approach to ensure eicient deliery o the subsidised inputs to armers. loweer,
particularly in remote rural areas, priate input suppliers may not exist because marketing costs are too
high and the customer base is too small. Programme designers thereore ace the dilemma o whether
to pursue a market oriented approach or to establish a state-managed supply system. 1he ormer option
would be more eicient and sustainable, but might not hae the capacity to reach the remote areas. 1he
latter option could more eectiely ensure a broader geographical coerage, but could also risk
undermining the deelopment o a more competitie priate input supply sector.
1he choice is essentially one between greater eiciency and sustainability on the one hand and
,geographical, equity on the other. loweer, the distinction between high transaction costs and market
ailures as causes o missing priate markets inluences the balance between eiciency,sustainability
and equity. I the armers` lack o access to priately supplied inputs is mainly due to high transaction
costs, priate input suppliers may be less willing to establish local outlets as they may expect demand to
disappear again ater the subsidies are phased out. On the other hand, i the missing markets are mainly
due to market ailures, a market oriented approach may help alleiate these market ailures, or instance
by enhancing competition in input supply. As mentioned aboe, the distinction is hard to make in
practise, and the situation is likely to ary greatly between countries, or een regionally within countries.
Bearing these challenges in mind, we will make an assessment o the likely eect o subsidies on market
distortions and the proitability o inputs, outline possible trade-os, and identiy areas where more
knowledge is needed or a clearer assessment.
6

G=<=?= !M%'&-
Agricultural input subsidies can be a useul instrument or promoting greater equality by targeting
subsidies speciically at the poorest smallholders. loweer, it is not entirely clear whether such
redistributie objecties are compatible with the eiciency criteria. On the one hand, the poorest
smallholders are most likely the ones that are most constrained by market ailures, such as credit
constraints and ulnerability to the risks o crop ailures. On the other hand, poor subsistence armers
may lack complementary resources, such as skills, scale o operation, productie assets, or the inancial
resources to pay een the subsidised prices, to make eectie use o the subsidised inputs. In other
words, use o agricultural inputs by poor smallholders may simply be unproitable een i
unconstrained by market ailures.
1hus, there may be a trade-o between equity and eiciency objecties. I the primary aim o a subsidy
programme is to achiee pro-poor growth, targeting the most ulnerable households may increase
equality at the expense o eiciency. Similarly, an objectie o increasing national sel-suiciency in
grain production will require the programme to target the most productie households, who may be
somewhat less-poor. Most o the programmes reiewed in this study hae both objecties in some
orm. \e proide an assessment o the programmes` intended targets, how well these targets are hit,
and how well the programmes perorm in terms o a general equity criterion.
G=<=G= +%3&#')#4'$'&-
Subsidy programmes are sustainable i they can be maintained oer the long term without draining the
public resources, or i the outcomes in terms o wider adoption o agricultural inputs and improed
agricultural productiity persist ater their termination. 1he uniersal input subsidy programmes
pursued by many SSA countries during the 0`s and 80`s largely ailed on both accounts.
Long term subsidy programmes may be economically justiied as long as they meet eiciency and
equity objecties. 1here are, howeer, political economy reasons to be sceptical about long term
programmes. Subsidies represent a signiicant alue, which is transerred rom the state to armers,
suppliers and other stakeholders. As such, stakeholders hae a great and obious interest in the
continuation and expansion o subsidies. In particular, when subsidies are rationed and targeted at
speciic groups, the people controlling how subsidies are targeted may exploit their power or personal
gain. Policy makers may also be inclined to expand the goernment support irrespectie o its
perormance, as it signals leadership and willingness to act. 1he politics o input subsidisation thereore
carry a risk that the programme gains a lie o its own, grows more ineicient and less equitable, and
eentually becomes unsustainable.
1o counter these eects, smart subsidies are meant to be a temporary measure designed with a clear
exit strategy detailing the termination o the programme. A sustainable smart subsidy programme seeks
to aect a permanent impact by a short term boost, or in other words to kick-start` the market or
agricultural subsidies. 1he permanent impacts can be achieed by alleiating the market ailures
plaguing the input markets directly or by raising the productie capacity o poor smallholders to a
suiciently high leel that the market ailures are no longer constraining. lor instance, i the subsidy
programme succeeds at permanently deeloping a more competitie priate input supply, the lower
prices will make inputs more widely accessible to smallholders. Similarly, i the programme helps
smallholders accumulate productie and inancial assets rom a ew years o surplus harests, the
armers may be able to inance ull-priced inputs rom their own saings ater programme termination.
On the other hand, i market ailures simply maniests again, once the programme ends the eects are
likely to proe short-lied. 1he ealuation study will assess whether and how the subsidy programmes
considered here are likely to hae a lasting eect on potential market ailures or households` assets.


G=?= 975 2($'&'1#$ 51()(C- (8 ')2%& 3%43','53
\hile the ealuation study will primarily ocus on the economics o input subsidy programmes, the
political economy o input subsidies cannot be completely ignored. Ideally, policies would be
implemented to maximise national welare, but it is naie to beliee that personal political motiations
do not play a role. In act, Dorward ,2009, argues that political economy diiculties are particularly
problematic in poor rural societies, as 1, the potential personal and political gains rom subsidy rents
are ery large relatie to other income opportunities, so incenties or political manipulation are strong,
and 2, iscal resources are ery scarce and costly to collect, so the aderse consequences o wasteul
policies are great.
Irrespectie o the economic justiications or large scale input subsidy programmes, their political
beneits may be substantial Input subsidies are eectiely transers o alue rom the goernment
directly to recipients, so beneits are immediate and easily recognised. 1hey may generate relatiely ast
and easily obserable results in terms o greater ood production, which allows policy makers to signal
strong leadership and decisieness. Subsidies can be narrowly targeted at speciically aoured
constituents, while excluding others, and they may just as easily be taken away again i political
objecties are not met. 1hus, it is possible that the popularity o large scale input subsidy programmes
in SSA is mainly due to their political attractieness rather than economic superiority. Banul ,2010b,
suggests that historical anecdotal eidence supports this iew. 1he uniersal subsidy programmes were
maintained or many years in spite o strong indications o their ineiciencies and unsustainable drain
on iscal resources. It took heay pressure rom outside donors and the threat o imminent iscal
collapse to push through liberalising reorms.
G=G= +&0%1&%05 (8 &75 3&%,-
\e structure the study o the input subsidy programmes into the ollowing headlines, which consider
the most important elements o smart subsidy programmes:
! N#1E/0(%),I \hat were the motiations or the subsidy programmes
! O%&1(C5I \hat are the oerall outcomes o the subsidy programmes in terms o increases in
input use, agricultural productiity and output, changes in market prices, etc. and how do the
beneits compare to the costs o the programme
! +1(25I \hat is the size o the programme, how wide does it coer and what is the extent o
subsidisation
! 9#0/5&')/I \hat type o targeting and rationing mechanism ,e.g. oucher, is used \ho
controls beneiciary selection and which criteria sere as guidelines low well were the
intended targets reached
! P5$'"50-I low are the inputs deliered to the end users \ho supplies and who distributes the
inputs. 1o what extent is the priate sector inoled \hat is the risk o displacing commercial
input supplies Are inputs deliered in a timely and eectie ashion
! !F'& 3&0#&5/-I Do the programmes exhibit a clear exit strategy, is it credible and has it actually
worked as planned
! .33533C5)&I Based on the aailable inormation presented aboe, how will we assess the
perormance o the programme in terms o eiciency, equity and sustainability
In section 4 we consider each o the our cases in turn, and in section 5 we summarise the lessons
learned using the same headlines.
8

G=Q= +5$51&'() (8 1#353 #), $'&50#&%05
\e hae selected relatiely ew cases to orm the basis o the ealuation study to allow a certain depth
and detail in our scrutiny. 1he our cases are
! Malawi: Agricultural Input Support Programme 2005,6 - ongoing ,AISP,
! Zambia: lertilizer Support Programme 2002,3 - ongoing ,ZlSP,
! Ghana: lertilizer Subsidy Programme 2008 - 2009 ,GlSP,
! 1anzania: National Agricultural Input Voucher Scheme 2008 - ongoing ,NAIVS,
1hese our cases are the ones that show up most requently in our literature searches. 1hey proide a
airly wide ariation in terms o programme size, objecties, targeting and deliery mechanisms, degree
o success and empirical eidence. By choosing these cases, we aim at orming a reasonably clear
picture o recent experiences with input subsidies.
1he ealuation study is based on two types o literature, i, oicial programme documents, reiews and
ealuations, and ii, empirical research papers. 1he irst set o literature typically coers a broad range o
issues, though not always in great depth. 1he second group inestigates selected eatures ,such as
targeting, outcome, etc., in more detail, usually based on relatiely narrow samples. \e ocus on the
most recent literature, partly to coer the latest eidence and partly because some o the programmes
,particularly the ones in 1anzania and Ghana, are relatiely new.
Q= +5$51&5, ')2%& 3%43',- 20(/0#CC53 ') +%46+#7#0#) .80'1#
1his section reiews and ealuates the our selected input subsidy programmes. 1able 1 summarises the
main eatures o the our programmes.

9

9#4$5 <I O"50"'5R (8 &75 8(%0 1#35 ')2%& 3%43',- 20(/0#CC53
Malawi
Agricultural Input Support
Programme
Zambia
lertilizer Support Programme
Ghana
lertilizer Subsidy Programme
1anzania
National Agricultural Input
Voucher Scheme
Outcome


! Output
23 - 54 89 - -
! Lconomic returns
Positie, but small and olatile Likely negatie - -
Scope
! Programme costs
USD 114 -285 million USD 4 million USD 14-26 million USD 100 million,year
! GDP
2-3 0.4 - -
! GDP in 2008,9
0.9 0.05 - 0.1 0.4
! Subsidy
64-9 60 - -
! Subsidy in 2008,9
91 80 50 50
! households coered
50-65 - - 45
1argeting
! 1argeting mechanism
Voucher larmer registry Voucher Voucher
! Package size
50 kg ert. - 2-4 kg seed 400 kg ert. - 20 kg seed No standard package 100 kg ert. - 10 kg. seed
! Beneiciary selection
Village Deelopment
Committee
larmer cooperatie Lxtension oicers Village Voucher Committee
,locally elected,
! 1argeting criteria
Priority to ulnerable
households
Relatie less poor households None Lmphasising ability to pay or
and use inputs
Deliery
! Input supply
Contracted priate irms Contracted priate irms Large priate importers Priate sector
! Input retail
Mainly parastatal Contracted priate warehouses Priate sector Priate sector
! Subsidy redemption
Retailers larmer cooperatie lertilizer importers National Microinance Bank
Lxit strategy
! Clear exit strategy
None 2-year beneit, 3-year prog. Single year programme 3-year beneit, 3-year prog.
! Lxit strategy enorced
- No One-year extension Ongoing
10

Q=<= D#$#R'S ./0'1%$&%0#$ >)2%& +%22(0& T0(/0#CC5UH#0C >)2%& +%22(0&
T0(/0#CC5 V.>+TW
<

Q=<=<= N#1E/0(%),
In Malawi 88 o the population lies in rural areas and slightly more than hal o these are poor. 1he
rural households are almost exclusiely maize producers, but only 10 o them are net sellers, whereas
around 60 o smallholders are net buyers o maize. 1his dependency on market purchases o maize
leaes poor households ulnerable to the high and olatile maize prices usually obsered in Malawi.
1hus, the political moties or supporting improements in agricultural productiity are largely drien
by a desire to increase smallholder sel-suiciency in maize production and reduce their exposure to
maize market risks. As ormulated by Chinsinga ,2011,, In Malawi, maize is politics, and input
subsidies are central to this`.
1he Agricultural Input Support Programme ,AISP, in Malawi, initiated in the 2005,6 season, builds
upon a long tradition o subsidising agricultural inputs.
2
1he recent wae o smart` subsidies started
with the establishment o a programme to distribute ree starter packs` in 1998,99, initially to all
households, but by 2001,2 to a more limited number o targeted households ,1argeted Input Program,
1IP,. 1his starter pack programme proided aluable experience in beneiciary targeting and
establishment o logistics systems to distribute the inputs to rural households. Partly as a result o these
early attempts, Malawian armers were een prior to the AISP some o the most intensie ertilizer
users in Sub-Saharan Arica, using around 30 kg,ha compared with a SSA aerage o 9 kg,ha. As the
1IP was scaled down in the 2004,5 season, Malawi was hit by bad weather resulting in a ery poor
maize harest, which translated into high prices and acute ood shortages ,Chibwana et al, 2010,. In
response, the Malawi goernment reintroduced large scale input subsidies in the orm o the AISP in
2005,6. 1he oerall objectie o this programme is to ...increase resource poor smallholder armers`
access to improed agricultural inputs in order to achiee ood sel-suiciency and to increase resource
poor smallholder armers` income through increased ood and cash crop production` ,Dorward et al,
2010, p. 12,.
Q=<=?= O%&1(C5
1he AISP appears to hae had a substantial eect on maize output, but some uncertainty surrounds the
estimates due to other actors ,e.g. the weather, inluencing the data. Oicial estimates suggest that
national maize harests increased by around 1 million tonnes in 2005,6 rising to more than 2 million
tonnes in the 2008,9 season ,around 54 and 114, compared to the 2002,3 and 2003,4 seasons.
1hese estimates are, howeer, highly contentious. More conseratie estimates by Dorward et al ,2010,
based on assumptions about aerage maize responses to ertilizer, put the increase in maize output at
around 400,000 tonnes in 2005,6 to 1,000,000 tonnes in 2008,9 ,corresponding to an increase o 23
and 54, compared to pre-AISP harests. 1hese estimates by Dorward et al ,2010, are deried by
multiplying the estimated increase in ertilizer use with an estimated aerage ertilizer response rate,
adjusting or weather, use o improed seeds and deliery delays. lor the lack o better statistics, this
approach seems reasonable, but the estimates must be iewed as highly uncertain.
A ew studies hae tried to quantiy the impact o the input subsidy based on a more stringent scientiic
,econometric, methodology. Ricker-Gilbert and Jayne ,2010, try to estimate the dynamic eects o the

1
1his section is based on Dorward et al ,2010, and Dorward and Chirwa ,2011, unless otherwise speciied.
2
1he programme changed its name to larm Input Support Programme ,lISP, in 2008,9. As the new programme is
essentially a continuation o the old AISP, we reer to the whole programme by AISP.
11

AISP. 1hey ind that a ertilizer subsidy signiicantly increases maize production within the same year,
and there are some indications o positie eects on maize production in subsequent seasons but these
are surrounded by greater uncertainty. On the other hand, they ind little eidence o a long-term eect
on household assets or general wellbeing.
An impact assessment based on household sureys by Chibwana et al ,2010, suggests that the
programme increased maize yields o recipient armers by 44 kg,ha ,around 42,, o which just oer
hal ,249 kg,ha, can be attributed to ertilizers and the rest to improed seed. Such production
increases are within the range estimated by Dorward et al ,2010, mentioned aboe. Chibwana et al
,2010, also report that the AISP caused some change in cropping patterns, as armers reallocated land
rom alternatie ood crops such as cassaa or sweet potato towards maize. 1o the extent that ertilized
maize is more productie, this shit represents a urther expansion in ood production, but it may also
increase risks o crop ailures as cassaa is more resistant to droughts than maize ,Barratt et al, 2006,. It
is not possible to assess the seerity and implications o these eects based on the aailable material.
1he eidence reiewed aboe indicates that the expansion in maize output is sizeable, and it is thereore
all the more puzzling that the eect did not translate into an obserable decline in maize prices. During
the decade preceding the start o the AISP, aerage local maize prices oscillated around a price o USD
0.2,kg ,1990 prices,, perhaps with a slightly declining trend. 1he larger harests obtained rom 2005,6
onwards did not result in the anticipated drop in maize prices, and prices ollowing the 2008,9 record
harest actually rose by almost 100 oer the preious season. A number o possible explanations are
oered, among which the most important is that demand seems to hae increased at least as ast as
supply. lor instance, partly due to oicial oerestimation o maize harests in 200, the goernment
contracted with the goernment o Zimbabwe to export 400,000 tonnes o maize to Zimbabwe. 1he
goernment managed to export only around 283,000 tonnes beore suspending the contract due to
rapidly increasing domestic prices ,Minde et al 2008,. Similarly, in 2009,10 the goernment added
130,000 tonnes o maize to the strategic grain resere and priate traders accumulated a urther 100,000
tonnes ,Dorward et al 2010,. A second reason or the exceptionally high maize prices ollowing the
2008,9 season could be the extremely high ertilizer prices during this season, which are partly passed
through to output prices. Although beneiciaries o the AISP were shielded rom the ertilizer price
increase, it is likely that more commercially oriented armers supplying the maize market as well as
maize importers were more exposed.
Lidence on the eects on poerty is harder to obtain. Dorward et al ,2010, reports on indings rom
ocus group discussions, which suggest that rural real wages increased continuously oer the AISP
lietime een or poor non-beneiciaries. As maize production by AISP beneiciaries increases, the
households` dependence on o-arm work is reduced and more jobs are aailable or non-beneiciaries
and land-less poor. It is not possible to judge how strong or widespread such eects were, or to which
extent the reported reductions in poerty rates can be attributed to the AISP.
Dorward et al ,2010, calculates the economic returns o the AISP based on their estimates o the eect
on production. 1he economic returns, deined as the net beneits relatie to total costs, ary
considerable depending on the weather, maize and ertilizer prices, assumptions about yield responses,
etc. Assuming a moderate yield response, the estimates suggest that the ery good conditions preailing
in the 2006, season produced decent economic returns o around 54, implying that a USD 100
inestment in programme actiities generated a USD 154 worth o output. More modest returns were
achieed in the 2005,6 ,despite plentiul rains, and 200,8 seasons ,12 and 6 respectiely,.
loweer, the 2008,9 season generated negatie returns despite good weather and high maize prices
due to extremely high ertilizer prices. In conclusion, these estimates suggest that economic returns are
likely to be modestly positie on aerage but with a high degree o olatility.
12

Q=<=G= +1(25
1he AISP is a massie programme, and the scope o the programme has expanded continuously oer
its lietime, as illustrated in table 2 below. By the 200,8 season, the programme proided what
amounted to an aerage 9 subsidy to 59 o all arming households. 1otal costs were around USD
11 million, corresponding to 3.4 o GDP ,9 o the total goernment budget,. 1he ollowing
season 2008,9 was exceptional due to extremely high ertilizer prices. 1he goernment decided to keep
the prices paid by armers constant and absorb the input price increase. As a result, the subsidy ratio
aeraged 91, and the total costs o the programme jumped to almost USD 285 million ,6.6 o GDP
and 16 o the public budget,. lertilizer prices hae come down again somewhat since 2009, but we
do not hae any later igures on programme expenditure. Initially, the programme subsidised mainly
maize ertilizers and to a lesser extent tobacco ertilizers, but it has since then branched out to include
maize seeds ,o which 84 were hybrid,, as well as a bit o cotton seed, legume seed and cotton
pesticides.
9#4$5 ?I +1(25 (8 &75 .>+T ') D#$#R'
2005,6 2006, 200,8 2008,9
Programme budget
,USD million,
36 54 82 139
Programme costs
,USD million,
51 91 11 285
o GDP 2.1 3.1 3.4 6.6
budget oerrun 42 69 43 105
household
coerage
N,A 54 59 65
subsidy 64 2 9 91
Source: Dorward et al ,2010,
Q=<=Q= 9#0/5&')/
1he AISP is based on a oucher system. Selected recipient households receie two coupons, each o
which can be redeemed or a bag o maize or tobacco ertilizer or a bag o maize seed ,hybrid or Open
Pollinated Variety, OPV,.
3
1he ertilizer bags carry an additional ixed price ,MK 950,, but the seed
bags require no extra expenditure. 1hus, in eect AISP oers an input subsidy as well as an input price
control mechanism, under which the programme absorbs all input price ariations. 1his is both a
strength and a weakness o the programme. One the one hand, the stability o input prices ensures that
smallholders can gradually build capacity or applying inputs, maintain soil ertility, and procure
improed seeds season ater season. Small input suppliers may deelop and solidiy their businesses
without sudden demand disruptions. On the other hand, when the goernment carries all the risk,
programme costs are diicult to control and the interention risks becoming unsustainable. In all years,
the realised programme costs were oer budget, in 2008,9 by more than 100 ,see table 2 aboe,.
Coupons are printed centrally and distributed to the district leel. Initially, the distribution o ouchers
to districts was based on historical cropping patterns, but rom the 200,8 season greater emphasis was

3
Maize ertilizer consists o a 50kg bag o 23:21:0 -4s or Urea, tobacco ertilizer coers a 50 kg bag o Compound D or
Calcium Ammonium Nitrate ,CAN,, and maize seeds come in 2kg bags o hybrid seeds or 3-4kg bags o OPV seeds.
13

placed on the number o arming households. At the district leel, 1raditional Authorities allocated the
ouchers among illages and the local Village Deelopment Committees were responsible or
identiying recipients. 1he oicial criteria or determining beneiciaries are not ery precise. 1hey
mainly stipulate that recipients should own land being cultiated in the releant season and that priority
should be gien ulnerable groups, particularly emale-headed households. As a result, much discretion
was let to illage chies and there appears to be considerable ariation between regions. lousehold
sureys suggest that coupons were disproportionately allocated to households with relatiely more land
more assets and to male-headed households. Such eidence is supported by Chibwana et al ,2010, who
ound that the most ulnerable and emale-headed households were less likely to receie ouchers,
whereas long term residents o illages were more likely to be selected.
Inappropriate targeting o programme beneits risks displacing commercial input sales, i recipients
would hae bought agricultural inputs anyway in the absence o subsidies. Commercial sales
displacement represents a shit in input sales rom non-subsidised to subsidised sales and does not
increase the total use o agricultural inputs. A high degree o displacement is thereore detrimental to
programme eectieness. No irm data exists on the extent o displacement, but Dorward et al ,2010,
suggest that around 30 o subsidised maize ertilizer sales would hae taken place on commercial
terms i there had been no support. 1hese estimates are based on examinations o changes in aggregate
sales in 2005,6 and 2006,, and the authors reer to supporting eidence rom a panel data analysis o
armer purchases. I such estimates are correct, displacement represents a substantial reduction in the
eectieness o the programme. Len worse it undermines the eorts to deelop a competitie priate
input supply sector, as priate suppliers that are not part o the programme ace strong state-subsidised
competition.
Q=<=X= P5$'"50-
larmers can redeem ouchers at retail outlets o irms, which are selected each season through tender.
\holesale supply ,imports, o ertilizers is mainly undertaken by large priate irms, but two large
parastatals, ADMARC and SllRlM, dominate the distribution and retailing o ertilizers to the
households. In the 2006, and 200,8 seasons, a ew large priate companies ,also inoled in
importation and wholesaling, with deeloped retail networks were authorised to market some o the
inputs and were responsible or 24-28 o subsidised ertilizer sales, but in the 2008,9 season no retail
contracts were awarded to priate distributors. 1he ery limited inolement o the priate sector in
input distribution may in part hae been due to distrust between the goernment and the priate sector
,Dorward et al, 2008,.
1he market or improed maize seeds is largely priate, with both large retail chains and smaller
independent agro-dealers distributing subsidised seed. Supply o improed seeds is, howeer,
dominated by multinational corporations, including Monsanto, which alone controls 50 o the market
,Chinsinga, 2011,. 1his concentration o market power is exacerbated by the AISP, partly due to
political priorities. 1he subsidy programme oers both subsidised hybrids, which are exclusiely
imported by multinationals, and OPV maize seeds, which are supplied by local seed companies.
Chinsinga ,2011, argues that oer the lie time o the programme attention has shited more and more
towards hybrid seeds. lybrid maize seeds generate higher yields than OPVs and are thereore more
attractie or policy makers, who want to show ast results. loweer, NGOs argue that OPVs are
more suitable or smallholders, as they are more resistant to pests and diseases, more drought resistant
and more amiliar to armers. Crucially, harested OPV maize may be retained as seeds or the next
season, unlike hybrid seeds, which must be bought rom the market each season. 1hus, adopting
subsidised hybrid seeds may generate a dependency on the multinational producers, which may proe
deastating or smallholders once subsidies are phased out.
14

1he eiciency o the programme is also aected by the timing o delieries and the extent o raud and
corruption by programme stakeholders. In Malawi, agricultural inputs should be aailable to armers by
end o Noember to ensure their eectie use. Although the timing o input delieries has improed
oer the lietime o the programme, only 30 o all sales had arried by end Noember in 2008,9
season, most o the rest was deliered during the ollowing month. 1aken together with relatiely high
displacement rates, late deliery is detrimental or the objectie o increasing ertilizer use. Recipients o
input ouchers may hold o purchases o inputs on commercial terms in the expectation o receiing
subsidised inputs shortly. I these inputs are deliered too late, armers may actually end up reducing
input use rather than merely displacing commercial inputs.
1he extent o raud and corruption is diicult to determine, but ragmental inormation based on ocus
group discussions and household sureys suggests that some problems do exist ,Dorward et al, 2010,.
A small number o households ,5, report haing to pay a ee or gaining access to ouchers and an
estimated 14-20 o ouchers were redeemed together with a small tip` on top o the regular
beneiciary co-inancing. 1here were also reports o oucher countereiting, which in 2008,9 accounted
or 2 o sales by parastatals and 3 o priate retail sales. Dorward et al ,2010, suggests that the
better perormance o priate retailers could be attributed to the act that they were much aster at
returning ouchers to the programme or inal settlement, which allowed rapid identiication and
termination o countereiting schemes.
Q=<=Y= !F'& 3&0#&5/-
1he programme does not appear to hae any exit strategy. It has been in eect or 6 years, and its
budget has expanded continuously throughout its lietime, rom MK 5.1 billion in 2005,6 to MK 21
billion in the 2011,12 budget ,Goernment o Malawi, 2011,. 1here are no indications that a phase out
o the programme is planned in the uture.
Q=<=Z= .33533C5)&
All eidence indicates that the Malawi AISP has a substantial positie eect on the use o agricultural
inputs, agricultural productiity and ood production. loweer, the gains come at a massie cost to the
Malawi goernment budget ,direct donor support only coered 5 o total programme costs in
2008,9,, which could alternatiely be used or inesting in inrastructure, education, health, etc. 1he
best estimates aailable on the economic returns o the programme appear to be positie but also
relatiely modest and highly olatile depending on input and output prices, weather conditions,
displacement rates and eiciency o programme administration.
One o the most important sources o uncertainty is the design o the instrument as a ariable subsidy
with a ixed armer payment. 1his proides the greatest degree o security to armers, as they are
shielded rom input price olatility. loweer, it also reduces the oerall eiciency o the programme.
1he high ertilizer prices in 2008,9 were a signal rom the market that ertilizers were in short supply.
lrom an eiciency point o iew it is better in such a situation to economise o the use o ertilizers,
which would most likely hae taken place i armers aced at least a part o the price rises. In the eent,
total subsidised ertilizer sales declined only slightly resulting in massie budget oerruns.
Usually, some eiciency must be sacriiced in aour o a more equal distribution o resources.
loweer, the reiew o the AISP suggests that the objectie o targeting the most poor and ulnerable
households is ery diicult to achiee or two reasons. lirstly, there must be some mechanism or
identiying worthy beneiciaries, in this case Village Deelopment Committees ,VDCs,. Although there
seems to be scope or ine-tuning targeting criteria, VDCs will still hae considerable discretion in
allocating ouchers, leaing room or abuse o powers, rent seeking and inluence by local politics. 1he
15

most ulnerable are unlikely to rank highly with such priorities. Secondly, households need to hae a
certain minimum productie capacity to be able to use the subsidised inputs eectiely, such as skills,
land, inancial resources, complementary assets and labour. On the other hand, although the poorest
households are not targeted directly, they may still gain indirectly by lower ood prices and greater job
opportunities proided by the general improements in agricultural productiity. Lower ood prices did
not materialise in Malawi ollowing the implementation o the AISP, but Dorward et al ,2010, suggest
that rural real wages may hae increased anyway by a tightening o the rural labour markets.
1here is considerable risk related to the long-term sustainability o the programme in its present orm.
1otal costs are extremely high relatie to the goernment budget and the total economy. \hat is worse,
the budget has been increasing steadily throughout its lietime and it has proed diicult to control as
eidenced by high and increasing budget oerruns. 1hese trends are exacerbated by the lack o credible
exit strategy. It is questionable whether the Malawi goernment is able to bear the burden in the long
run.
A second aspect o sustainability is the extent to which the higher adoption o agricultural inputs is
likely to persist ater a possible uture programme termination. 1his is doubtul. 1here is little to
suggest that programme actiities hae addressed the underlying barriers to a well-unctioning
agricultural input market, such as a thin priate input supply network, lack o access to credit and high
costs o input deliery. Distribution and retailing o ertilizers is still dominated by parastatals, and there
seems to be little scope or new suppliers to capture a share o the market, though the market or
improed seeds appears to be more competitie. 1he high inancing and transaction costs are not
reduced but merely shited rom armers to goernment. A sustainable outcome could be achieed i
beneiciaries were able to accumulate suicient inancial and productie assets to oercome the market
barriers on their own ater programme termination. loweer, the little eidence that exists ,Ricker-
Gilbert and Jayne, 2010, suggests that such capital accumulation is not happening. Should the AISP be
dismantled in the uture, there is a substantial risk that much o the gains in terms o more widespread
agricultural input adoption will be reersed.
In conclusion, the Malawi AISP has a large eect on productiity and output, but the programme is
ery costly, it largely ails to target the most ulnerable households and its long term sustainability is
questionable.
Q=?= [#C4'#S H50&'$'J50 +%22(0& T0(/0#CC5 V[H+TW
Q

Q=?=<= N#1E/0(%),
1he lertilizer Support Programme ,ZlSP, in Zambia ollows earlier attempts at stimulating the
adoption o agricultural inputs, mainly ertilizers and hybrid seeds, in the production o maize. Larlier
programmes ocused less on direct subsidies and more on controlling input prices and making sure that
inputs were aailable to smallholders through state-managed production and distribution. Indirect and
unintentional subsidisation was proided in the orm o state-proided credit, o which only 5-10
was recoered.
In 2001 the goernment estimated that only 30 o smallholders had access to improed seeds and just
20 had access to ertilizers. Small-scale armers had too ew inancial resources to generate suicient
demand to support a competitie priate input supply sector. In this context, the ZlSP was launched at

4
1his section is based on \orld Bank ,2010, unless stated otherwise.
16

the start o the 2002,3 agricultural season. It sought to break rom earlier programmes by subsidising
inputs directly rather than proiding credit and by ocusing on the deelopment o a competitie
priate input supply sector rather than relying on state-managed distribution. Speciic objecties o the
programme can be summarised as
1. 1o ensure timely, eectie and adequate access o smallholder armers to agricultural inputs in
the orm o ertilizer and hybrid maize seeds
2. 1o acilitate the deelopment o a competitie priate sector in the supply o agricultural inputs
3. 1o acilitate the process o armer organisation, dissemination o knowledge and creation o
other rural institutions that will contribute to the deelopment o the agricultural sector.
Q=?=?= O%&1(C5
Although eidence suggests that the ZlSP was less eectie than anticipated by the goernment, it
does appear to hae a substantial eect on maize yields and production o participants. In designing the
programme, the goernment expected that armers were able to achiee a maize yield o 3 tonnes per
hectare, almost a trebling o aerage yields among smallholders in Zambia. Instead, a surey conducted
by the end o the 200,8 season showed that participants achieed an aerage yield o around 2 tonnes
per hectare, albeit with large regional ariations.
In aggregate, \orld Bank ,2010, estimates that total production in Zambia increased by 146,000 tonnes
o maize 200,8, corresponding to 89 growth in output as a result o the ZlSP. 1his increase coers
output due to higher yields ,estimated as 82,000 tonnes or 50 yield increase, as well as expansion in
the area cultiated by maize ,around 64,000 tonnes,. 1hese estimates are characterised by considerable
uncertainty. 1hey are based on a household surey o ZlSP beneiciaries, the results o which are
extrapolated to the national leel. In the surey, armers are asked how much they produced in the
current season compared with seasons prior to support rom the ZlSP. So the estimates are deried
rom armers responses ,partly based on their recollection o past production,, and there are apparently
no attempts at controlling or actors unrelated to the programme. 1hey are, howeer, the best
estimates aailable.
1he \orld Bank ,2010, study also estimates the total costs o the programme, amounting to ZMK 183
billion, or USD 4 million, including direct costs o the inputs, administration and logistics, as well as
the indirect costs o salaries paid to goernment sta in proportion to the resources spent on the
programme and armer contributions. 1hese cost estimates imply that the increase in maize supply was
made possible at a cost o around USD 325 per tonnes at the arm gate. In comparison import prices
luctuated between USD 295 and USD 406 per tonnes during the period under inestigation ,200-
2009,.
On the ace o it, this suggests that the programme has been reasonably proitable rom a national
perspectie. 1here are, howeer, costs that are not suiciently accounted or in the estimate. Applying
agricultural inputs, particularly ertilizers, is more labour intensie and the opportunity costs o labour
are not included. Also, arm gate prices do not include the oten considerable transport costs to urban
centres. linally, a substantial part o the incremental production is attributable to an increase in the area
cultiated by maize. 1o the extent that the increase in maize plantings is brought about by displacing
alternatie crops, the alue o the displaced crops should be added as a urther opportunity costs. Len
where land is ree`, continued land expansion may not be sustainable in the long run. I the eect on
land expansion is actored out, the resource costs o the ZlSP amount to USD 59 per tonnes,
substantially higher than the import prices. As a means o increasing ood security in outlying areas, the
1

ZlSP may proide reasonable alue or money, but as a source o increased national ood supply the
programme appears to be largely uncompetitie.
Q=?=G= +1(25
1he ZlSP is designed to reach around 125,000 arming households, although in 2006, and 2008,9
the goernment budgeted or some 200,000 households. It is not entirely clear why the planned
number o beneiciaries changed so much oer time, but it may hae something to do with the
goernment budgeting process. According to goernment unding rules, the total budget or the
programme must be negotiated each iscal year and substantial ariation in the budget rom year to year
is possible. 1his is also a source o serious administratie diiculties as discussed in more detail below.
1he ZlSP subsidised inputs by around 60, but the goernment contribution increased to as much as
80 ollowing the extremely high ertilizer prices during the 2008,9 season. In total, the USD 4
million estimated cost o the programme in 200,8 corresponds to around 0.4 o GDP and 1.6 o
the public budget ,the budgeted costs or 2008,9 season amounted to around 0.9 o GDP. Although
not as massie as the Malawi AISP reiewed aboe, the programme is still considered large scale.
Q=?=Q= 9#0/5&')/
larmer cooperaties, speciically approed by the goernment, play a central role in identiying
beneiciaries and collecting the armers` payments, which are deposited beore the inputs are handed
out. According to the targeting criteria, recipient households should
! be an actie small-scale armer
! hae the capacity to cultiate between one and ie hectares o land,
! be able to coer 40 o commercial input prices,
! should hae no prior history as a deaulter in earlier goernment subsidised credit programmes.
In addition, armers need to be a member o the cooperatie to beneit rom the ZlSP. Lach
beneiciary household is entitled to pick up a package o agricultural inputs, consisting o suicient
amounts o ertilizers and hybrid seeds to cultiate one ha. o land using the dosage recommended by
the goernment.
5
Compared to the alue o the ouchers distributed by AISP in Malawi, the ZlSP
packages are considerably larger, by a actor o around 8-10.
1he selection criteria as well as the size o the input packages relect a ocus on relatiely less-poor
arming households. 1he stipulation that beneiciaries should be able to cultiate at least one hectare o
land eectiely excludes the 40 poorest smallholders, who own less land ,Minde et al 2008,. Also, the
large input packages and the requirement that beneiciaries are members o an approed cooperatie,
which demands a membership ee, also seres to discriminate against the poorest. 1he bias against the
poorest smallholders is relected in the eidence characterising the recipients. Sureys o recipients
reeal that more than 85 o armers receiing support in 200,8 cultiated one hectare or more o
land ,Minde et al 2008,, and 35 o beneiciaries owned drat animals compared to a national aerage
o around 11 ,\orld Bank 2010,. 1his suggests that the ZlSP largely hit their intended targets, but
these targets were not the poorest smallholders.

5
Speciically, one package consists o our 50kg bags o compound D basal ertilizer, our 50kg bags o urea top dressing, as
well as one 20kg bag o hybrid seeds ,in short 20 kg seed - 4x4 ertilizer,.
18

1argeting the less-poor households risks undermining the eectieness o the programme through
displacement o commercial input sales, as these households are more likely to be able to inance input
purchases on market terms. Lidence to this eect is mixed. A surey o ZlSP beneiciaries suggests
that 50 o recipients o subsidised inputs bought inputs rom priate shops beore receiing ZlSP
support generating a strong potential or displacement. loweer, the same surey reealed that 43
continued to complement their subsidised package with inputs purchased on commercial terms rom
priate suppliers. It is not clear whether these households reduced their purchases o priate inputs, or
to which extent the subsidised packages inspired armers to adopt agricultural inputs more widely. 1he
\orld Bank ,2010, study concludes that displacement constituted at least o subsidised sales,
generated by the recipients who stopped purchasing commercial inputs. loweer, this number could
be signiicantly higher i the remaining 43 commercial customers also purchased less than beore.
Q=?=X= P5$'"50-
1he supply and distribution o the subsidised inputs are relatiely centralised. Priate sector
inolement is sought through a tendering procedure, but in practise priate sector inolement is
limited. Compound D, one type o ertilizer, is usually supplied by a state-owned company ,Nitrogen
Chemicals o Zambia,. Contracts or urea, another type o ertilizer, are typically awarded the same two
large priate irms ,Omnia Small Scale Limited and Nyiombo Inestments Limited,.
1his apparent lack o competition in ertilizer supply is reportedly due to a general limited capacity o
the priate sector. loweer, when potential competitors to the 2-3 dominating irms are eectiely
excluded rom participating in the programme, there is little chance that they may deelop this capacity.
1his represents a undamental dilemma in input subsidy programmes between the need to ensure
smallholder access to subsidised inputs and the objecties o deeloping a more competitie priate
input supply sector to promote long term sustainability.
1he seed market appears somewhat more competitie, with seeral irms supplying dierent seed
arieties ,seen irms in 2008,. Contracts or distributing and storing the inputs until armers pick up
their packages are awarded smaller priate distributors and warehouse owners. 1he actiities are,
howeer, still centrally managed and there does not seem to be any trading and marketing o subsidised
inputs by priate dealers themseles.
One o the main diiculties related to deliery o subsidised inputs to armers is serious delays in
arrial. According to ZlSP guidelines, inputs should be aailable or retrieal by armers by end o
October so the inputs can be applied by the beginning o the agricultural season in Noember. A
surey o beneiciaries indicates that in 2008 less than 4 o subsidised inputs was distributed by end
October and 69 o recipients reported that they did not get their inputs until ater the start o the
rains.
1he timing problems are related to goernment budgeting procedures and programme administration.
1he iscal year in Zambia runs rom 1 April to 31 March. As budget allocations to the ZlSP hae to be
negotiated each year, stakeholders do not know how many subsidised input packages can be distributed
until the budget is approed by parliament in March. Ater inal approal, the tender or supply and
distribution o the inputs can be prepared. In 2008, the tender procedure ran throughout the summer
and the winners were announced in August, a ew months beore the inputs were to be deliered. It
should be noted that rom 2010 the goernment was to change the iscal year to match the calendar
year, giing ZlSP administrators three more months to prepare actiities. \e hae no inormation on
whether this change has helped oercome the timing issues.
19

Lidence o misuse is ery limited. 1here are some indications that input packs went missing in the
distribution process. 1he \orld Bank ,2010, notes an aerage 20 discrepancy between the number o
packages the district oicials say their released and cooperaties report to hae receied. It is not
known what became o the missing packs, and some o it may simply be due to accounting errors. 1he
large majority o armers sureyed reported satisaction with the quality o inputs they receied.
Q=?=Y= !F'& 3&0#&5/-
Originally, the programme was designed to run or three years, 2002,3 - 2004,5. Lach o the 125,000
beneiciary households was expected to graduate` rom the programme ater two consecutie years.
In the irst year, households were to beneit rom the ull subsidy and in the second year, the subsidy
would be haled. In practise, howeer, the exit strategy appears to hae had little eect. 1he
programme has been extended continuously, and the graduation mechanism is not working as planned.
1he responsibility or implementing the graduation mechanism rested with the armer cooperaties,
who identiied beneiciaries and prepared lists o selected armers, as well as with the District
Agricultural Committees ,DACs,, who approed the lists. loweer, in a surey o cooperatie leaders,
only 5 o them reported that preious support rom the lSP was taken into account when selecting
recipients - the most cited criteria or support was membership o the cooperatie and ability to pay
or the subsidised inputs. Also, the DACs do not appear to hae checked the lists or graduating
armers, let alone enorced the rule.
Q=?=Z= .33533C5)&
1he ZlSP in Zambia is plagued by many o the same diiculties as the AISP in Malawi. 1he greater
use o agricultural input appears to hae substantial eects on maize production, but the extra output
comes at a ery high cost. 1he best estimates aailable suggest that it would be cheaper to import maize
or consumption in urban areas than to increase production within Zambia through the ZlSP in its
present orm. Input subsidies could still be justiied i the apparent ineiciencies were outweighed by
equity considerations or a long term sustainable deelopment o the input sector. loweer, such
eects are doubtul as discussed below.
1he most important barriers to greater eiciency o the ZlSP appear to be inappropriate targeting o
beneiciaries and ineicient input deliery. 1he \orld Bank ,2010, suggest that a large part o the
beneiciaries already had reasonable access to agricultural inputs on commercial terms and around hal
o sureyed households purchased some inputs prior to beneiting rom the ZlSP. 1his indicates a
considerable risk that the subsidised input can displace input sales on commercial terms reducing the
oerall increase in input use and agricultural productiity. lurther, despite eorts to inole the priate
sector in supplying and distributing the inputs through public tenders, the input deliery system appears
highly non-competitie. \ear ater year ertilizer supply contracts are awarded the same 2-3 large irms,
one o them a parastatal, indicating general lack o competition. Distribution and storage o the inputs
is tendered out to priate transporters and warehouses, but it is still centrally managed and not subject
to much competition. In essence, the ZlSP creates a parallel state-controlled input marketing system,
which competes with ,on a subsidised basis, rather than utilizes the priate market.
It is striking that although the ZlSP objecties target smallholders in general, they do not emphasise
particularly the poorest and most ulnerable households. It appears that the programme was designed
to target the less-poor smallholders, possibly based on a presumption that the poorest households may
lack the capacity to ully utilise the subsidised inputs. 1he eidence we hae suggests that this targeting
objectie is largely met, which also means that the programme ails to beneit the poorest and most
ulnerable households directly.
20

It is ery doubtul that the increased use o agricultural inputs and higher agricultural productiity
achieed by the ZlSP is sustainable in the long run. Unlike the AISP in Malawi, the ZlSP does contain
an exit strategy but it is largely ignored. Programme actiities are not designed to address the underlying
causes o the low agricultural input intensity, and in some cases they may actually make things worse.
Major barriers to input adoption appear to be the lack o access to inputs in remote areas and, where
they are aailable, high prices partly due to imperect competition. Ideally, input subsidies will stimulate
demand and induce more irms to establish shops in remote areas, thereby increase access to inputs and
competition in the market. loweer, the lSP only stimulates demand or subsidised inputs, which are
not supplied by the market. In act, the programme risks reducing market demand through
displacement. 1he \orld Bank ,2010, ound that o beneiciaries, who purchased inputs rom the
priate market beore receiing subsidised inputs, stopped doing this ater enrolling in the programme.
I other recipients, who continued to buy commercial inputs, reduced their priate sector demand,
displacement could be higher. Unortunately, we do not hae more precise estimates o displacement.
Although the ZlSP encourages some priate sector inolement through open tenders, the contracts
are neer awarded new or potential entrants with the justiication that such new players lack the
capacity to ensure timely deliery o inputs. loweer, i the programme only inoles the ew priate
irms that already hae the capacity to delier the inputs, it is ery unlikely that it will contribute to the
deelopment o a more competitie priate input supply sector.
In conclusion, the ZlSP has a signiicant eect on ood production, but costs are too high. 1he main
beneiciaries are not the poorest households - in act the programme appears to be designed to
speciically target the less-poor. And long term sustainability is threatened by a state-drien and non-
competitie deliery system.
Q=G= \7#)#S H50&'$'J50 +%43',- T0(/0#CC5 V\H+TW
Y

Q=G=<= N#1E/0(%),
1he ertilizer subsidies implemented in Ghana in 2008, later extended to 2009, were ery dierent rom
the programmes implemented in Malawi and Zambia. It was neer meant to be a comprehensie
programme aimed at achieing a sustainable increase in smallholder adoption o agricultural inputs.
Rather, it was designed in great haste as an emergency measure to mitigate the aderse impacts o the
extremely high ertilizer prices.
1he ertilizer market in Ghana was one o the most liberalised in SSA prior to 2008 with irtually no
goernment interention. 1he termination o uniersal subsidy programmes through the 80s and 90s
coincided with a decline in ertilizer intensity rom 22 kg,ha in 198 to 8 kg,ha in 2006 ,\awson et al
2010,. As ertilizer prices grew rapidly through 200 and 2008, the goernment eared that ertilizer use
would decline een urther ,by an estimated 0,, reducing agricultural productiity and ood
production ,by potentially 20,, necessitating imports o ood crops, the prices o which also reached
an all-time high during this period. In March 2008, the goernment began discussing with leading
importers the possibility o subsidising ertilizer to counter the rising prices, and in May it announced
its intention to do so. loweer, it was not until early July that details o the programme were
published. On 4 July 2008, the subsidy took eect, too late to beneit the major season in southern
areas and the plantings in the northern regions, and barely in time or the second ertilizer application
in the north and the minor season in the south.

6
1his section is based on Banul ,2009, unless otherwise noted.
21

It is part o the story that 2008 was an election year in Ghana. 1he incumbent New Patriotic Party
,NPP, was seen as a party o the urban elite and stood in the polls to loose Noember`s election to the
National Democratic Congress ,NDC,, which garnered most o its support rom the rural population.
Banul ,2010b, interprets the input subsidy programme as an attempt to show armers that the
goernment had empathy or the rural population. In the eent, NPP lost the election by less than
0.01 o the otes. NDC took oer the preious goernment`s subsidy policies and continued the
programme or another year.
Q=G=?= O%&1(C5
It is not possible to estimate the outcome o the GlSP to any reasonable degree o conidence. 1he
programme was designed to aoid a decline in ertilizer use and agricultural production by keeping
ertilizer prices constant, so we cannot assess the outcome by obsering increases in output. \awson et
al ,2010, report that the all in ood output aoided could be as much as 20, but such estimates are
highly uncertain. Len i we take this estimate as an indication, it is not possible to calculate the alue
o the programme as we do not know exactly which crops are the main beneiciaries o the subsidised
ertilizers. Sureys suggest that the ertilizers are applied to a wide ariety o crops, including maize,
rice, egetables ,tomatoes, cabbage, etc.,, ruits ,oranges,, and oil palm.
Data rom laostat show that total production o maize and rice increased substantially in 2008 and
2009, by respectiely 21 and 10 ,maize,, and 58 and 30 ,rice,. 1hese numbers should not be
taken as outcome estimates, as a wide ariety o actors unrelated to ertilizer subsidies ,e.g. weather,
may hae aected production. loweer, they may sere as an indication that a massie decline in
output due to higher ertilizer prices has been aoided. 1o what extent this may be attributed to the
GlSP cannot be determined.
Q=G=G= +1(25
Compared to the input subsidy programmes implemented in Malawi and Zambia, the subsidies oered
in Ghana constitute a ery small scale programme. 1he total costs o the programme is estimated at
around USD 14 million in 2008 and USD 26 million was set aside to coer the costs o an extension o
the programme in 2009 ,\awson et al 2010,, corresponding to roughly 0.05 and 0.1 o GDP
respectiely.
1he size o the subsidy was set to keep the prices aced by armers roughly the same as in 200, around
GlS 14 - GlS 28 ,approximately the same amount in USD, per 50 kg bag o ertilizer, depending on
the type o ertilizer and geographical location. 1his subsidy corresponded to around 50 o the total
remunerations receied by importers. Initially, the goernment planned to issue 600,000 ouchers in
2008, each redeemable or the speciied rebate on one 50 kg bag o ertilizer, but in the end more than
1.1 million ouchers were printed, although less than 50 o those were eentually redeemed. 1he
reasons or the oerrun o the number o ouchers and subsequent low redemption rate are not
entirely clear. loweer, it appears that lack o clear criteria or the distribution o ouchers and general
uncertainty about how many ouchers were aailable in each district generated an initial shortage o
ouchers during the critical late summer months where ertilizers are most eectiely applied.

1his
prompted the goernment to issue more ouchers, and the subsequent glut caused shortage o
ertilizers. 1here were also reports o hoarding, in some cases een by armers who had no intention or

Second ertilizer application in the northern regions and ertilizers or the second ,minor, season in the Southern areas.
Other peaks or ertilizer application in Ghana are during the spring and early summer, but at that point the subsidy was not
yet in eect.
22

capacity to utilise the ouchers. \e do not hae detailed inormation on how many ouchers were
distributed in 2009.
Q=G=Q= 9#0/5&')/
1he ouchers were allocated to District Agricultural Directors, who passed on the ouchers to
extension oicers or inal distribution among armers. 1he distribution o ouchers between districts
was ormally based on ague notions o armers` need`. loweer, Banul ,2010b, argues that the
actual regional allocation o ouchers was more closely correlated with political actors than eiciency
or equity considerations. Speciically, he shows that districts, which the incumbent party lost in the
preious election in 2004, receied more ouchers than districts it won. lurther, the number o
ouchers allocated to a district increases with the ote margin, with which the district was lost. Banul
,2010b, interprets this result as attempts o ote-buying` by the goernment. 1he weaker the
goernment is in a district, the more it is aoured by the subsidy programme.
At the sub-district leel, there were no centrally stipulated criteria or who were eligible or receiing
ouchers, so any guidance was mainly proided by the district directors. Banul ,2010b, does not hae
any results on illage-leel allocation, but he argues that the most important politics is conducted at the
district leel anyway. As a result o the limited guidance, targeting aried greatly rom district to district
in terms o the number o ouchers each household could hae ,rom 2 to 10 or more, and any
characteristics o beneiciaries. Generally, it seems that most extension oicers handed out ouchers to
armers on a irst-come irst-sered basis.
Q=G=X= P5$'"50-
1he GlSP stands out as one o the most liberal ertiliser support programmes in SSA by extensiely
utilising the existing priate sector or input supply, distribution and retailing. 1he ouchers could be
used as partial payment or ertilizers at any retailer, who would accept them. 1he retailers would then
pass the ouchers up through alue chain to ertilizer importers, who would ultimately redeem
ouchers with the goernment.
Such a market-oriented oucher system is potentially ery eicient, as it allows armers to choose reely
between dierent suppliers spurring competition among existing businesses as well as new entrants. It
reduces the need or the goernment to set up and manage parallel distribution systems and select
suppliers and distributors through lengthy tender procedures. Also, there is less risk o hurting the
existing priate input supply sector through displacement o commercial input supplies, as these priate
suppliers are ree to engage in the subsidised input trade.
loweer, in practise a number o actors sered to diminish these beneits. Most importantly, only
ertilizer importers were able to redeem ouchers or cash with the goernment. In eect, a relatiely
small number o importers acted as gatekeepers controlling the low o ertilizers rom the world
market to Ghana as well as the low o subsidies rom the goernment to armers, which bestowed
considerable market power on importers. A consequence o this bias was that less than 40 o all
retailers accepted ouchers rom armers, primarily because they were unable to redeem ouchers with
their own suppliers or because it was too expensie or too diicult to do so ,Krausoa and Banul,
2010,. Most retailers, particularly the smallest ones in the most remote areas, hae no direct relationship
with importers. 1hereore, oucher would oten hae to trael through numerous steps in the supply
chain, each step subtracting a margin rom the oucher alue, beore reaching the goernment or inal
settlement ,Banul, 2010a,. Around hal the retailers, who sold subsidised ertilizer, submitted their
ouchers to another agricultural input dealer ,Krausoa and Banul, 2010,.
23

1here were other actors hampering programme eectieness. Due to the late launch o the
programme, the ertilizers arried late in the season signiicantly reducing their eectieness. Since the
programme was announced in adance, many armers had postponed input purchases in anticipation o
subsidies. 1he distribution o ouchers was characterised by conusion and general lack o inormation
regarding e.g. how many ouchers were distributed, how to get the ouchers and which retailers
accepted them. 1he conusion was exacerbated by the act that ouchers were speciic to ertilizer type
so a NPK 15:15:15 oucher could not be used to purchase urea. 1his led to a shortage o ouchers or
preerred ertilizer types and many armers had to settle or ertilizer they had little experience with. As
a result, a large majority ,92, o households sureyed by \awson et al ,2010, were dissatisied with
the accessibility o subsidised ertilizer.
Q=G=Y= !F'& 3&0#&5/-
1he programme was designed and announced as an emergency response to high ertilizer prices in
2008 and thereore intended to expire by the end o that year. In the eent, the subsidies were extended
into 2009 as well, een though the crisis had subsided by then. \e hae ound no indications that the
programme has been extended urther.
Q=G=Z= .33533C5)&
It was neer the intention o the GlSP to achiee a long term increase in ertiliser adoption by the
poorest smallholders, so it is perhaps unair to judge the programme on the basis o eiciency, equity
and sustainability related to such objecties. It is, howeer, interesting to ealuate some o the main
eatures o the programme by these criteria to proide insights into the perormance o dierent design
choices.
1he background and design o the programme suggest that ertilizer importers hae had a large
inluence on the goernment`s subsidy policies. 1he idea o an input subsidy was irst proposed by the
largest ertilizer importer in September 200, and the programme was designed through a series o
discussions between the goernment and the largest importers. 1here is little doubt that importers are
some o the main beneiciaries o the programme. Banul ,2009, reports that the total prices ,subsidy -
armer contribution, o ertilizer during the 2008 programme period ,July - December, as negotiated
by the goernment and importers were in most cases signiicantly higher than the market prices
preailing immediately prior to the launch o the programme ,June 2008,. lor instance, in the Ashanti
region NPK 15:15:15 and urea sold or GlS 35 in June whereas the negotiated prices or July-
December were respectiely GlS 50.50 and 51.50 ,the extra margins on sulphate o ammonia were
generally much lower,. Not only did importers aoid a large reduction in sales, they also appear to hae
extracted a higher margin. By restricting inal settlement o ouchers to relatiely ew large importers,
the programme adds to the market power o importers and diminishes the eiciency gains achieed by
implementing a highly market-oriented approach.
A major beneit o a oucher system is that the goernment can potentially target the most ulnerable
and poorest smallholders and thereby promote a more equitable distribution o productie resources.
Such possibilities hae largely been ignored in this case, although a relatiely wide geographical
distribution was achieed by making ouchers region-speciic thereby orcing suppliers to sere
remoter and more costly areas. 1he choice o making extension oicers responsible or inal
distribution o ouchers is praised by \awson et al ,2010,, as a smart and innoatie approach.
Lxtension oicers are supposed to be in close contact with armers and know them reasonably well
and they can complement ertilizer ouchers with knowledge and training. Also, they may be less
inclined to using ouchers as an instrument or political patronage than, say, illage chies because they
cut across tribes or ethnic groups, regions, religions and political persuasions.
24

1he long-term sustainability is less o a actor in this case, due to the programme`s short term and
narrow objecties. 1he subsidies are unlikely to hae signiicantly increased ertilizer adoption relatie
to pre-GlSP periods, as the main eect was to keep prices constant. It is possible that the programme
aerted long term negatie eects. I the ertilizers had not been subsidised, a large decline in demand
could hae caused struggling input suppliers to exit the market resulting in less competition and thinner
geographical market coerage. It is, howeer, questionable whether the programme saed many
suppliers as it mainly beneited importers and large ,better connected, retailers, which are likely to be
better equipped to withstand temporary shortalls in demand. 1he long term outcomes o the
programme are thereore expected to be limited.
In conclusion, the outcome o the GlSP is uncertain, and any eect is likely to be temporary. 1here
was no attempt at targeting the poorest households, and particularly large ertilizer importers appear to
hae beneited greatly rom the programme.
Q=Q= 9#)J#)'#S ]#&'()#$ ./0'1%$&%0#$ >)2%& ^(%1750 +175C5 V].>^+W
_

Q=Q=<= N#1E/0(%),
1he National Agricultural Input Voucher Scheme in 1anzania ,NAIVS, is a ery new and still ongoing
programme launched in 2009. 1hereore, little eidence is presently aailable on the general
perormance o programme. \e chose to include the case anyway, as a recent study ,Pan and
Christiaensen, 2011, on a pilot subsidy programme undertaken in 2008 oers detailed insights into the
targeting perormance o the oucher mechanism employed in the NAIVS. 1he ollowing will
thereore ocus mostly on experiences gained rom the pilot programme, but we will also comment on
some o the eatures o the expanded NAIVS.
1he pilot subsidy programme initiated in 2008 and later expanded into the NAIVS in 2009 was
launched by the Goernment o 1anzania in response to the high ood and ertilizer prices preailing in
200-2008. Agricultural input intensity is ery low in 1anzania, armers use on aerage 8 kg,ha o
ertilizers ,below SSA aerage,, and only 5. o rice armers and 0. o maize armers use improed
seed arieties together with ertilizers. Agricultural productiity is low by international standards and
relatie 1anzania`s own potential as measure by research ield tests and on-arm trials ,\orld Bank,
2009,. 1he goernment thereore argued that the best way to improe national ood security in the ace
o high international ood prices was to promote the use o agricultural inputs to raise productiity.
Q=Q=?= O%&1(C5
It is still too early to assess the outcome o the NAIVS. \e hae not ound any attempts in the
literature to make een preliminary estimations.
Q=Q=G= +1(25
1he NAIVS is designed to reach a total o 2.5 million households ,around 45 o all smallholders in
1anzania, in 65 districts, and subsidies are speciically targeted at producers o two major ood staple
crops, maize and rice. Lach beneiciary household is entitled to an input package suited or the
cultiation o ' ha o maize or rice at a 50 subsidy.
9
1hus, the input package is larger than the ones
oered to Malawi armers, but smaller than the Zambian equialents. 1he subsidy is smaller than in

8
1his section is based on Pan and Christiaensen ,2011, and \orld Bank ,2009, unless otherwise stated.
9
Speciically, 1, one bag o urea, 2, one bag o Di-ammonium Phosphates or two bags o Minjungu Rock Phosphate with
nitrogen supplement, and 3, 10 kg o hybrid or OPV maize seeds or 16 kg o rice seeds.
25

both o these countries, and it appears to be ixed in percentage terms, implying that armer
contributions will ary in proportion to input prices. 1he programme is budgeted at USD 299 million
oer three years, o which USD 139 million is coered by the Goernment o 1anzania and the rest
inanced by the \orld Bank. 1his corresponds to around 0.4 o GDP in 2009.
Q=Q=Q= 9#0/5&')/
1he programme establishes a chain o Voucher Committees at the Regional, District and Village leels
to oersee allocation and distribution o ouchers. 1he ouchers are allocated speciically to high-
potential maize and rice production regions and to areas where rice armers hae access to irrigation.
1he Village Voucher Committees ,VVC, are six-member bodies ,three women and three men, elected
by illage assemblies. 1he VVCs are responsible or selecting beneiciaries and prepare lists o
recipients or approal by the illage assemblies.
1he pilot programme in 2008 stipulated ery ew, broad targeting criteria. VVCs were adised to target
literate arming households willing to use the input ouchers or the purported crops and able to meet
the armers` co-inancing. Such criteria mainly pointed towards the less poor households and let
signiicant targeting powers at the discretion o the VVCs.
1he study by Pan and Christiaensen ,2011, sets out to estimate the targeting perormance o the 2008
pilot programme relatie to the programme`s two oerall objecties, 1, to increase oerall maize and
rice output, and 2, to increase access to modern inputs among poor and ulnerable smallholders. 1he
irst criterion is eectiely an eiciency criterion as they estimate to which extent the pilot programme
targets households, who are able to increase output the most. 1he second criterion relates to equality.
1he study is based on a household panel surey, which allows the authors to better control or actors
unrelated to the pilot programme.
1he study proides three striking results: lirstly, the estimated targeting perormance is ery close to
what would preail i ouchers were allocated randomly between intended ,poor or productie, and
unintended ,less-poor or less-productie, targets. 1argeting towards the most productie households is
slightly better than random, but targeting in aour o the poor is slightly worse. Secondly, the authors
ind clear eidence o what they reer to as elite capture. Speciically, they estimate that elected illage
oicials receie about 60 o the distributed ouchers, and other indicators o political connectedness,
such as access to 1V, radio and internet or participation in public meetings or armer`s associations,
also signiicantly increases the likelihood o receiing ouchers. As the political elite` tends to be less
poor, this bias goes a long way to explain the poor targeting perormance. 1hirdly, the targeting
perormance relatie to the poerty objectie tends to improe with the number o ouchers aailable
or distribution. 1his suggests that the bias in targeting resulting rom elite capture can be reduced by
ensuring a reasonable household coerage.
1hese results can be interpreted in the ollowing way. 1hat targeting on aerage is not too ar rom a
random allocation, and 60 o all ouchers go to the elite, could indicate that the remaining ouchers
were ery well targeted. Perhaps the political connections o the elite proide a priileged access to
ouchers. Once the demand o the elite is saturated`, the remaining ouchers go to the intended
targets. Such an interpretation is also consistent with the inding that targeting perormance increases
with household coerage. Passing a certain threshold, the elite has already taken its share and the ull
increment in ouchers is allocated to intended targets.
lollowing the experiences gained by the pilot programme, the expanded NAIVS programme document
deined the targeting criteria as:
26

1. lull time armers residing in the illage
2. larmers cultiate less than one hectare o maize or rice
3. larmers use the subsidised input o maize or rice production
4. larmers agree to sere as good examples in how to use good agricultural practices
5. larmers are willing and able to coer the co-inancing
6. lemale-headed households are gien priority
. larmers, who hae not used inputs in the past ie years, are gien priority
1hese criteria attempt to clariy the intended targeting, but inconsistencies still exist. lor instance, some
criteria ocus on the most ulnerable smallholders ,less than one ha o land and emale-headed,,
whereas others eectiely excludes the poorest households ,ability to coer co-inancing,. Also, seeral
criteria may be ery diicult to document ,no recent input use, or enorce ,sere as a good example,.
\hether the redeinition and clariication o criteria improes targeting perormance remains to be
seen.
Q=Q=X= P5$'"50-
Input supply, distribution and retail are largely undertaken by priate sector actors. larmers turn in
ouchers or a rebate at speciically certiied agro-dealers, who can redeem the ouchers directly with a
branch o the National Microinance Bank ,NMB,. Certiication is open to any agro-dealer, who
completes a business and management training programme organised by the Citizens Network or
loreign Aairs ,CNlA,, an NGO. 1he agro-dealers procure the inputs rom the open market, which is
reportedly rather competitie.
Due to the recent launch o the programme, we hae ound no eidence on the perormance o the
deliery system. Pan and Christiaensen ,2011,, who ocus exclusiely on targeting, do not discuss
deliery.
Q=Q=Y= !F'& 3&0#&5/-
1he programme plans or a duration o three years, and armers are eligible or a maximum o three
years o support. 1he beneiciaries o the pilot programme continue to receie ouchers under the
expanded NAIVS, and new entrants are expected in each year o the programme duration. 1hereore,
additional goernment support is needed or two years ater programme termination to complete the
three-year cycle o late entrants.
Q=Q=Z= .33533C5)&
1he experiences gained rom the pilot input subsidy programme in 1anzania highlights the potential
trade-o between objecties o raising national ood production on the one hand and beneiting the
poorest and most ulnerable households on the other - or in the economic terminology - between
eiciency and equity. 1he armers, who are most capable o translating an increase in agricultural input
use into expanded output, are not necessarily the poorest armers. Indeed, Pan and Christiaensen
,2010, suggest that targeting the poorest households has a lower impact on crop production than
beneiting the less poor.
1hese indings suggest that it may be necessary to decide whether the primary objectie o the input
subsidy programme is pro-poor growth or improed national agricultural production. 1he dual
objecties o the 1lSP risks worsening the targeting perormance and reducing the chance to meet any
o the objecties.
2

Under the NAIVS, locally elected VVCs are responsible or selecting beneiciaries, and the list o
recipients must be approed by the illage assemblies. 1his design choice is made to enhance
transparency and accountability. loweer, it also places signiicant power with the local political elite,
which may diert programme beneits away rom intended targets and thereby reduce programme
perormance in terms o both eiciency and equity. Pan and Christiaensen ,2010, ound that targeting
perormance may be improed by ensuring a relatiely high household coerage. loweer, such a
strategy implies a tacit acceptance o the act that some programme beneits will be captured by
unintended targets.
Due to the recent launch o the programme, we hae little eidence, on which we can ealuate
programme sustainability. loweer, a ew obserations on programme design can be made. O the
our input subsidy programmes inestigated in this ealuation study, the deliery system o NAIVS is
perhaps the one that is most in line with the market-based approach characterising an ideal smart
subsidy programme. It is designed to interere ery little in the existing priate market, and in contrast
with the GlSP in Ghana, it places the oucher settlement with an actor outside the input supply alue
chain ,the NMB, thereby dispersing some o the market power. 1his may be suicient to aoid
signiicant distortion o the priate market. On the other hand, it also presumes that a reasonably
competitie priate market already exists, and it is not clear how inputs may be deliered to more
remote areas, where the priate sector may be less deeloped. Apart rom the market oriented
approach, there is little to suggest that the programme to any large extent addresses the potential
underlying barriers to widespread input adoption, such as credit constraints, risk o crop ailure, etc. It
is thereore questionable whether any rise in the use o agricultural inputs will persist ater termination
o the programme.
In conclusion, the NAIVS in 1anzania represents an example o a highly market oriented input subsidy
programme. Signiicant resources are dierted to beneiting the less-poor with good political
connections. \e hae no inormation on the short term outcomes o the programme, and there is little
to suggest that outcomes will be sustainable in the long term.
X= ;533()3 ;5#0)5,
1he assessment o the our cases aboe illustrates the complexity o subsidising agricultural inputs and
highlights some o the challenges that must be addressed to ensure that a subsidy programme enhances
eiciency, equity and long term sustainability o agricultural input use. 1his section will summarise
some o the main lessons learned rom the our cases and where the most important gaps are in the
current state o knowledge.
X=<= N#1E/0(%),
\hat hae we learned
! Most o the programmes reiewed here were implemented in response to aderse shocks to
agricultural markets. 1he AISP in Malawi was launched ater a seere drought and poor
harests in 2004,5, and the programmes in Ghana and 1anzania were implemented in response
to high ertilizer prices in 2008. 1his suggests that the political motiation behind the
programmes was a desire to signal leadership and decisieness in response to an emergency.
Liciency, equity and sustainability may hae been secondary objecties at best.
28

X=?= O%&1(C5
\hat hae we learned
! Signiicant increases in agricultural productiity and ood production is possible, and the
potential or improing agricultural productiity by subsidising agricultural inputs exists.
loweer the estimates are somewhat uncertain. Costs are ery high, and gien uncertainties it is
unclear whether the programmes proide alue or money.
! 1here is ery little conincing eidence to suggest that outcomes are likely to persist ater
termination o the programmes. loweer, the subsidy programmes are designed to address the
distortions created by market imperections rather than the market imperections themseles.
\hen ,i, the programmes are phased out, input use is likely to decline again.
\hat do we need to know
! It is ery diicult to judge i input subsidy programmes are a good inestment gien the
inormation aailable at this point. In particular, more precise estimates are needed o impacts,
as well as opportunity costs in terms o returns on alternatie inestments oregone. Also,
estimates o dynamic ,long term, impacts are irtually non-existent, so there is little eidence
upon which to ealuate the long term sustainability o the programmes.
! More knowledge is needed on how to promote sustainability o a subsidy programme. A more
sustainable programme would ocus more on alleiating the market ailures, which distorts
agricultural markets, and less on mitigating the eects o the distortions.
X=G= +1(25
\hat hae we learned
! Subsidy design inoles trade-os between eiciency, equity and sustainability. I the subsidy
aries with input prices ,as in the programmes in Malawi and Zambia,, poor armers are partly
or completely shielded rom high prices, but economic returns rom the programme becomes
more ariable and likely negatie when input prices are high. Also, high input prices risk
inancial sustainability o the programme. On the other hand, i the subsidy is ixed in
percentage terms ,as in Ghana and 1anzania,, armers share a part o the burden o higher
input prices, which could exclude the poorest armers and cause disruptions in the unctioning
o input markets.
! 1he nature and scope o the subsidy is closely related to targeting perormance. Greater
household coerage improes targeting perormance, as demonstrated in 1anzania. 1he size o
the subsidy determines how easy it is or the poor to meet co-inancing requirements. Larger
input packages make the subsidy less releant or poor armers ,as exempliied in Zambia,.
linally, a relatiely large scale programme is needed to achiee a reasonable eect on national
ood security and ood prices.
\hat do we need to know
! More knowledge is needed on the consequences o input price olatility or programme
perormance and, particularly, long term sustainability. Does an input price spike, like the one
obsered in 200-2008 simply lead to a temporary decline in demand and agricultural
29

productiity, or does it hae more serious long term repercussions or smallholder adoption o
agricultural inputs, priate input markets, credit constraints, etc.
X=Q= 9#0/5&')/
\hat hae we learned
! None o the our programmes examined here succeed at targeting the poorest and most
ulnerable households. Indeed, some o the programmes did not intend to. A oucher scheme
has a potential or targeting subsidies at speciic groups o armers, poor and ulnerable
households, armers who do not already use inputs, most productie armers, etc. loweer,
realising the potential is ery diicult.
! 1he institutions with the responsibility or selecting beneiciaries hae considerable discretion
oer subsidy allocation, which generates the potential or aouritism, corruption and political
patronage. 1here is considerable scope or clariying and ocusing the targeting criteria in all
our programmes, but the question remains whether and how such targeting criteria can be
enorced.
\hat do we need to know
! low are targeting mechanisms best designed to aoid or minimise targeting bias Is a certain
degree o aouritism a necessary cost or a reasonably well targeted subsidy Is such a cost
acceptable
X=X= P5$'"50-
\hat hae we learned
! Smart subsidy programmes seek to promote the deelopment o a competitie priate input
distribution network to enhance eiciency and long term sustainability. \et some programmes
employ large existing irms, to the eectie exclusion o new smaller entrants, precisely
because the large irms already hae reasonably well deeloped distribution networks. 1he
desire to ensure eectie deliery o inputs to armers risks undermining the deelopment o a
competitie priate deliery system.
! 1here is a trade-o between eiciency and ,regional, equity in input deliery. Liciency can be
enhanced by utilizing existing priate input supply sector ,proided the priate sector is
reasonably competitie,, but this may come at the cost o regional coerage, by eectiely
excluding remote areas where markets are thin and priate agro-dealers largely absent. Is it
possible to complement the priate sector in dense areas with state-managed supply in remote
areas Or can a subsidy programme induce the priate sector to expand into remote areas
! Most o the programmes experienced serious delays and uncertainty in the deliery o inputs to
armers, mainly due to administratie challenges. Such delays may hae seere consequences
or programme eiciency and sustainability. Postponing application o inputs diminishes yields
directly, and uncertainties about input subsidy entitlements may cause armers to reduce
commercial purchases.
\hat do we need to know
30

! Apart rom the timing issues, we hae encountered little eidence on the actual perormance o
the input deliery system ,with the possible exception o Ghana,. lor instance how
competitie is the priate input supply sector ,what are the costs o market power,, how well
deeloped are the input markets in remote areas and how eicient are alternatie state-
controlled distribution channels
X=Y= !F'& +&0#&5/-
\hat hae we learned
! Lxit strategies are largely absent, non-credible or neer enorced in the Malawi and Zambia
programmes. In Ghana the exit strategy was implicit in the single-year scope ,expanded to two,
o the programme. In 1anzania, the exit strategy has not yet been tested. Lxit strategies are
based on the premise that positie eects o the programme persists ater programme
termination. 1here is ery little conincing eidence to suggest that this is likely to happen.
! Absence o a clear programme duration can lead to poor programme perormance by
increasing uncertainty about uture entitlements and preenting the continuous low o unding
and inputs. I programme continuation and scope has to be decided each iscal year, the
structure o the public budget-cycle may cause delays in the inal deliery o inputs ,as
demonstrated by the ZlSP in Zambia,.
\hat do we need to know
! More knowledge is needed on how to make an exit strategy eectie and credible. Subsidy
programmes are oten popular and there is considerable political pressure or their
continuation and expansion. low can such pressures be countered to improe chances o long
term sustainability
Y= :()1$%3'()
Based on the indings presented in this ealuation study, we conclude that agricultural input subsidies
are not likely to be an appropriate use o scarce iscal resources, i subsidy programmes are designed
and implemented in ways similar to the our programmes studied here. It is true that input subsidies
appear to raise agricultural productiity substantially. 1he programmes in Malawi and Zambia hae
shown impressie eects on national ood production. loweer, the so-called smart subsidy
programmes considered in this study are still plagued by many o the same problems associated with
the earlier generation o uniersal input subsidy programmes. Programme perormance could
potentially be improed by addressing the issues outlined below, but the challenges are huge.
lirst, the programmes are too costly and ineicient. 1hey ail to properly utilise the eiciencies oered
by the priate input markets by channelling resources through parastatal entities ,Malawi,, state-
managed distribution networks ,Zambia,, or large non-competitie corporations ,Malawi, Zambia and
Ghana,.
10
Other actors hae reduced the eectieness o input subsidies: Poor targeting hae resulted
in displacement o non-subsidised priate sector sales, limiting the total expansion in the use o
agricultural inputs. And administratie problems hae led to signiicant delays, generating uncertainty

10
Input deliery in 1anzania may be more competitie, but we do not yet hae suicient inormation on the deliery
perormance o the NAIVS.
31

about entitlements and postponing the application o inputs. Additionally, the sheer size o the
programmes, particularly in Malawi and Zambia, signiicantly limits the resources aailable or other
inestments and threatens the sustainability o goernment inances.
Second, most programme beneits accrue to less-poor and politically well-connected households as well
as large input suppliers. In some cases ,Malawi and 1anzania,, priority is oicially gien to the poorest
and most ulnerable households. In practise, they are largely excluded due to relatiely high co-
inancing requirements ,Ghana and 1anzania,, input package size ,Zambia,, and conlicting or ague
targeting criteria ,Malawi and 1anzania,. 1he institutions responsible or beneiciary selection ,local
councils, cooperaties and extension oicers, enjoy considerable discretion oer who will receie the
subsidised inputs, and this authority proides opportunities or rent seeking and political manipulation.
Large input suppliers are some o the main beneiciaries o the programmes as they ace guaranteed
demand at oten inlated prices, and as they in some cases are bestowed special priileges ,Malawi,
Zambia and Ghana,.
1hird, the subsidy programmes mainly attack the symptoms o low input use and poor agricultural
productiity rather than the underlying disease` o high input procurement costs and market ailures.
Long term sustainability o the programmes is thereore doubtul. It is possible that a truly market
oriented programme may strengthen the competitieness o the agricultural input markets, but i input
demand collapses ater termination o the programme such gains may disappear. 1he greater use o
agricultural inputs may persist i smallholders manage to accumulate enough assets to oercome credit
constraints and become less ulnerable to crop ailure. 1here are, howeer, no signs o this happening.
lourth, there is a substantial risk that an exit strategy will be ineectie ,as demonstrated by Malawi
and Zambia,. Once an input subsidy programme is launched, most stakeholders hae a great interest in
its continuation and expansion. Recipients beneit directly, the local elite may use subsidies as a tool to
reach political objecties, input suppliers enjoy a stable demand and possibly greater market power, and
national policy makers may point towards tangible results o their policies in terms o higher
agricultural output. Likely consequences are that the programme becomes more entrenched in the
political system and more subject to political manipulation and rent seeking.
It may be possible to improe eiciency, equity and sustainability o the input subsidy programmes, but
the challenges are huge. A oucher scheme is potentially an eectie instrument or disconnecting
input deliery rom subsidy targeting, thereby allowing the priate sector to handle input deliery while
the state organises targeting. 1he programmes in 1anzania and Ghana demonstrate that such a diision
o labour between the state and the priate sector is possible. But they also show that institutional
details, such as who has the power to redeem ouchers with the goernment, greatly aect programme
perormance. A more eicient or pro-poor targeting may be diicult to achiee. No matter how clearly
target criteria are speciied, they need to be enorced to be eectie. Most likely, those who identiy
beneiciaries will always hae some discretionary power which can be exploited. As input subsidies are
unlikely to alleiate market ailures directly, long term sustainability o programme impacts largely
depends on recipients` ability to accumulate inancial assets rom production surpluses generated by the
subsidies. \e hae seen no eidence o this happening.
32


Z= L58505)153
Banul, Aua Branoah ,2009, Operational Details o the 2008 lertilizer Subsidy in Ghana - Preliminary
Report. GSSP Background Paper 18, \ashington D.C., IlPRI.
Banul, Aua Branoah ,2010a, Market-Smart Lessons rom the 2008 and 2009 lertilizer Subsidy
Programs in Ghana. CP ^ote., tbe ver.tetter of tbe Cbava trateg, vpport Prograv, IlPRI.
Banul, Aua Branoah ,2010b, Old problems in the new solutions PR Di.cv..iov Paper 01002,
\ashington D.C., IlPRI.
Barratt, N., D. Chitundu, O. Doer, J. Llsinga, S. Lriksson, L. Guma, M. laggblade, S. laggblade, 1.
O. lenn, l. R. Locke, C. O`Donnell, C. Smith and 1. Steens ,2006,, Cassaa as drought
insurance: lood security implications o cassaa trials in Central Zambia, .gre/ov 45,1,, March
2006
Chibwana, Christopher, Monica lisher, Charles Jumbe, \illiam Masters and Gerald Shiely ,2010,,
Mea.vrivg tbe ivpact. of Matari`. farv ivpvt .vb.ia, prograv, Contributed paper presented at the
Arican Association o Agricultural Lconomists ,AAAL, conerence, Cape 1own, South Arica
2010
Chinsinga, Belssings ,2011,, Seeds and subsidies: 1he political economy o input programmes in
Malawi, D vttetiv 42,4,, Institute o Deelopment Studies
Craword, Lric \., 1.S. Jayne and Valerie A. Kelly ,2006, Alternatie Approaches or Promoting
lertilizer Use in Arica. .gricvttvre ava Rvrat Deretopvevt Di.cv..iov Paper 22, \ashington D.C.,
1he \orld Bank.
Dorward, Andrew ,2009, Rethinking Agricultural Input Subsidy Programmes in a Changing \orld.
London, School o Oriental and Arican Studies.
Dorward, Andrew and Lphraim Chirwa ,2011, 1he Malawi Agricultural Input Subsidy Programme:
2005-6 to 2008-9. vtervatiovat ]ovrvat of .gricvttvrat v.taivabitit, 9,1,.
Dorward, Andrew, Lphriam Chirwa, Duncan Boughton, Lric Craword, 1hom Jayne, Rachel Slater,
Valerie Kelly and Maxton 1soka ,2008, 1owards smart` subsidies in agriculture Lessons rom
recent experience in Malawi. ^atvrat Re.ovrce Per.pectire. 11. London, Oerseas Deelopment
Institute.
Dorward, Andrew, Lphriam Chirwa, and 1.S. Jayne ,2010, 1he Malawi Agricultural Input Subsidy
Programme 2005,6 to 2008,9 |Prepared or the \orld Bank Arican Success Stories[. Project
Report. \ashington D.C.: \orld Bank.
luture Agricultures ,2010, Raising agricultural productiity in Arica. Optiov. for actiov, ava tbe rote of
.vb.iaie.. Arica Progress Panel Policy Brie, luture Agricultures.
Goernment o Malawi ,2011,, vaget tatevevt 2011,12, 3
rd
June 2011
Krausoa, Marika and Aua Branoah Banul ,2010,, Oeriew o the agricultural input sector in Ghana,
PR Di.cv..iov Paper 01021, International lood Policy Research Institute, September 2010
Minde, Isaac, 1.S. Jayne, Lric Craword, Joshua Ariga and Jones Goereh ,2008, Promoting lertilizer
Use in Arica: Current Issues and Lmpirical Lidence rom Malawi, Zambia, and Kenya. Paper
33

preparea for tbe Regiovat trategic .gricvttvrat Kvorteage vpport ,.tev ;Re.K) for ovtberv .frica,
!or/ivg Paper ^o.1, lood Security Group, Michigan State Uniersity.
Pan, Lei and Luc Christianensen ,2011, \ho is Vouching or the Input Voucher Decentralized
1argeting and Llite Capture in 1anzania. !orta av/ Potic, Re.earcb !or/ivg Paper erie. 5651.
\ashington D.C., 1he \orld Bank. Aailable at SSRN: http:,,ssrn.com,abstract~183315.
Ricker-Gilbert, J. and Jayne, 1. S. ,2010,, !bat are tbe a,vavic effect. of fertitier .vb.iaie. ov bov.ebota rett
beivg. riaevce frov Matari, Contributed paper presented at the 3
rd
Arican Association o
Agricultural Lconomists ,AAAL, and 48
th
Agricultural Lconomists Association o South Arica
,ALASA, Conerence, Cape 1own, South Arica, September 19-23 2010
1iba, Zontal ,2009, 1he role o input subsidies: Operational guidelines on implementation. Poticie. for
Cooa covovic Mavagevevt of ooa Price rivg. iv .fricav Covvtrie.. Rome, lAO 1rade and Markets
Diision.
\iggins, Stee and Jonathan Brooks ,2010, 1he Use o Input Subsidies in Deeloping Countries. Paris,
OLCD.
\orld Bank ,2009,, .cceteratea fooa .ecvrit, prograv of tbe |vitea Repvbtic of 1avavia vvaer tbe gtobat fooa cri.i.
re.pov.e prograv, Program Document, May 2 2009
\orld Bank ,2010,, Zambia Impact Assessment o the lertilizer Support Program, Analysis o
Lectieness and Liciency. Report ^o.:11, \orld Bank, Arica Region.
\awson, Daid O., lrederick A. Armah, Lrnest K.A. Aria amd Samuel K.N. Dadzie ,2010, Ghana`s
lertilizer Subsidy Policy: Larly lield Lessons rom larmers in the Central Region. ]ovrvat of
v.taivabte Deretopvevt iv .frica 12,3,, pp.191-203.


Published by:
Ministry of Foreign Affairs of Denmark
Evaluation Department
Asiatisk plads 2
1448 Copenhagen K
Denmark
E-mail: eval@um.dk
The publication can be downloaded from:
www.evaluation.dk
ISBN: 978-87-7087-586-8
ISBN: 978-87-7087-587-5 (internet version)
Other Evaluation Studies
2007/1: Assistance to Development Research
2007/2: Synthesis of Evaluations on Technical Assistance
2007/3: Evaluating the Impact of Rural Roads in Nicaragua
2008/1: Synthesis of Evaluations of HIV/AIDS Assistance
2008/2: Synergies between Bilateral and Multilateral Activities
2008/3: Assessing Multilateral Organisation Effectiveness
2008/4: Economic Empowerment of Women
2008/5: Public Private Partnership Programme
2009/1: Evaluating Aid Effectiveness in the Aggregate:
Methodological Issues
2009/2: Addressing the Underlying and Basic Causes of
Child Undernutrition in Developing Countries:
What Works and Why?
2009/3: Experiences with Conducting Evaluations Jointly with
Partner Countries
2009/4: Synthesis of Impact Evaluations of Microcredit
2009/5: Synthesis of Evaluations on Support to Business
Development
2010/1: Evaluating Aid Effectiveness in the Aggregate:
A critical assessment of the evidence
2010/2: Gender and Value Chain Development
2011/1: CCPAs Open Fun Football Schools Programme
EVALUATION STUDY
2011/2
Agricultural input subsidies
in Sub-Saharan Africa