Lorebgmipllsumdolyorsltzmhhabc Land Governance POLICY BRIEF WORLD BANK ✦ DEVELOPMENT RESEARCH GROUP ISSUE 2 ✦ NOVEMBER 2017 Performance of large farms and benefits for smallholders: challenges and opportunities in Malawi’s estate sector ✳✳ Klaus Deininger and Fang Xia The agricultural estates in Malawi (1.5 million expansion of large farms since 2008, has land transfers is needed to avoid that results hectares), mostly held by domestic commer- raised many concerns, and a large body of are due to sample or case selection. Second, cial farmers, are underperforming. Data from literature now exist on its risks. time series information will be desirable three sources (digitized agricultural leases, for assessing how different policies affect a georeferenced farm survey, and satellite Malawi is of interest for measuring the outcomes. imagery) are used to analyze this. It shows performance of estates and effects for that estates are underachieving compared to smallholders, given the scale of large farm Traditionally, such data has come only from smallholders in terms of yield, productivity, and investments and the length of time for censuses or sample surveys. This paper, land use intensity. Estates also fail to generate which these have been in operation. The shows possibilities for using additional data positive spillovers for smallholders. Moreover, government leased about 1.5 million hect- sets and new analytical approaches. Three more than 70 percent of estates have expired ares to commercial farmers in the 1980s data sets are used: Georeferenced survey leases, so government is losing ground rents and early 1990s, representing 20% of the data, digital administrative data held by while it is also a source of tenure insecurity for country’s agricultural land. The goal was to land agencies, and satellite imagery based estate holders. In addition, overlapping claims help commercialize the agricultural sector, land use analysis. affect 140,000 hectares (9 percent of total area while overcoming limitations in regulatory under estates). Clarifying boundaries and regimes for customary tenure. Land registries administrative records con- lease status for estates should be the first step tain a wealth of information, allowing for toward systematic demarcation and prepara- The over 25 years of experience with routine monitoring of large investments’ tion of customary estates under the recently large farms in Malawi allows discerning performance, but is often locked up in enacted Land Acts, once gazetted. Failure to longer-term impacts and identifying chal- analog form. To enable the Malawi gov- follow this sequence could, however, add an- lenges not yet apparent in other African ernment to better manage its estate sector, other layer of unconnected rights, exacerbate countries where land transfers have hap- digitization of all leasehold contracts was tenure insecurity and discourage investment pened more recently. Such analysis is also supported by a World Bank project in 2016. in productive performance. highly policy relevant as it can directly feed into the policy debate on commercial agri- Machine learning algorithms using medium- culture and tenure security, and the review scale imagery at rather high frequency Measuring estate of the estate land policy. The analysis also is now available for free on cloud-based performance provides guidance for the 2016 enacted platforms. The methodology has shown new land laws, the implementation of to generate information on land use, and No consensus has yet emerged on how which still needs to be regulated. potentially even crop type or yield at field to analytically tackle impact assessment of level, as long as fields are of a minimum size. large-scale land based investment. If access These combined data enable us to trace to modern technology is limited or factor A novel analytical the evolution of Malawi’s estate sector and markets are imperfect, commercial farm methodology allow making inferences on the longer-term establishment may benefit neighboring impact of large farms. smallholders by improving their knowledge Most available studies on large farms of improved techniques and allowing easier assess the impact of specific investments. access to factor and output markets. The Although providing valuable insights on Evolution of Malawi’s conversion of customary land into lease- the dynamics of establishment and per- estate sector holds also generates public revenue in formance of these estates, it is difficult terms of ground rent. Beyond such poten- to determine the extent to which these Malawi has traditionally been characterized tial beneficial effects, the literature has long findings are representative of the estate by a dualistic land tenure structure geared pointed out that large farms may compete sector at large. Addressing this subject at towards cash crop production. In colonial with local smallholders for resources, most the sector level would require dealing with times, cultivation of tobacco, the country’s prominently land but also water. The rapid two issues. First, data on the universe of main cash and export crop, was restricted to white settlers who had preferential access and increased frequency of land transac- The 2006/07 National Census of Agriculture to land, labor, and credit, and guaranteed tions with outsiders, and causing disputes. and Livestock (NACAL) reported that about market access via a quota system. After Perceived tenure insecurity increased with a third of estates have tenants; the share of independence in 1964, this estate land was negative impact on farm productivity and estates with tenants peaks at close to 50% transferred to Malawians. output, especially by females. in the 10–100 size category. To boost commercial crop production, the The launch of a more comprehensive land NACAL also illustrates that most estates government introduced 21-year leases policy reform process in 1996, culminated (73%) are owned by Malawian persons; 11% for public land in the late 1980s. A large in 2016 in Parliamentary approval of a series are owned by “others”, most likely legal enti- number of estates, most sized from 10 to of Land Bills. One change is the introduc- ties, and 10% by expatriates. The ownership 30 hectares, were carved out of what was tion of the ‘customary estates’ defined as all share of expatriates and government peaks deemed unutilized customary land and land owned, held or occupied as private at 100–500 ha and that of ‘others’ in the > transferred to aspiring commercial farmers, land within a Traditional Land Management 500 ha group. covering almost 800,000 hectares in 1994. Area (TLMA). Only estates were allowed to produce and Textual data in the digitized administrative sell tobacco directly, while smallholders had records shows that in 2016, the estate trans- to sell their output to the marketing board Characteristics of the fers cover some 1.5 million ha, (1.35 and 0.14 at low prices. The surplus thus generated estate sector in agricultural and non-agricultural estates, by the government was funneled to estate respectively) in 58,733 leases (35,140 and owners in the form of soft loans, providing The 1997 Estate Lands Utilisation Study, 23,593 for agricultural and non-agricultural an implicit subsidy that reinforced the dual- or ELUS, is a key source of information land, respectively). istic structure of the country’s agriculture on the establishment of the sector. Most sector. estates are established by Malawians (table Using the original data of establishment, 1). Some 57% of estate land were newly Table 2 illustrates that from a basis of 16,725 Dissatisfaction with the results of such a cleared with the remainder used as cus- ha registered estates in the pre-indepen- strict distinction between estates and the tomary land before. In fact, a sizeable share dence period, transfers to the estate sector customary sector led to a moratorium on of new estate owners seems to have con- in the form of leaseholds, accelerated con- lease issuance in 1994. Tobacco quotas verted land they previously farmed under siderably after independence in three main were gradually extended to smallholders. customary tenure, either to be able to phases, differing in volume of leases issued, Rapid take-up led to marked improvements grow tobacco or to increase tenure security. and mean size. Following the 1994 mora- in socio-economic indicators and soon About half of owners were absentee, with torium, overall issuance of new agricultural brought small farmers’ share in tobacco a quarter indicating that they rarely visited leases dropped sharply, while issuance con- production to some 70%. their estates. Encroachment was an issue on tinued apace for non-agricultural estates. 52% of estates above 500 ha, but affected Yet these reforms did little to improve small- a much smaller share (5%) of estates below Agricultural estates measure 39.8 ha on holders’ security under customary tenure 20 ha. average and most agricultural estates are that historically allowed egalitarian land in the 10–30 ha group (73%), 8% (111,884 access and high levels of tenure security by ELUS reported that tenancy on estates was ha) and 0.7% (99,752 ha) of estate area is in community members, but over time, how- widespread, with some 72% reporting to estates larger than 50 or 500 ha, respectively ever, these systems came under increasing employ tenants who were estimated to (see Table 2). stress. Land scarcity became a growing con- account also for 52% of estates’ labor force. cern due to population growth, migration, Quality of land records Table 1: Agricultural Estate characteristics by size The formal process to obtain a lease com- Size category in ha prised four steps: An application stating size, All 5–10 10–50 50–100 100–500 >500 intended use, and location of the desired Estate ownership piece of land (“sketch map”), together with a ‘no objection’ document by the chief, certi- Years run by the current owner 18.99 12.54 15.28 21.13 19.84 30.77 fying that neither chief nor village headman Owner is Malawian (%) 72.58 82.76 92.42 80.21 50.52 29.75 object to the proposed transfer, had to be Owner is expatriate (%) 10.48 0.00 1.18 4.17 29.17 20.66 submitted. Having validated the appli- Owner is government (%) 2.19 0.00 0.00 3.13 5.21 4.96 cation, Government issued an offer that details the length of the lease, permitted Land and labor use intensity land use, assessed fees, and annual ground Area owned 433.86 8.52 21.61 74.28 272.36 2,544 rent, ideally accompanied by a survey plan Area operated (%) 15.4% 67.7% 47.5% 36.9% 29.4% 11.6% describing the property’s location more Total wage bill per ha (US$) 131.95 144.42 133.60 174.62 138.21 77.64 precisely. Acceptance transformed the offer letter into a preliminary lease contract that No. of obs. 860 29 422 96 192 121 would then be formalized by an officially Source: Own computation from 2006/07 NACAL. registered deed. As each step normally required side payments, the actual pro- of Agriculture and Livestock (NACAL), which Beyond such beneficial effects, large farms cess followed was often quite different or contains information for both smallholders may compete with local smallholders for remained incomplete. and estates. resources, most prominently land. Spatial proximity as a channel for transmission of Data suggest that only 36% of all leases Comparing smallholders to estates provides spillover effects between investors and (42% of agricultural ones) are supported by interesting insights in several respects. First, neighboring households used to investi- a deed; 34% (37% of agricultural ones) have for estates, 15% of allocated land is oper- gate economic and social impacts of mine only a letter of offer; and 30% (21% of agri- ated on average. This share decreases from openings or closings, including on female cultural ones) remained at the application 88% in the group below 5 ha, a figure that empowerment and, although more limited, stage. Quality of spatial documentation on is comparable to the intensity of land utili- to assess the impacts of large farm invest- the administrative records varies. While 2% zation by smallholders, to 12% in the above ment on neighboring small farmers. of leases for agricultural estates (and 18% 500 ha group. Prima facie this provides some for non-agricultural ones) are surveyed and support for claims about un- or underused While lack of panel data on smallholders accompanied by a deed plan, 52% (and estate land. Idle estate land has been a recur- in Malawi makes it impossible to identify 66% for non-agricultural ones) have not rent theme in Malawi’s policy debate. causal impacts, we can use simple regres- advanced beyond the sketch plan whereas sions as a descriptive device to assess for 46% (and 16% for non-agricultural ones) Second, production structure and cropping whether, after controlling for other factors, the sketch was redrawn by the survey patterns differ between smallholders and smallholders’ location on or distance to an department (see also table 3). estates: 42% of estate area is devoted to estate, with or without a valid lease, affects tobacco, followed by maize (39%), ground- their production outcomes. By showing that, for 7,819 agricultural nuts (7%), and other crops. The 2006 data estates with a total area of 404,584 ha, also suggest that for all crops except cas- Results of our analysis suggest that, mainly documents lack data on lease duration, dig- sava, smallholders’ yields are significantly because of a larger area cultivated, location itization of all records also points towards above those by estates. While these are on or proximity to an agricultural estate is gaps in quality of existing documenta- yields rather than profits, the share of estates associated with higher levels of output. This tion that may create challenges for good using purchased inputs and the mean per does, however, not translate into higher governance. hectare value of such inputs by those who levels of productivity. In fact, for “squatters” use them is significantly above the equiva- on agricultural estates, output and profit Leases’ spatial components also point lent figure for smallholders. This suggests per hectare are negative and significant. towards significant overlaps, a notion that is that the relationship between farm size and Per-hectare profits are higher only for small- indeed supported by categorization of land profits is unlikely to be positive. holders in closer proximity to the boundary use based on overlays with medium reso- of estates with non-expired leases. While lution imagery: 28% of agricultural estates further exploration of this issue with better have at least 20% of their area overlap with Benefits for smallholders data would be warranted, this suggests that another estate, an issue affecting a total of any indirect benefits from estates for small- 137,064 ha. Commercial farms establishment has the holders will be quite limited. potential to generate positive effects for smallholders, such as access to technology In contrast to other countries where large Productivity of estates and markets for credit, input, labor, and farms produce bulk commodities and compared to smallholders output where access to modern technology often generate little employment, many is limited or factor markets are imperfect. of Malawi’s estates are labor intensive (see Overlaying administrative data on estate An additional source of positive spillovers table 3). Permanent or temporary male boundaries with satellite imagery allows is through employment on estates. Small (female) labor is hired by 64% (27%) and for a rough assessment of land under farmers who work on estates as casual 70% (56%) of estates respectively. Demand crops. Information on production and workers may also acquire knowledge about for permanent labor per ha increases with yields requires survey-based information. new techniques or pick up specific skills size to about 0.9 males and 0.6 females in We use again the 2006/07 National Census that will be useful on their own farms. the largest category though the pattern for temporary labor is more volatile. Table 2: Evolution of number and area under agric. and non-agric. estate leases Ground rent collection Agric estates Non Agric estates Leases/year Area/year Total area Leases/year Area/year Total area Government potential for collecting transferred transferred ground rent is eroded over time as leases No. 1000 ha No. 1000 ha have expired and were not renewed. In 1965–86 104 10.79 254050 107 0.18 5080 2016, leases for 74% of agricultural estates 1987–94 2626 64.85 772850 124 0.14 6210 had expired. In fact, only 5% of agricultural 1995–2006 230 6.71 853340 267 0.42 11290 estates (vs. 41% of non-agricultural ones) had remaining lease terms beyond 10 2007–16 112 9.08 944180 532 0.46 15890 years and thus a time horizon long enough Source: Own computation from the National Geographical Estates Database. to make longer-term investments. This situation may reduce estate productivity Conclusion and policy near real time. Policy decisions on proce- directly, as well as limiting the scope for implications dures for lease renewal, particularly setting efficiency-enhancing transfers of land to levels of ground rent that are realistic, and operators with higher levels of ability. Our paper contributes to the debate on developing procedures on how to deal how to analytically tackle impact assess- with accumulated ground rent arrears, are The mean annual rent of less than US$ 1/ ment of large-scale land based investment one precondition for such a process to be ha for agricultural estates and US$ 27/ha for from a methodological and a substantive feasible. non-agricultural ones, is implying that yield point of view. may be below the cost of collection. Low A second precondition is that a clear hier- ground rents may also encourage specula- The descriptive statistics presented are archy of evidence among competing tive instead of productive land use. derived from the digitized leasehold data claims be established, and procedures to base, showing that computerization of ascertain and adjudicate rights in a way To illustrate the potential revenue from administrative data locked up on paper, that deals with discrepancies or overlaps agricultural leases, we note that, the mean enables their use for policy analysis and through administrative mechanisms be market price of an existing lease is US$ 58/ action. Methodologically, we show how developed. The recent enactment of a set ha. Even a compliance rate of 50% could combining administrative records with of Land Acts creates enormous opportuni- generate annual lease revenue of some US$ georeferenced survey data and remotely ties to link clarification of estate leases to 35 million, or 5% of Malawi’s total public sensed imagery can help address many of securing smallholder land rights. However, spending, in addition to providing strong the issues that traditionally impeded routine not respecting this precondition holds a risk incentives for effective land use. monitoring of large agricultural investments’ of unintended consequences. performance, and that could then trigger Table 3: Descriptive statistics of swift action in case of deviations. Using sporadic adjudication processes to a set of land records that already contains estates by lease status Substantively, using 2006/07 data we find numerous overlaps, will add yet another that Malawian estates failed to live up to layer of complexity, exacerbate tenure inse- Agric. Non-agric. Total estates estates their potential. With few exceptions, yields curity, and increase the level of disputes. To were below and input use above small- avoid this, it will be essential to allow spo- General characteristics holders’, who also seem to have derived radic adjudication processes only in areas Total area (1,000 ha) 1,487.44 1,348.76 138.68 few spillover benefits, either in terms of where TLMAs have been determined and Mean area (ha) 27.10 39.80 6.60 technology or market access. estate leases have been clarified in a process followed by adjudication and registration of Lease length <=21 years (%) 47.88 73.62 9.55 Detailed investigation of land records customary land ownership in an integrated Size less than 10 ha (%) 42.60 8.37 97.83 suggests that difficulties for government and systematic process. Size 10–30 ha (%) 45.62 73.22 1.09 to collect ground rent revenue are ema- Size 30–50 ha (%) 5.90 9.40 0.25 nating from expired lease records, while a Future research on estate sector perfor- substantial program for lease renewal, and mance and spillover effects, building on Size 50–100 ha (%) 3.21 5.04 0.26 improving quality of land records is not in more recent data, would be desirable to Size 100–500 ha (%) 2.12 3.24 0.33 place. Tenure insecurity may reduce inten- provide important input into the design Size above 500 ha (%) 0.55 0.74 0.25 sity and productivity of land use. A failure and possibly evaluation of a forward- Formal documentation to maintain administrative records not only looking program of tenure regularization to undermines generation of public revenue, improve the productivity and resilience of Has deed (%) 35.80 42.03 26.51 but may also give rise to a collective action Malawi’s agricultural sector and, given this Has offer but no deed (%) 34.47 37.09 30.56 problem whereby each private lessee will sector’s importance, the country’s overall Sketch plan (%) 56.76 51.65 65.81 not have an incentive to keep records up to economic performance. date although society would greatly benefit SD plan (%) 35.53 46.47 16.14 from it. Deed plan (%) 7.71 1.87 18.05 Annual rent (US$/ha) 10.69 0.79 26.66 To unleash the potential of properly run Lease term in 2016 estates to contribute to diversification of Malawi’s agricultural sector, there is need Lease expired (%) 45.35 69.65 9.16 to renew, cancel, or renegotiate existing No. of obs. 58,733 35,140 23,593 estate leases using a systematic process Source: Own computation from the National Geographical Estates that could then form the basis for contin- Database. ued monitoring of lease performance in This policy brief is based on Deininger, Klaus and Xia, Fang. 2017. Assessing Effects of Large-Scale Land Transfers: Challenges and Opportunities in Malawi’s Estate Sector. Policy Research Working Paper; No. 8200. World Bank, Washington, DC. http://bit.ly/2xj90GI Klaus Deininger is a lead economist at the World Bank, Fang Xia is based at the University of International Business and Economics in Being. ✦  CONTACT: kdeininger@worldbank.org. The views presented are those of the authors and do not necessarily represent those of the World Bank, its Executive Directors or the member countries they represent.