Jump to ContentJump to Main Navigation
African Economic DevelopmentEvidence, Theory, Policy$

Christopher Cramer, John Sender, and Arkebe Oqubay

Print publication date: 2020

Print ISBN-13: 9780198832331

Published to Oxford Scholarship Online: July 2020

DOI: 10.1093/oso/9780198832331.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use. date: 14 June 2021

Technical Change and Agricultural Productivity

Technical Change and Agricultural Productivity

(p.219) 9 Technical Change and Agricultural Productivity
African Economic Development

Christopher Cramer

John Sender

Arkebe Oqubay

Oxford University Press

Abstract and Keywords

The evidence does not support gloomy generalizations about an irreversible African environmental crisis or pessimistic arguments that barriers to adopting Green Revolution technologies are insuperable. Although evidence on agricultural technology in Africa is often unreliable, food output and grain yields do appear to have risen strongly in some African economies.. Huge variations in crop yields, including within similar agro-ecological zones, suggest massive potential for policies to promote a rapid increase in yields. Agricultural research and development (R&D) within African countries—and production on many large-scale farms—has shown that dramatically higher yields are possible. Crop yield improvements—with the aid of suitable high-yield varieties (HYVs), public agricultural research spending, and especially investment in irrigation—are possible without draconian resettlement schemes, without wasteful extension service spending, and without recourse to micro-finance schemes. The methods underpinning commonly produced estimates of yields are unreliable, calling into question conventional wisdom that small farms are more efficient than larger farms.

Keywords:   small farms, crop yields, irrigation, R&D, micro-finance, extension

9.1 Introduction: From the Missionary Position to Modern Charitable Fantasies

Rural development policy in Africa has been weighed down by a long tradition of pessimism and catastrophic prediction. Missionaries, colonial officers, and settlers all had their reasons for vociferous anxiety about disastrous land use; and their lurid visions of degradation, drought, food shortages, and starvation still colour many current official, academic, and non-governmental organization (NGO) publications.

We begin this chapter by illustrating the remarkable persistence and popularity of gloomy views about how difficult, if not impossible, it will be to sustain increases in agricultural output. We find little support in the available data for generalizations about an irreversible African environmental crisis. And we are not convinced that the barriers said to limit the adoption of Green Revolution technologies are all insuperable; on the contrary, we point to evidence, in Section 9.4, of interventions and economic policies that have been remarkably successful in increasing agricultural productivity. We argue that there is considerable agro-ecological potential to build on the success of recent interventions to increase yields.

As in Chapters 7 and 8, we insist that policymakers would be wise to question the quality of the data used in debates about agricultural technology, especially data used to justify ineffective interventions to reduce land degradation and hunger—such as brutal interventions by agronomists and others to police (and resettle) rural populations and demand changes in how they farm, or massive expenditure on the salaries of agricultural extension officers and on subsidies for micro-finance institutions (MFIs).

9.2 Panic and Paternalism

In the 1820s, the Scottish missionary Moffat ‘was pre-disposed to designate the inhabitants of all the dry lands north of the Orange River as responsible for a situation of moral and environmental disorder’. Moffat depicted the Tswana as (p.220) ‘environmental destroyers’ in urgent need of European Christian tutelage. This Methodist missionary’s disapproval was echoed by colonial forestry and conservationist officials, ‘directly equating veld-burning and tree-felling carried out by Africans with moral degeneration and criminality’.1 By the 1930s it had become increasingly fashionable to warn of the extreme dangers posed by soil erosion. The Colonial Office in London and administrators in East Africa had become convinced of the need to intervene coercively to regulate the husbandry practices of African farmers, because ‘the apparently increasing incidence of drought conditions in many parts of East Africa over the period 1926 to 1935 indicated that the region was becoming progressively more arid’.2 After the Second World War, ‘the dominant colonial view was that African farmers were incompetent as they were responsible for environmental decay in the reserves’.3 The Head of the Economic Bureau of the Colonial Office (and his French counterparts) doubted the possibility ‘of any technical change at all’.4

In the post-colonial period, neo-Malthusian alarmism influenced agricultural policy in many African countries. The historical evidence for Africa does not provide much support for these policies, suggesting that there is no direct link between soil degradation and population growth: ‘There are cases where “more people” accompanied “less erosion”, as well as cases where soil degradation occurred in spite of declining population pressure.’5 Nevertheless, officials, agricultural economists, and agronomists in Africa continue to claim that population growth leads to environmental degradation. For example, the Director of Statistics at the African Development Bank warns of the dangers of downward spirals and an immiserating rural process throughout Africa:

Rapid population growth, inadequate food production … and increasing degradation of natural resources have created a vicious circle of poverty and environmental degradation, especially in rural areas.6

United Nations Conference on Trade and Development’s (UNCTAD’s) Economic Development in Africa Report highlights the same trends.7 Some of the most influential American agricultural economists repeat these warnings, claiming that African farmers are making the wrong choices:

[A]t present most African smallholders appear not to be choosing sustainable paths, hence the interlinked crises of rural poverty, declining per capita agricultural productivity, and environmental degradation.8

(p.221) This long history of anxieties and periodic panics about impending rural catastrophe can partly be explained by elite preoccupations with political risk. There might be threats to those in power if land degradation and declining agricultural productivity cause accelerated migration by ‘climate refugees’ out of rural areas, swelling the slum population and the ranks of the urban dangerous classes, or the number of squatters invading more fertile capitalist farms. In Nairobi, for example, rapid migration into squatter settlements from rural areas appears to have reinforced ‘the government perception that informal settlers are uneducated, unhealthy and dangerous. Indeed, the informal settlements are perceived as crime zones.’9

These conventional wisdom fears about migration are very weakly supported by the evidence. There are other—and well-established—facts about migration from rural Africa that subvert the conventional views: first, only a very small proportion of the projected increase in Africa’s urban population will be the result of climate change impacts—about 0.1 per cent according to recent estimates; second, urban conflict and food riots in Africa have complex causes and cannot simply be attributed to an influx of ‘climate refugees’ or to anthropogenic climate change, variability in food production, food availability or prices;10 third, there is evidence that climate change may—paradoxically for those unfamiliar with Hirschman’s hiding hand (see Chapter 6)—create important opportunities for rural–rural migration and encourage investment to increase crop production and rural incomes, for example in new areas of coffee cultivation in the south-west highlands of Ethiopia.11

Internationally, moral panic about ‘climate refugees’ has been encouraged by some of the more xenophobic media, by NGOs, by political demagogues and the leaders of populist movements in the European Union (EU).12 The imagined political consequences of Africa’s agricultural retrogression or stagnation do not require further discussion here. Having suggested that it is important to look closely at the evidence before indulging in fervid speculation about political futures, we turn to a more detailed examination of the most prevalent (negative) assessments of rural technological change in Africa, especially the conclusion that Africa’s Green Revolution has been ‘delayed and weak’.

9.3 Chronicle of a Failure Foretold

The Green Revolution, when cereal yields jumped from one to two or more tonnes per hectare in a few years after the late 1960s, has been described as the most (p.222) important episode of agricultural innovation in modern history; the World Bank (and others) have spent more than a decade trying to explain why ‘it didn’t happen in Sub-Saharan Africa’.13

[T]he impact of agricultural innovation in sub-Saharan Africa cannot be compared to its success in transforming rural economies of many Asian and Latin American countries during the 1960s–1990s.14

In the period after 1960, public funding was used ‘to apply scientific understandings of genetics to the development of improved crop varieties that were suited to the growing conditions of developing countries’. The initial research to develop high-yielding varieties (HYVs) of wheat and rice was extraordinarily successful, resulting in an extremely rapid spread of HYVs from the research centres to many farms in similar agro-ecological areas, especially to those areas in countries with extensive irrigation and/or reliable rainfall. In other areas and for other crops, including crops that are particularly important in Africa, diffusion has taken much longer.15 For example, in the 18 years between 1965 and 1983, HYVs of rice and wheat were adopted on more than 120 million hectares in Asia and Latin America (but only about 0.7 million hectares in sub-Saharan Africa).16

Africa is said to have had ‘a delayed and weak Green Revolution’ and this failure is explained in different ways: Africa does not have large tracts of agro-ecologically similar land, comparable to the irrigated lowlands of South East Asia or the Indo-Gangetic plane of South Asia; Africa’s production environments are extremely heterogeneous and localized so that no single regional or district package of HYVs and agrochemicals can be recommended and distributed; Africa’s soils are generally poor and degraded;17 very low and stagnant levels of fertilizer use exacerbate land degradation;18 much of the land in Africa’s dry zones is currently poorly responsive to agrochemical inputs and ‘generally unfavourable for agriculture’; many dryland zones are badly served by transport infrastructure so that many of the agricultural producers living in these zones (about 171 million people) face extremely long travel times to reach the nearest large town;19 if they purchase agricultural inputs the costs will be relatively high and the farmgate prices they receive for their output are likely to be relatively low;20 although less than 3 per cent of Africa’s total cultivated area is irrigated—compared to about 39 per cent in South Asia and 29 per cent in East Asia—the rate of expansion of the irrigated area since 1961 has been very much slower than in tropical Asia, averaging only about 1 per cent per year since 1995; in several African countries (p.223) there has been a decline between 2000 and 2015 in the percentage of arable land equipped for irrigation; the cost per hectare of irrigation projects in Africa has been very much higher than the cost in other regions, partly because African irrigation projects are relatively small—too small to take advantage of scale economies to reduce unit costs.21

The rapid development of rice and wheat HYV production appears to have depended on two earlier specific forms of investment: first, a long history of accumulating the infrastructure required for irrigation and for rural transport; and, second, decades of appropriate earlier research on wheat and rice in advanced capitalist countries prior to the 1960s. There was no similar stock of scientific knowledge about the crops—cassava, yams, millets, and sorghum—and the crop mixes that are particularly important in Africa.22 Sorghum and pearl millet account for about one third of Africa’s cropped area, but hardly any of Africa’s scientists are currently working on these crops—less than 5 per cent of full-time equivalent (FTE) researchers. Since the 1960s, national agricultural research programmes in Africa have received relatively little funding, especially for their operating budgets, and their expenditure was particularly volatile and squeezed in the period from the mid-1980s to the end of the 1990s.23

One policy conclusion that could be drawn from this abbreviated list of the difficulties involved in cultivating areas with degraded soils and erratic rainfall patterns would be to decide to concentrate on lower hanging fruit, that is, to allocate far more resources to zones with much higher immediate production potential and lower risks of climate shock. The slogan that each and every administrative area should receive a similar amount of state resources to promote rural development may be politically seductive—because it addresses regional inequalities, historic neglect, or ethnic discrimination—but is likely be costly in terms of lost opportunities for accumulation and technological progress. It is a false remedy for the problem of insufficient ‘inclusiveness’ in resource allocation.

If Africa has failed, for all the reasons given (and others), to replicate the record-breaking speed of Asia’s Green Revolution, there is no need for policymakers to despair or accept all the pronouncements made by the prophets of ecological doom. For it is not difficult to make certain negative types of prophecy: we can confidently predict, for example, that most countries in the world will fail to produce many marathon runners as fast as Eliud Kipchoge or Mary Keitany. But, if we use a less demanding criterion, we can predict great success for many non-Kenyan aspiring athletes, that is, we can be confident that the average time they take to complete a marathon will fall substantially. Policymakers should not be browbeaten by long lists of reasons for failure; they can reject the usual story of (p.224) African agricultural production as shrouded in an air of dismal inevitability, of preordained productivity stagnation—failure foretold. Less stringent criteria can be used to assess African agricultural success, such as the speed of output and yield growth relative to Africa’s own past performance, or relative to the agricultural growth rates achieved when Europe was industrializing.

It is important too for policymakers to recall the failures of past predictions about future trends in African agricultural output. There were many confident predictions, for example in the 1980s and 1990s, of progressive declines in agricultural production as a result of the HIV epidemic; but some of the academic prophets later published a rare mea culpa, admitting that ‘as advocacy took over, so science flew out of the window’. When the areas that were the original source of the data used to support these catastrophic predictions were resurveyed, it was discovered that ‘the AIDS epidemic seems not to have had the profound long-term impact in this part of East Africa that was predicted 20 years ago’.24 There has not been enough reflection on the reasons for these mistakes, including the severe limitations of the researchers’ aggregated conceptual model of ‘a Ugandan farming system’ and the unreliable (but trusted) methods of data collection.

9.4 Measuring Agricultural and Food Production

Assertions about food production in Africa have often been quite gloomy. ‘Comparing the 1970s with the present’, as one example has it, ‘it is readily apparent that the state of African agriculture is far worse now.’25 There is a particular anxiety that food production has suffered throughout Africa because of greater integration into global markets. Thus:

Most developing countries have succumbed to the demand to open up and engage in free trade. This produced area diversion to export crops, led to decline in the food grains growth rate which fell below the population growth rate, resulting in falling per capita output and availability of food grains … we find declining per head food grains output combined with fast growing per head exportables output in every important developing region … and the whole of Sub-Saharan Africa.26

But we will argue that these assertions—often made by aid bureaucrats as well as by Marxist or neo-Marxist scholars—are questionable. This is a continuation of an (p.225) argument made more than three decades ago by one of us, who queried pessimistic pronouncements about African food production that were as popular then as they are now.27

Too little has changed since the mid-1980s: there are still many reasons for extreme caution when using African food and agricultural production statistics, including the fact that agricultural censuses have become less common. The Food and Agriculture Organization (FAO) itself considers that only two countries in sub-Saharan Africa have high standards in data collection, while standards in 21 other countries remain low, and the remaining African countries could not be rated.28 Published data available on the Food and Agriculture Organization Statistical Database (FAOSTAT) and CountrySTAT sites do not always agree about the level of cereal and starchy root crop production, for example in Nigeria or Zambia.29 The FAO has argued since the 1950s that high-standard, reliable estimates of yields can only be obtained by using their recommended ‘gold standard’ method, that is, a crop-cutting survey. But in a rigorous discussion of crop-cutting samples to measure yields on maize plots in Uganda several concerns were raised regarding the accuracy of even the most costly crop-cutting methods.30

Most commentators on hunger continue to rely on the estimates of domestic food production in the FAO’s published data on food balance sheets (FBS), which are the basis for estimates of food consumption per capita in all African countries. But more reliable, individual-level national dietary surveys have established that the FAO estimates tend to either over- or underestimate consumption of most types of food.31 These well-known problems with FAO food production and consumption data, as well as with the Gallup data on self-reported degrees of hunger, have not prevented the United Nations Development Programme (UNDP), for example, from expressing the gloomy view—based on these data—that African dietary intakes have been growing too slowly and that African food production has been growing ‘at a very slow rate’.32 Other United Nations (UN) officials including, for example, a UN Assistant Secretary-General for Economic Development, are also pessimistic about nutrition, but for very different reasons: ‘In Zambia, greater use of seeds and fertilizers from agribusiness tripled maize production without reducing the country’s very high rates of … malnutrition.’33 No evidence is provided to support this paradoxical assertion: in fact, the prevalence of undernutrition (PoU) and of child stunting fell in Zambia between 2004 and 2006 and 2015 and 2017. Gloom has also been spread by Michael Lipton, who (p.226) is well aware that ‘Africa’s data for output of food staples are largely worthless’, yet who nevertheless asserted in 2012 that ‘calorie output and intake per person in most of Africa are no higher than in the early 1960s’.34

Despite all the dismal pronouncements quoted, it is surprisingly easy to use FAO (and United States Department of Agriculture (USDA)) data to question gloomy assessments of food and staple production in Africa and of the supply of calories available per person. We do not believe that the data used in Figures 9.1 and 9.2 are reliable. We will continue to provide policymakers with much ammunition to criticize them. But the aggregate official data on food production and food supply used in these figures can serve a similar purpose to the International Labour Organization’s (ILO’s) suspect data on African trends in employment presented in Chapter 7: they frame and provide a corrective perspective to the widespread pronouncements by economists about Africa’s allegedly tragic performance.

Technical Change and Agricultural Productivity

Figure 9.1 Food production and average food supply in sub-Saharan Africa, 1960–2015

Note: Countries included: Angola, Benin, Botswana, Burkina Faso, Cabo Verde, Cameroon, Central African Republic, Chad, Congo, Côte d?Ivoire, Eswatini, Ethiopia, Gabon, Gambia, Ghana, Guinea, Guinea-Bissau, Kenya, Lesotho, Liberia, Madagascar, Malawi, Mali, Mauritania, Mauritius, Mozambique, Namibia, Niger, Nigeria, Rwanda, Sao Tome and Principe, Senegal, Sierra Leone, South Africa, Togo, Uganda, United Republic of Tanzania, Zambia, Zimbabwe.

Source: FAOSTAT (2018).

Technical Change and Agricultural Productivity

Figure 9.2 Maize and rice production and yield in sub-Saharan Africa, 1960–2018

Source: USDA (2019).

We now turn attention to the rising trends in the output and yield of those basic food grains that are so critical for the consumption of Africa’s poor. The trends in maize production are particularly important because maize is the most commonly (p.227) cultivated cereal in Africa.35 Because crop failure, drought, and yield fluctuations are so common in Africa, it is always important to look at long-run data: the production data for maize and rice in Figure 9.2 cover more than fifty years.

It is also important to stress (again) the heterogeneity of African experiences: recent agricultural performance has been very much more impressive in some African countries and areas than others. For example, between 1990/2 and 2012/14, rather rapid growth rates in total and per capita agricultural production were recorded for Malawi, Ethiopia, Ghana, Mozambique, and Tanzania, while, over the same period, total and per capita agricultural output plummeted in Zimbabwe and the growth rate of per capita output was relatively slow in Senegal and the Côte d’Ivoire. Within countries, a high and probably increasing proportion of total marketed agricultural output was produced by a small number of farms, with many farmers only producing tiny or even declining amounts. The policy implications of more disaggregated agricultural production data will be discussed in Sections 9.5 and 9.7, while the mediocre growth of agricultural exports and the rapid growth of food imports—which still account for a very small percentage (about 2 per cent) of all food consumed in Africa36—were noted in Chapter 5.

(p.228) The trends shown by the unreliable data in Figures 9.1 and 9.2 may be considered plausible to the extent that they are consistent with much more reliable evidence, such as such as the evidence on undernutrition (measured by child stunting and provided in the United Nations Children’s Fund (UNICEF), World Health Organization (WHO), World Bank Group Joint Malnutrition Estimates). These estimates suggest that, between 1990 and 2016, there was a fall in the proportion of sub-Saharan African children under the age of 5 suffering from stunting—from 45.7 per cent to 33.6 per cent.

If food production, availability, and calorie intake were actually failing to improve in sub-Saharan Africa, then well-documented and widespread declines in child stunting become difficult to understand, because improved nutritional intake does appear to play an important role in the growth performance of African infants.37 Apart from the decline in stunting, a more comprehensive measure also shows a very large reduction since 1990 in the health burden of chronic hunger in sub-Saharan Africa. Disability-adjusted life years (DALYs) have been used to measure the number of healthy life years lost in a population that can be directly attributed to chronic hunger. One econometric analysis suggests that Africa’s rising total food supply has been associated with a lower burden of chronic hunger in Africa.38

The FAO (in collaboration with other international organizations monitoring food, hunger and undernutrition—International Fund for Agricultural Development (IFAD), UNICEF, WHO, and the World Food Programme (WFP)) publishes an annual report that appears to be designed to convince bilateral and other funders of the need to replenish the budgets of these organizations. In 2018, this flagship report claimed that ‘Undernourishment and severe food insecurity appear to be increasing in almost all subregions of Africa.’ The necessary caveats that the FAO itself makes about this self-serving headline about recent trends in the PoU are relegated to an appendix:

Due to the probabilistic nature of the inference and the margins of uncertainty associated with estimates of each of the parameters in the model, the precision of the PoU estimates is generally low.39

9.5 Against the Poor Measurements behind a Defence of the Small Farm Orthodoxy

The latest donor efforts to improve the quality of data on crop production and yields, the Living Standards Measurement Study—Integrated Surveys on (p.229) Agriculture (LSMS-ISA), are funded by the Gates Foundation and managed by the World Bank rather than the FAO or African statistical agencies. Implementation began in 2010 but to date these household panel surveys cover just eight African countries. We already noted, in Chapter 7, one very important methodological limitation of the LSMS-ISA surveys: their incomplete coverage of the national agricultural sector and exclusion of the most dynamic large-scale capitalist farm enterprises. Moreover, the method used by LSMS-ISA to measure the crop yields in the sampled agricultural households is unreliable: these surveys solicit farmer-reported information on crop production at the plot level; but research in Uganda and Ethiopia has shown that farmer-reported data suffers from systematic measurement error and leads to biased estimates of production and yield. When compared to more reliable measures of yield—based on remote sensing and full-plot cutting, as opposed to cutting from a small sample within a plot—the farmer-reported yields are particularly likely to be overestimates on smaller plots; on average, these self-reported yields in Uganda were almost double the yields measured by more accurate methods.40 Similarly, surveys of maize production in Ethiopia indicate that self-reported production is likely to be over-reported by up to 50 per cent on small plots, but under-reported by 25 per cent on larger plots.41

Evidence this faulty immediately calls into question an ideologically powerful piece of conventional wisdom. For the implication is that African household survey evidence supporting an inverse size–productivity relationship is deeply unreliable. Policymakers should not assume that smaller farms usually achieve higher yields per hectare than larger farms, or that large farms in Africa are unlikely to benefit from scale economies. Faith in the relative efficiency and productivity of smallholder farmers—a powerful nugget of common sense (see Chapter 3)—underpins almost all policy advice on rural development and poverty reduction, but this advice appears to be based on the naive view that the largest and richest farmers have actually been included in the national sampling frame used by Agricultural Household Surveys (and will usually agree to tell official surveyors the truth about output and yields on their farms) and that smallholder farmers give a precise account of their own yields.

The conventional wisdom is that ‘the most productive farms will be small in most places in Sub-Saharan Africa’ and that these farms, because they are ‘managed and worked primarily by family members’, can reduce hired-labour supervision costs and avoid shirking to achieve high yields. Furthermore, differentiation or a change in the size distribution of farms in Africa and the increasing dominance of capitalist farmers is unlikely in the foreseeable future: ‘Poorly functioning land markets … work to help keep farms small, because they keep the risks and (p.230) costs of renting or purchasing land unnecessarily high.’42 This conventional wisdom fuels campaigns:

Melinda and I started to realize that the poorest people in the world shared an occupation in common: They were small farmers. The conclusion was obvious: They could lift their families up by growing more food.43

The Agriculture and Rural Development Team in the World Bank confirm that ‘Most rural development strategies in Sub-Saharan Africa focus on improving the productivity of smallholder farms.’44 When the evidence is ambiguous or flatly contradicts the hegemonic faith in the productive potential of these farms, some mini-farm champions—especially those trained in mainstream micro-economics—can reach for another weapon to defend their belief: they can deploy the simple neoclassical model of a unitary farm household to prove that small farmers ‘could’ (theoretically) be more efficient ‘if’ market imperfections and information asymmetries disappeared. These neoclassical economists can confidently anticipate the disappearance of such market glitches as soon as appropriate institutions, for example, cooperatives, a high-tech electronic commodity exchange, and competitive financial/insurance services, are conjured out of the air by waving a magic wand or brandishing a mobile phone.45

The two most important policies and expenditure priorities derived from the small farm household model and pessimism about technological progress in Africa involve efforts to provide additional credit and extension advice. Even economists critical of neoclassical methods and optimistic about African developmental prospects agree that these policies are urgently needed. Ha-Joon Chang, for example, argues that ‘credit provision to small farmers has been one of the most important challenges that has faced the policy-makers in the early stages of economic development’.46 Chang quite rightly emphasizes the significance of state provision of, subsidies to, and regulation of rural credit, but seems to put too much faith in what we have shown to be a very vague and problematic category of ‘small’ farmers. We question the impact of these popular policies to support extension services and private sector credit to small farmers in Section 9.6; and in Section 9.7 we discuss the potential for alternative policies and investment priorities.

(p.231) 9.6 Do Farmers Fail without Advice and Subsidized Credit?

Farmers have good reason not to tell the truth to outsiders and rural development tourists participating in focus group discussions. But we do think that some of what they say about extension officers, or about the advice they have received from young people with technical qualifications employed by the Ministry of Agriculture, should be taken seriously. We have met successful capitalist farmers who simply laugh at the idea that inexperienced, wet-behind-the-ears advisors know what they are talking about and might have something useful to say. The poorest, especially women who cultivate very small areas and do not have the cash to purchase agrochemicals and HYVs, have told us that no advisor has ever visited their farms; they do not even know the name of the local extension officer or the precise location of his or her office.

But we do not have to rely on our own anecdotes to question the impact of agricultural extension services in Africa. Historians of the most successful episodes of technological change in rural Africa emphasize the role of African traders and capitalists—not agricultural extension officers—in explaining the introduction and extraordinarily rapid expansion of cocoa in Amansie between 1900 and 1916.47 More recently, there have been hundreds of evaluations of the impact on agricultural production of expenditure on extension. Policymakers may find it tedious to wade through all these studies, because when polite and judicious mainstream economists are funded to evaluate the impact of extension, they usually hedge their bets on the evidence: ‘Results from these studies are mixed and few generalizations can be made’; ‘evidence from systematic reviews regarding the effects of extension and advisory services on poverty is extremely thin’; ‘there is limited rigorous evidence on the effectiveness of … potentially scalable extension services provided by public agencies in developing countries’.48 These diplomatically fudged conclusions suggest there is an elephant in the room—an awkward fact that no one wants to mention.

The uncomfortable truth is that, after many decades of devoting a high proportion of government recurrent expenditure on agriculture to paying the salaries and wages of extension staff, it cannot be shown that their work has improved agricultural productivity. Negative evaluations of extension services have sometimes led to changes in the types of extension projects donors are willing to fund. For example, the high recurrent cost training and visit model was all the rage at the World Bank, promoted from 1975 to 1998 in over fifty countries, but it fell out of favour in Africa to be replaced by models that purport to be more participatory, geared to farmer field schools and to the average rather than to ‘progressive’ (p.232) farmers, sensitive to gender relations and to farmers’ demands, based on farmer-to-farmer interactions, and decentralized.49 But there is no evidence that the new fads have made much difference. In Kenya, for example, a field experiment on the impact of text messaging concluded that ‘we do not find consistent evidence that advice delivered through mobile phones was effective in increasing knowledge or use of recommended inputs’.50 In Ethiopia, research on an NGO programme in North Shewa Zone (to train farmers in soil conservation) concludes that ‘there were no differences in crop productivity between attendees and non-attendees, and higher rate of participation in farmer trainings was not associated with changes in land-use intensity at the village level’.51 An evaluation of a larger NGO programme providing training in soil fertility management to the members of farmer-based organizations in Ghana’s Upper Volta Region also could not demonstrate a positive impact.52 Government-provided advice through salaried cadres of advisors appears to have been equally ineffective in Ethiopia and Malawi, for example.53

Why then do so many governments, donors, and NGOs continue to devote resources to training and advising smallholders in Africa? There are several possible answers: it may be politically difficult to retrench large numbers of government workers or to disappoint college-trained young aspirants to posts in the bureaucracy; it may also be politically convenient to have eyes in the villages and a means of policing potential troublemakers, for example, by fining and imprisoning them or refusing to supply them with subsidized inputs and credit. Rwanda and Ethiopia provide good examples of rural areas where the line between state officials and party officials is blurred and an important function of the development and extension agents, the umudugudu leaders or ‘the delegates’, may be to eliminate opposition and entrench the power of the existing elite.54

Insistent demands to increase expenditures on agricultural extension, like proposals to increase the level of resources devoted to agricultural credit, micro-credit, and financial inclusion, have often received ideological support from mainstream economists; these demands have also been amplified by domestic and international political pressures that policymakers cannot easily ignore. According to mainstream theory, farmers and other potential capitalists in rural Africa suffer because they are cut off from the opportunities for investing, risk-taking, and risk spreading that would be available through better financial integration into larger national and global financial markets. They do not have access to critical financial services because they are ‘afflicted’ by fragmented, imperfect markets; in rural Africa markets do not look anything like the neoclassical ideal type—information asymmetries and the high costs of attempting to enforce (p.233) contracts in rural areas are major obstacles to adopting new inputs to accelerate development.55

The policies prescribed by those who believe in reducing market imperfections to allow untrammelled individual entrepreneurs to spearhead African economic development are predictable: states and donors should focus on supporting institutions that ensure the legal enforcement of financial contracts, make property rights more secure, and diffuse information. Of course, states should also remove all forms of ‘financial repression’—directed credit and the regulations that hinder the entry and expansion of new private (and especially foreign corporate) lenders in Africa. More recently, mainstream policy conclusions have become a little more nuanced: it is sometimes acknowledged that the private lenders and financial institutions may have to be ‘crowded in’; they will need to be bribed by access to privatized assets at very low prices and by additional state subsidies, especially if the aim is to introduce innovative financial products (such as crop insurance now heavily subsidized by United States Agency for International Development (USAID) and the Gates Foundation).56

The advocates of conventional liberalization policy packages cannot avoid the unpalatable conclusion that there is a case for state and donor expenditure to subsidize rural credit, because they have been unable to find any evidence of the surge in private sector lending anticipated by the mainstream theorists of ‘financial repression’ (see Chapter 4), despite the space for private sector initiatives created by savage cuts in public sector agricultural credit between the 1980s and the end of the 1990s.57 The dominant view now is that Africa’s rural credit needs can effectively be provided by subsidizing MFIs that benefit from lower transaction costs and informational hurdles than larger-scale, bureaucratic, and more formal lenders such as national rural development banks. By the 1990s, the cost of the subsidies provided annually by donors to the newly fashionable micro-finance sector was about US$1 billion.58

Despite massive subsidies and repeated policy recommendations to increase interest rates, very few MFIs in sub-Saharan Africa show any signs of achieving operational sustainability or reaping the cost advantages of operating at scale. Those institutions that are least effective in reaching poor farmers have received more subsidies than institutions making smaller loans to the poor.59 Most credit in rural Africa continues to be supplied by family, friends, usurious shopkeepers/traders, and employers seeking to secure labour at below the local market wage rate. Where there is access to micro-credit, it does not appear to have a discernible impact on the use of new agricultural inputs such as chemical fertilizer. This was (p.234) the conclusion of case studies in Tanzania and Kenya, for example, where panel data also suggest that both rice yields and household income have failed to increase with credit use.60

There is even better evidence that most MFIs fail to provide any credit to the smallest farmers and the poorest rural people. Access to any type of formal credit by African households at the bottom of the distribution of income is very low and barely increased between 2011 and 2017.61 The much-hyped success of the fintech innovation for digital money transfers in Kenya (M-Pesa) cannot be shown to have reduced the poverty of those Kenyans who do not have links to the relatively wealthy. In western Uganda, women’s access to a micro-finance cooperative did not prove ‘an unconditional blessing’. In South Africa, where the micro-credit bubble expanded at a quite extraordinary rate, the National Income Dynamics Survey panel data 2008 to 2015 suggest that access to micro- and informal loans by the poorest rural women has a negative effect on their quality of life.62

The distribution of credit in Africa usually excludes poor borrowers in rural areas and results in problems of financial sustainability. Most donors and the public relations staff of the MasterCard Foundation and Citibank now prefer to focus on MFIs’ contribution to the nebulous goal of ‘financial inclusion’, rather than sticking to the old and widely missed target of poverty reduction.63 Ignoring evidence on the impact of expenditures to subsidize rural credit may be convenient for other reasons: regimes often seek to attract support from influential power brokers by channelling short-term credit to wealthy and powerful rural borrowers—not merely at de jure rates of interest way below those charged by private sector informal lenders, but also on the implicit understanding that de facto there will be no repayments at all. We have argued that most expenditure to promote MFIs and on financial inclusion continues for ideological and political reasons: investment resources have not been allocated to those national institutions that have historically (in Japan, Brazil, South Korea, the Netherlands, Austria, and Switzerland, for example) been extraordinarily successful in providing savings and money transmission facilities to relatively poor people—post offices.64 On the contrary, there has been donor pressure to limit the role of post offices and, in some African countries (Mali, for example), a structural adjustment programme led to the demise of the post’s facilities and network.65

(p.235) 9.7 Massive Potential to Sustain and Accelerate Increased Yields per Hectare: Policy Opportunities

There are several ways to make a more realistic assessment of Africa’s potential to improve its agricultural performance. We begin by emphasizing the immense gap between the yields actually achieved on some farms and the low yields achieved by others farming in the same agro-climatic zones. An important implication of these carefully measured cropland yield divergences within homogeneous zones is that there is a huge potential to increase output and productivity, if currently observed best cultivating practices are diffused throughout the zone.66 Satellite-derived images have made an important contribution to identifying hotspots of yield divergence within each African country; they could be used by policymakers to argue for concentrating investment in specific sub-districts and for closer monitoring of the impact of policy on yields.

The current level of adoption of HYVs also provides a good indicator of a potential for dramatic increases in crop production. Although the HYV adoption rate in sub-Saharan Africa probably increased by as much in the 2000s as in the four preceding decades, adoption is very far from universal, as suggested by the data on yield divergence. By 2010 the total area sown to improved varieties of food crops exceeded 37 million hectares, more than double the estimated area in 2000; but this impressive achievement represented only 35 per cent of area planted to these crops.67 Adoption rates are, for example, only 23 per cent for improved varieties of cowpea in sub-Saharan Africa; they are as low as 20 per cent for improved varieties of a key food crop (cassava) in some countries. While improved varieties of maize are now cultivated on 57 per cent of the total maize area in West and Central Africa, some countries in this region have only managed to plant improved varieties on less than 10 per cent of the maize area; in Mozambique and Angola, improved varieties of maize are also planted on only 10 per cent of the maize area, compared to about 70 per cent in Kenya and 84 per cent in Zambia.

Increasing the proportion of African farmland cultivated with varieties developed through modern crop-breeding techniques could make a huge contribution to improving living standards and to economic growth.68 Already, according to the estimates of one econometric model, the adoption of improved varieties in sub-Saharan Africa has boosted average net crop yields by about 0.55 tonnes per hectare, equivalent to an increase of 47 per cent above average yields achieved during the period 1976–80. Part of the explanation for the faster rate of adoption of HYVs in some African countries than others is that these countries have invested much more in agricultural research, enabling them to release larger (p.236) numbers of new varieties and achieve more rapid rates of diffusion.69 Unfortunately, in a rather large number of sub-Saharan African countries the growth rates of public agricultural research spending between 2000 and 2014 were stagnant (seven countries) or negative (five countries).70 Even worse, in several countries a smaller share of public agricultural expenditure was devoted to research and development (R&D) than to extension services, despite the much higher estimated returns to expenditure on R&D than other types of public agricultural spending.71 There is an urgent need for a better-informed policy debate about how to increase the level and improve the functional allocation of public agricultural spending. We argue that proposals for reallocating spending both to R&D and to export crops should now move to the top of the policy agenda. But a precondition for improving government policy is the availability of much more transparent, consistent, and accurate data on current patterns of public agricultural expenditure in Africa.72

The rate of adoption of HYVs, especially of new varieties of maize, rice, and vegetables, could be accelerated by prioritizing public sector investment in irrigation and water control because, as noted in Section 9.3, these investments in ‘the leading input’ improve the prospects for introducing not only HYVs, but also complementary inputs—such as fertilizers and agrochemicals—that farmers can use to maximize crop yields. In South Asia the irrigated area represents about 42 per cent of the cultivable area; in sub-Saharan Africa most of the area currently cropped (at least 95 per cent) is cultivated in the rainy season without irrigation; producers face the risk of extremely uncertain water availability both in terms of quantity and timing.73 The potential to reduce the (often life-threatening) risks of relying on rainfed cropping systems is considerable: estimates of the precise magnitude of this potential do differ and are sensitive to underlying assumptions about cost, yields, internal rates of return, and so on, but all agree that it is possible to escape some of the risks of rainfed production; that the scope for economically and technically viable investment to expand the area under irrigation is considerable.

In the drylands of sub-Saharan Africa, about 5.2 million hectares are currently irrigated. There is a well-established potential to irrigate an additional 14 million hectares, equivalent to about 8 per cent of the total area that is currently being cultivated in the drylands. The impact on sub-Saharan African crop production of this realistically assessed addition to the irrigated area would be ‘transformational’: one estimate is that dryland cereal production could increase by 52 per cent.74 The impact would be particularly dramatic in those countries where the area that could be developed for irrigation accounts for a large proportion of the total cultivable (p.237) area. For example, in Malawi fully 70 per cent of the cultivable area in the drylands could be brought under irrigation and in Ghana and Tanzania the comparable proportion is more than 25 per cent. Nigeria has the potential to irrigate millions of hectares of dryland, but many other countries also have the potential to irrigate large areas—200,000 hectares or more—including Angola, Burkina Faso, Chad, Ghana, Madagascar, Mali, Niger, Senegal, Somalia, South Africa, and Uganda.75

Successful cultivation of HYVs has usually been associated with improved water management and has depended on the extremely positive response of these seeds to fertilizer. An increase in the use of fertilizers appears to have played a key role throughout the world in raising cereal yields and promoting structural change.76 Many farmers in sub-Saharan Africa are currently not benefitting sufficiently from the potential yield gains offered by plant genetic improvement because they are not applying adequate nutrients to their soils.77 Despite recent relatively fast rates of growth of consumption in some African countries, fertilizer consumption remains way below the levels recorded in South Asia: in Bangladesh, for example, average consumption per hectare is about 280 kg, compared to the sub-Saharan African average of 15 kg per hectare. There is massive potential to increase and improve fertilizer use in Africa from a very low base, which has motivated investors such as Dangote to invest billions of US dollars in fertilizer production in a number of African countries.78

Policies that combine increases in Africa’s irrigated area with increased (but appropriate) levels of nutrient application would increase yields in Africa. Enhanced investment in agricultural R&D should result in more disaggregated and efficient fertilizer recommendations linked to the push for public sector investment in irrigation and transport infrastructure in precisely those rural areas with the greatest potential for wage employment and export production. (p.238)


(1) Grove (1989: 166, 184).

(2) Anderson (1984: 322–3).

(3) Shanguhyia (2015: 2).

(4) Cooper (2004: 20).

(5) Koning and Smaling (2005: 5).

(6) Lufumpa (2005: 369).

(7) UNCTAD (2018: 17–18).

(8) Reardon et al. (1999: 377, 389). For similar views see: Nhamo et al. (2019: 2); and Rigaud et al. (2018: 82).

(9) Kyed, Stepputat, and Albrecht (2017: 24). For the fears of the urban bourgeoisie during the British Industrial Revolution see Enzensberger (1974).

(10) Rigaud et al. (2018: 87); Selby (2014); Mach et al. (2019).

(11) Moat et al. (2017).

(12) Kelman (2019: 11); Bettini (2013).

(13) World Bank (2007: 55); Goyal and Nash (2017: 7).

(14) Ogundari and Bolarinwa (2018: 19, 3).

(15) Gollin, Hansen, and Wingender (2018: 2).

(16) Fuglie and Marder (2015: 339).

(17) Tittonell and Giller (2013: 88).

(18) World Bank (2008: 55).

(19) Cervigni and Morris (2016: 115 and 49–63). See also You (2008: 1).

(20) Porteous (2020: 2).

(21) Nhemachena et al. (2018: table 3); Inocencio et al. (2007).

(22) Gollin, Hansen, and Wingender (2018: 12, 8–9).

(23) Alwang (2015: 15); Beintema and Stads (2017: figures 11 and 13).

(24) Barnett, Dercon, and Seeley (2010: 958); Seeley, Dercon, and Barnett (2010: 333).

(25) Bryceson (2009: 56).

(26) Patnaik (2016: 145). For similar views see Sundaram and Chowdhury (2017); and Traore and Sakyi (2018: 6).

(27) Sender and Smith (1986: 100).

(28) Carletto, Jolliffe, and Banerjee (2015: 137, 134).

(29) Luan et al. (2019: 15).

(30) Gourlay, Kilic, and Lobell (2017: 10).

(31) Del Gobbo et al. (2015). Grünberger (2014: 4). Additional criticisms of the FAO’s FBS approach are summarized in de Weerdt et al. (2016).

(32) Chauvin, Mulangu, and Porto (2012: 4).

(33) Sundaram (2019).

(34) Lipton (2012a: 3).

(35) Wiggins (2018: 28). African evidence suggests that income elasticities are low for cereals and tubers (Choudhury and Headey, 2016: table 2.1).

(36) Vorley and Lançon (2016: 8).

(37) Stewart et al. (2019).

(38) Gödecke, Stein, and Qaim (2018: 25–6).

(39) FAO et al. (2018: 142). Given these margins of error, it is hard to assess the meaning of the finding that the PoU in sub-Saharan Africa ‘increased’ from 22.3 to 23.2 (ibid.: table 1).

(40) Lobell et al. (2018: 13).

(41) Desiere and Jolliffe (2018: 90–1).

(42) Larson, Muraoka, and Otsuka (2016: 9).

(44) Larson, Muraoka, and Otsuka (2016: 2).

(45) The weak economic logic, patchy evidence, and dubious claims about rural poverty advanced by believers in the inverse size–productivity relationship are discussed in more detail in Sender and Johnston (2004). The myth that poor producers in rural areas use mobile phones to obtain more accurate and lower-cost information on market prices has been criticized by Burrell and Oreglia (2015).

(46) Chang (2009: 489, 494).

(47) Austin (1987).

(48) Alwang et al. (2019: 2); Bernstein, Johnson, and Arsalan (2019); Kondylis et al. (2017: 13); Fabregas et al. (2017: 8); Waddington et al. (2014: 37). See also Ragasa et al. (2013) and Berhane et al. (2018: 22).

(49) Anderson, Feder, and Ganguly (2006).

(50) Fabregas et al. (2017: iii).

(51) Chesterman et al. (2019: 7).

(52) Andam, Makhija, and Spielman (2018).

(53) Bowser (2015: 22); Ragasa (2019).

(54) Cramer and Sender (2019a); Huggins (2017: 729).

(55) Conning and Udry (2007: 2859–60).

(56) De Bock and Gelade (2012). The uptake and renewal rates for micro-insurance in Africa are still very low.

(57) De Bock and Gelade (2007: 2867).

(58) Hudon and Traca (2011: 966).

(59) Cull, Demirgüç-Kunt, and Morduch (2018).

(60) Nakano and Magezi (2019).

(61) Awaworyi Churchill (2018); Bernards (2019: 4); Adjognon, Liverpool-Tasie, and Reardon (2017: table 5a).

(62) Bateman, Duvendack, and Loubere (2019); Bateman (2014); Meier zu Selhausen and Stam (2013); Greyling and Rossouw (2019: 10).

(63) Bateman and Chang (2012: 17).

(64) Mader (2018: 479); Rillo and Mlyamoto (2016).

(65) D’Alcantara, Dembinski, and Pilley (2014: 9).

(66) Luan et al. (2019).

(67) Fuglie and Marder (2015: 338).

(68) Gollin, Hansen, and Wingender (2018: 32–3); Walker and Alwang (2015: 74–122 and 206–227).

(69) Fuglie and Marder (2015: 356–7).

(70) Beintema and Stads (2017: 7).

(71) Benin, McBride, and Mogues (2016: 115).

(72) Mogues and Caceres (2018).

(73) Walker (2016: xvi).

(74) Ward (2016: 53, 64).

(75) Ward (2016: 58).

(76) McArthur and McCord (2017).

(77) Tittonell and Giller (2013).

(78) Oxford Business Group (2019: 18).