A green revolution betrayed? Seed technology and small-scale maize farmers in Zimbabwe

Abstract

Since the 1960s both large- and small-scale Zimbabwean maize farmers have been replacing open pollinated varieties (OPVs) with locally developed hybrids. By the 1990s, most were buying hybrid seed, though the adoption rates of new seed types were slowing. With the collapse of the Zimbabwean economy many small farmers returned to planting OPVs and saving seed, not only because hybrid seed was unavailable but also as a rational response to economic risks. Initially these risks were tied to Zimbabwe's economic structural adjustment programme, which cut extension services, reduced short-term credit and destabilised maize prices. Subsequently risks increased as land invasions on seed producing farms forced the importation of seeds with which small-scale farmers were unfamiliar, and when escalating inflation precluded the use of money as a store of value. Control of inflation, better marketing and restored supplies of local seed should see restored planting of hybrid seed.

Keywords:

• maize
• seed technology
• OPV (open pollinated variety)
• HYV (high yielding variety)
• economic risk
• inflation
• Zimbabwe

JEL classification:

• Q19

1. Introduction

In the 40 years following the end of World War II, land productivity on Zimbabwe's maize farms rose by over 300%. On small-scale farms much of this improvement occurred in the first decade following independence in 1980. In the second decade of Zimbabwe's independence the rate of increase slowed, and with the third decade the productivity of maize lands began to decline. On large-scale commercial farms this was easily attributed to the land invasions that dominated the country's agricultural life, but the decline was also observed on small-scale farms owned by black Zimbabweans. For both small farmers and large, the economic context was clearly a central issue. This paper discusses their adoption (and later rejection) of hybrid seeds as a new technology, the central thesis being that small-scale and communal-land farmers were reacting as rational economic agents to the institutional challenges of the past 60 years. Historical background for the paper is drawn from agricultural census data for the periods 1970–75, 1980–86 and 1995–2000, supplemented by data from the US Department of Agriculture's Foreign Agricultural Service.

In the context of recent history, it is well known that the period 1998–2009 saw the collapse of Zimbabwe's economy, its infrastructure and its currency. From a farmer's perspective, fertiliser and seed became difficult to access, the quality of commercial seed became uneven, money failed to function as a store of value in the period between harvest and planting, and grain prices became state controlled. We argue that small-scale farmers responded rationally to these supply- and demand-side problems by shifting into a survivalist mode in which the planted area decreased and lower yielding open pollinated varieties of maize (OPVs) replaced hybrids. We suggest, however, that such tendencies considerably antedated the recent inflation. In particular we stress those aspects of the country's structural adjustment package that increased the risks faced by farmers: declining extension services, restrictions on short-term credit, a deteriorating maize collection infrastructure and price instability.

Eicher (1995) described the rapid adoption of hybrid maize varieties by Zimbabwe's smallholders and communal farmers in the six years after independence in 1980 as a ‘maize-based green revolution’, presenting this as proof that biological technologies could be successfully adopted by smallholders. In fact, this was less a revolution than a reversion to form: Zimbabwean farmers had been planting improved OPVs since the turn of the century, and locally produced hybrid seeds since the early 1960s. Small-scale and subsistence cultivators were already comfortable with the use of hybrid seeds before the disruptions of the independence war.

The adoption rate of new hybrids slowed markedly in the 1990s, and maize productivity first stagnated and then fell. Was this slowdown in the adoption of new seeds simply an artefact of earlier successes, and was it irrational? Bourdillon et al. (2003) mention that by the late 1990s hybrid seed was already in almost universal use by Zimbabwe's small farmers. They do not mention, however, that most small farmers were still planting the old R200 series seeds even though the newer, more disease resistant, SC series was already available. They also fail to mention that one reason for the widespread use of hybrid seed was a de facto prohibition on the sale of OPVs in Zimbabwe until shortly after 2000. In consequence, while nearly 100% of maize land had been planted to hybrids, these were neither necessarily the best hybrids nor the most appropriate seeds.

This paper ties the stagnation and decline in maize land productivity that began in the late 1990s to rising farming risk. This had multiple causes, some exogenous (the 1990s had more drought years than usual), but most endogenous. These initially flowed from the economic structural adjustment programme (ESAP) of the 1990s. We argue that these curtailed the adoption of new hybrid seeds, thereby stunting productivity growth. However, the economic uncertainty of the structural adjustment era was outweighed by more absolute issues. The 2001 growing season was hampered by drought, but though rainfall returned to normal patterns for the next six years, small-scale farmers then suffered rapid cost inflation together with shortages of fertiliser, pesticides and good quality local seed (following the land invasion of many seed growing farms). In the face of these supply-side problems they resorted to saving seed from season to season, and the earlier productivity gains were lost. Unfortunately merely solving these immediate problems may not be enough to restore the health of Zimbabwe's maize sector: the long-run recovery of Zimbabwe as a granary for southern Africa will also depend on reducing the endogenous risks and uncertainties driven by government policy.

2. Risk and the rate of agricultural change

Neoclassical theory portrays the farmer as a rational economic agent whose technology adoption decision is driven by a desire to maximise expected profit subject to a vector of constraints (Just & Zilberman, 1983; Feder et al., 1985). While we endorse this as a general view of the Zimbabwean small-scale farmer, we note that this definition does not distinguish objective constraints (such as the uncertainty of future yields and prices) from subjective constraints in the form of uncertainty on the part of the farmer. This subjective uncertainty is the result of decision variables such as culture that can influence farmers' ostensibly rational reactions to specific incentives or policies.

Zimbabwe's seed producers are certainly aware of such subjective issues. On its website Seed Co, the major national producer, contended that: ‘Cultural agricultural practices are very contributory to the success or failures to maize production’ (Seed Co, 2009). Characteristics such as taste and appearance are subjective and can influence seed choice. Bourdillon et al. (2003) report that, despite their lower yields, OPVs such as Hickory King are often planted for household use because they taste better. And risk aversion is itself subjective. Roumasset (1976) argued that when high yielding varieties are seen to provide greater yields, but at higher risk, the likelihood of adopting the HYV (high yielding variety) falls as the probability of suffering a disaster-level yield rises, especially on rainfed farms (as most Zimbabwean farms are).

It could be argued that these are issues of perception rather than reality: Mosley (2002) described the hybrids responsible for the African Green Revolution as ‘offering a free lunch’ – increasing yields and decreasing the risks of drought and disease. We argue, however, that adopting new seed technologies can entail significant risks if supplies of their complementary inputs (and indeed of the seeds themselves) are uncertain. Uncertain availability of commercial fertilisers and herbicides could be a barrier to the use of HYVs by risk averse farmers (see also Feder, 1980:269).

Information asymmetry between seed producers and farmers adds a further dimension to the issue. With full information, farmers taking a minimax (strictly risk averse) view should be more likely to adopt drought and disease resistant seeds; in other words to maximise yields in worst case situations, even if this requires a commitment to fertiliser and weed killer purchases during the season. Fast-maturing drought resistant seeds provide these benefits at the cost of lower yields. They are classic risk aversion choices. Allowing farmers to extend their planting season, they further insure against low yields if the rains fail at critical times (post-germination and during tasselling). Where farmers are fully informed, seed technology choice should be driven by the prior probabilities of drought and disease. However, where the probabilities are unknowable (Knightian uncertainty) – as in post-1990 Zimbabwe, where exchange rates, inflation, and seed, fertiliser and maize prices were volatile, and the situation was outside the farmers' range of previous experience – this could curtail the adoption of new ostensibly rational technologies.

Evenson (1997:106) makes the point that mere adoption of a technology need not lead to increased productivity; this requires technical and allocative efficiency, and appropriate infrastructure and market institutions. This certainly accords with the Zimbabwean experience, where much of the post-1980 rise in small-scale farmers' output was attributable to infrastructural improvements such as new maize collection points. The subsequent fall in output may be in part attributed to the closure of these points under ESAP's financial stringencies (Jayne et al., 1994; Eicher, 1995; Smale & Jayne, 2003).

Zimbabwe's recent experience shows how uncertainties in the supply and price of production complements such as fertiliser can affect seed choice decisions. The sources of uncertainty have multiplied: fertiliser has been in short supply, seed quality has been uneven and hybrid seeds have often arrived late in the season. Established hybrids with which farmers were familiar (such as R215) have also been taken off the market, adding to informational uncertainty.

As a result, farmers have returned to lower yielding OPVs, and to saving seeds from hybrids, despite knowing that hybrid vigour is lost in the second generation. The loss of vigour when hybrid seeds are saved is not a trivial issue, especially on poor soils and in low rainfall areas. Morris et al., citing a 1997 study by Waddington in which yield depressions in second generation seed were estimated for popular Zimbabwean hybrids, say that:

At the F1 generation, the hybrids tested … out-yielded the improved OPVs, but only under favourable conditions; under poor conditions, the yield advantage of the hybrids was negligible. At the F2 generation, the yield rankings were reversed: the improved OPVs out-yielded the hybrids, with the yield advantage particularly pronounced under poor conditions. (Morris et al., 1999:53)

Similarly, Pixley and Bänziger (2001) found that the appropriate hybrids outperformed improved OPVs significantly (by approximately 30% in normal conditions, and by roughly 18% in poor conditions and in unfertilised fields), but that saving seed had the opposite effect: hybrids typically showed yield depressions in excess of 30% and top crosses lost 15%, but OPVs were only slightly depressed.

Saving hybrid seed clearly lowers overall yields. The implication is critical: where seed and fertiliser prices are high relative to the maize price, and where yields are low (below 1.5 tonnes/ha – a common yield for many communal farmers), use of improved OPVs, and seed saving from these, is economically rational. Farmers' reversion to ‘lower yielding’ OPVs despite the advent of improved hybrids was the logical response to Zimbabwe's infrastructural and economic collapse.

Commercial credit is commonly regarded as a complement to the use of new biotechnologies, though recent research has suggested that the presence or absence of credit had little impact on maize output in Zimbabwe (Musuna & Muchapondwa, 2008). Indebtedness clearly entails risk. Under the structural adjustment regime, and during the period of high inflation, small-scale farmers found it difficult to access credit. Indeed, post-1997 credit to communal farmers fell to negligible levels (CSO, 2001), though in 2000 the state did attempt to redress this by establishing an agricultural input credit scheme targeting small-scale farmers, controlling their input and output prices, and providing seed and fertiliser directly to them. Unfortunately, overzealous state controls and collapsing infrastructure meant that these schemes paralleled falling local fertiliser production (documented in the FEWS/AGRITEX National Early Warning Unit, 2002, food security update for March).

Instability in the maize price received by farmers is not a new phenomenon. Unstable maize prices, and falling real incomes for maize farmers, had already begun to characterise the maize market in the 1980s, despite the Grain Marketing Board's single channel marketing system that was then in place. In 1987 the government committed itself to a more ‘market based’ approach. Three years later this culminated in ESAP, which added considerably to the uncertainties that farmers faced in the 1990s, bringing fluctuations in the cost and availability of credit and the value of money, making the maize price even more volatile, and causing the maize marketing infrastructure to deteriorate.

Kim et al. (1992:250–2) note two implications of rising price instability for new technology adoption rates: it increases demand for technologies that save on variable inputs such as fertiliser and seed, and it reduces the adoption rate of yield increasing technologies. By the mid-1990s Zimbabwe's inflation rates were over 40%, and continued to rise, culminating in the hyperinflation of 2003–2008. Such high and unstable inflation precluded the use of cash as a store of value. Farmers were no longer able to sell a crop and save the proceeds to buy inputs in the following season, and banks were unwilling to extend credit. In such inflationary conditions, storing seed is one of the few effective means of saving, which makes the rejection of high yielding commercial hybrids appear increasingly rational.

Even if we abstract from Zimbabwe's inflationary history, experience suggests a rationale for maize price stabilisation. Structural adjustment packages that free agricultural markets can increase output prices where these have been historically depressed (as in Zambia); however, market liberalisation is often associated with increased price instability. While the modern commercial farmer can address this through futures markets, the small-scale farmer enjoys no such option. In consequence, free market macroeconomic policies increase the uncertainties confronting small farmers who are considering planting hybrid or genetically modified seeds and buying the associated inputs.

The rejection of new biotechnologies by Zimbabwe's small farmers has both sociopolitical and economic implications. The increased maize yield and production stability provided by drought and disease resistant hybrids could lower the retail prices of maize without the state having the need to resort to the price controls of the last decade, or the marketing board subsidies of the 1980s, whose suspension during ESAP so increased small farmer uncertainty.

3. Institutions and risk

A commercial maize farmer's success or failure often depends on the viability of the institutions with which he deals. Seed producers, maize marketers, fertiliser producers, extension services and credit providers: such institutions provide the foundations on which viable commercial agriculture stands. The growth of agriculture after World War II paralleled the growth of the country's broader economy, but was aided by a number of interventions, including agricultural marketing boards, interest-free loans and a free technical support programme. It was also helped by the formation of the Seed Maize Association, which was set up by interested farmers to work with agronomists to provide consistent high quality seed. This initially provided improved OPVs such as Salisbury White, and then, in 1949, the first generation of maize hybrids. These seeds gave higher yields in both normal and drought years (Bourdillon et al., 2003:5). Their success led to rapid and widespread adoption: in 1957, only eight years after these hybrids were introduced, 80% of maize land on white commercial farms was planted with them (CSO, 1965a). In 1960, a second, more successful hybrid, SR-52, was released and was adopted on two-thirds of commercial land within eight years. This was the technological reform Eicher called ‘Zimbabwe's first green revolution’ (1995:807). The key point is that these seeds were also rapidly adopted by African commercial and communal farmers on lands suited to maize.

New seeds were only a part of the sector's success. There are varying estimates of the productivity increase – some idea of the range can be found by close reading of Friis-Hansen (1995:166). The estimate that, between 1945 and 1950 and the early 1980s, commercial maize land productivity rose by 325%, is not extreme. Friis-Hansen, following Tattersfield (1982), says that of this increase only 45 percentage points accrued from improved seed while 200 percentage points were provided by fertiliser use and 60 percentage points by better agronomy and use of pesticides.

The sanctions that followed the Unilateral Declaration of Independence (UDI) in 1965 forced white farmers to move away from export crops, such as tobacco, to crops with larger local markets such as maize and cotton (see Figure 3A in Appendix). Eicher (1995:808) suggests that the move from tobacco to maize in the drier regions provided the incentive to develop short-seasoned hybrids better suited to these drier regions. These hybrids (R200, R201 and R215) which emerged in the early 1970s also suited small-scale and communal-land farmers.

Despite the bush war of the mid and late 1970s, hybrid seed was increasingly planted by smallholders. Smale and Jayne (2003) suggest that between 20 and 40% of the maize planted was hybrid, while Rusike (1998:314, Figure 17.2) suggests that by the last years of the war adoption rates were above 70%. Nonetheless, Eicher attributes the post-independence peace dividend largely to improved smallholder access to R200 hybrids, though the increase in productivity clearly had multiple institutional roots. In particular the state ensured the marketability of crop production by increasing the number of crop buying stations in smallholder areas and increased credit extension to smallholders, allowing them to purchase the fertiliser necessary to improve yields.

Although foreign firms have a growing share of the market, the institution at the heart of the post-independence Zimbabwean hybrid maize seed system has been Seed Co, a public company formed in 1983 when the Seed Maize Association and Crop Seeds Association merged. In the 1990s Seed Co focused on improving the disease resistance of hybrids. It is these improved seeds that small-scale farmers have been slower to adopt, even though the previously popular R201 and R215 varieties have not been available since 2000.

The declining use of improved seed has had both demand and supply dimensions. Seed quality assurance clearly reduces farmer risk. Seed Services and the Seed Act require certification of maize seed; however, these institutions faced problems, especially after the land invasions of seed producing farms. Seed producers did use emerging farmers as seed growers, but with mixed success. We noted seed being late to arrive and of mixed standard. With Seed Co and other local firms unable to meet demand, some seed was imported and some provided as aid. Here too, however, despite institutional safeguards, the seed quality has been mixed, and farmers have not always known the seeds' characteristics, which has further increased uncertainty.1

A variety of other institutions also influenced the rate at which new seeds were adopted, in particular extension services, credit providers and agricultural marketing boards. Pre-1980, the Department of Conservation and Extension (CONEX) provided advisory services to white commercial farmers, and the Department of Agricultural Development (DEVAG) provided these services to black smallholders. At independence, these were combined to form the Department of Agricultural, Technical and Extension Services (AGRITEX) (Hanyani-Mlambo, 2002:3). Pazvakavambwa and Hakutangwi (2004:218) claim that DEVAG played a significant role in the expansion of maize production in the 1970s, but that smallholders received less support from CONEX. After independence, with the help of a World Bank loan, AGRITEX further increased extension services to communal areas. However, institutional factors again increased the uncertainties faced by small farmers in the 1990s when conflicting interests within AGRITEX (Pazvakavambwa & Hakutangwi, 2004:231), political instability and reduced investment caused the deterioration in extension services which paralleled the falling adoption rates of new seeds.

In 1978 the largest supplier of short-term credit, the Agricultural Finance Corporation (AFC), launched a campaign targeting communal farmers. This included maize packages made up of seed, fertiliser and insecticide (Mashingaidze, 2004:371). However, the number of defaults on loan repayments forced the AFC to be more selective in its lending. By 1994 they were extending credit to only 3% of smallholders in Zimbabwe, while the value of AFC loans to commercial farmers was three times the value of its loans to smallholders (Eicher, 1995:809).

Until 1992 the Grain Marketing Board reduced risks for farmers by setting an annual pre-planting price for maize. The liberalisation of the maize market with ESAP (Shumba &Whingiri, 2004:588) increased maize price instability, making production decisions (such as seed selection) more difficult. These problems were further aggravated post-2000 by the Grain Marketing Board Act, which required that farmers register and deliver their product within 14 days of harvest. This monopsony power eventually led to grain confiscations and price controls that, combined with payment delays in an inflationary climate, seriously exacerbated the difficulties faced by small-scale farmers (RATES, 2003:33).

The traditional family is often forgotten as an institution, but its importance should not be downplayed. Observation suggests that much of the seed being planted in the rural areas was initially purchased by family members who were in urban employment and who then sent it back to the rural areas. There was preference for commercially produced hybrid seed, and widespread belief in its merits. A recovery of Zimbabwe's urban economy could be a useful catalyst for the recovery of productivity by small-scale maize farmers.

4. Risk and seed technology adoption in Zimbabwe

The process whereby new technologies are adopted in agriculture is naturally of concern to development theorists. We follow the broad approach to risk and technology adoption introduced by Hayami and Ruttan (1985) and Mosley (2002) and explore the sources of risk in the period when productivity stagnation among communal farmers began (1995–2000). We compare this to two earlier periods when productivity was increasing and new seeds were being adopted. The first is 1970–75, pre-independence and also prior to the escalation of the liberation war. The second is 1980–86, immediately post-independence. Risk in the 1995–2000 period is related to the weakening of extension services, a significant reduction in AFC short-term loans leading to declining fertiliser use by communal farmers, and escalating price instability.

Grain Marketing Board deliveries of maize between 1955 and 1965 (the early era of hybrid seed use) provide a useful starting point. Although this period preceded UDI and the sanctions which caused many commercial farmers to switch from tobacco to maize, there was already a widening gap between deliveries from large-scale white commercial farmers and smaller-scale black farmers.

Figure 1: Maize deliveries by African and European farmers to Grain Marketing Board 1955–1965

In contrast, the productivity gap between small-scale farmers and commercial farmers narrowed sharply between the early 1970s and the early 1980s. Table 1 shows average production statistics over the three periods of interest for the three main crops cultivated by smallholders in Zimbabwe: maize, Burley tobacco and cottonseed. The narrowing productivity gap is even more remarkable if we recognise that the ‘commercial sector’ included small-scale farmers until 1981 while the ‘communal sector’ included resettlement areas from 1981.2

Table 1: Mean production, maize tobacco and cottonseed: 1970–75, 1980–86, 1995–2000

	Commercial sector			Communal sector
	1970–75	1980–86	1995–2000	1970–75	1980–86	1995–2000
Area planted to maize ('000 ha)	315	278	185	645	1040	1259
Production ('000 mt)	1296	1069	779	384	865	1059
Yield (kg/ha)	4104	3808	4224	578	836	840
Area planted to Burley tobacco('000 ha)	n\a	1.5	2.2	n\a	1	4.3
Production ('000 mt)	n\a	2.5	3.9	n\a	0.9	4.1
Yield (kg/ha)	n\a	1705	1696	n\a	848	970
Area planted to cottonseed('000 ha)	81	72	47	34	76	250
Production ('000 mt)	126	137	78	27	56	170
Yield (kg/ha)	1590	1894	1629	798	713	658
Source: CSO (2001).

The yield figures for maize among communal farmers increased by approximately 42% between the first period and the second. Although the latter included two poor seasons, the state encouraged production by creating new buying stations and increasing credit facilities. Fertiliser use increased and the R200 series of hybrids was increasingly adopted by smallholders.

During the third period, however, smallholder productivity gains slowed. Although the SC series of hybrids (which could have further increased yields) were commercially available, their adoption rates were lower. Following Bourdillon et al. (2003) and Mudege (2005), we assume that the stagnation in productivity growth was caused by this non-adoption of new varieties, but suggest that the failure of extension services and declining fertiliser access compounded the issue. These factors both increased the risk involved in adopting new seed varieties and reduced the potential for productivity gains where they were adopted.

The cultivation of marginal lands by smallholders could also explain the slower increase in productivity. Table 1 shows that the total area planted increased for all three crops over the three periods, while Figure 2 shows the steady expansion of the extensive margin. Classical (Ricardian) rent theory suggests that the marginal land added will have lower potential. In addition the increase in land planted by the communal sector is partly attributable to the communal resettlement farms which were formerly under commercial maize. While this improved the average land quality in the communal sector, it also lowered the overall productivity of the country's maize land.

Figure 2: Increasing the extensive margin: Land area cultivated (Zimbabwe 1970–2001)

Crop selection can also affect cultivation at the extensive margin; resettled farmers not only had to choose between old or new hybrid maize seeds, but also between OPVs and hybrids, or abandoning maize tobacco or cottonseed. This is reflected in the drop in maize's share of total cropland.

Mudege (2005:74) claims that AGRITEX played an important role in the dissemination of hybrid maize in the period 1980–86. It is difficult to establish how many extension staff were active in each of the three periods on which this paper focuses, but he does note that staffing declined from 2500 in 1989 to 2000 in 2001. Although real funding of extension services rose in the post-independence period, it declined thereafter (see Table 2 for the overall reduction in government agricultural spending). Staffing is necessary but not sufficient for extension work; Mudege (2005:77) describes extension officers equipped with motorbikes, but lacking the petrol to use them. Budget cuts meant that fewer farmers were benefiting from information about the nature, suitability and requirements of new hybrids, which increased the general uncertainty about them and reduced the likelihood of their adoption.

Table 2: Zimbabwean fertiliser use and risk (pre- and post-independence)

	1970–75(pre-independence)	1980–86 (immediate post-independence)	1995–2000
Price instability index^a	9.6	11.5	13.0
Fertiliser consumption (mt)	140	154	170
Fertiliser consumption (kg/ha)	57.4	58.2	52.9
Price of ammonium nitrate fertiliser(nominal Z$/mt)	n\a	306^b	3415
Price of ammonium nitrate fertiliser(constant 2000 Z$/mt)	n\a	98^b	77
AN price: Maize price	n\a	2^b	1.5
AFC short-term loans(constant 2000Z$ million)	1770	4032	2834
Government agricultural expenditure(constant 2000 Z$ million)	3981	5562	2717
Rainfall (mm – national average)	791.88	741.66	773.28
Drought years (<= 500 mm)	1 (1973)	1 (1987)	1 (1995)
^aThe price instability index is defined by UNCTAD (2004:13), *** where y(t) is the exponential trend of prices and Y(t) is the actual price. The index is calculated for each year and then averaged over the specific period. ^bFigure for 1982–86 only.
Sources: CSO (1987, 2001); World Bank (2007); FAO (2009); Zimbabwe Meteorological Services (2009).

Even if information is available, risk is increased and adoption slowed if new technologies require that farmers purchase larger amounts of variable inputs. This risk has both subjective and objective elements. The subjective element can be mitigated by extension services. Bourdillon et al. (2003) describe survey respondents concerned that chemical fertilisers are detrimental to the natural fertility of the soil. While this is a justified concern if the correct precautions and procedures are ignored, it is the role of extension staff to ensure that farmers understand the benefits of fertiliser and how to mitigate their adverse effects.

Some aspects of subjective risk cannot be dealt with so simply. Bourdillon et al.'s survey found some mistrust of AGRITEX agents. Mudege (2005:76) also saw perceptions as problematic, with poorer farmers accusing extension officers of focusing their attention on rich farmers, while the wealthier farmers accused extension officers of not fulfilling their duties. The politically volatile climate also affected perceptions; Bourdillon et al. mention farmers' claims that some AGRITEX extension officers provided poor advice so as to discredit the government. Increasing political risk thus subjectively curbs the effectiveness of extension services and hinders the adoption of new seed technologies.

Objective risks, including uncertain fertiliser supply and price instability, can also influence the behaviour of rational farmers. The consumption of fertiliser has clearly changed over the three periods studied, but we suggest that credit systems can mitigate the risk caused by fertiliser dependence.

Table 2 shows that, while total fertiliser consumption increased over the three periods, fertiliser consumption per hectare decreased; this despite a decline in the price of fertiliser relative to maize. Note, however, that the fertiliser consumption, production and price figures in Table 2 are for the entire agricultural sector, while our primary interest is the behaviour of smallholders. The decline in fertiliser use per arable hectare may be an artefact of land ownership shifting from large commercial enterprises to resettlement farmers who use less fertiliser per hectare.

Figure 3 shows credit, fertiliser supply and fertiliser consumption for 1970–2002. Note that this is not a simple price-clearing equilibrium market: the relationships appear fairly stable until 1990, the year of Zimbabwe's structural adjustment, with supply consistently tracking (i.e. lagging behind) demand. Although a Chow test does not confirm a structural change in the relationship between fertiliser supply and consumption, it is worth noting that the relationship between supply and consumption was statistically significant prior to structural adjustment, but not thereafter.

Figure 3: Short-term credit extension, fertiliser supply and fertiliser consumption

After independence in 1980, the AFC increased its extension of short-term loans to smallholders, allowing them to purchase both hybrid seeds and fertiliser. This facility declined following structural adjustment and rising defaults. Table 2 shows the increased extension of AFC short-term loans in the second period and their reduction in the third. While short-term credit helps to enable the adoption of new and more productive technologies, there are also other factors involved.

Many respondents in Bourdillon et al.'s survey (2003) claimed to avoid planting maize varieties that require more fertiliser because their cultivation was too expensive. However, we see from Table 2 that both the real price of fertilisers, and their price relative to the maize output price, actually declined between the second and third periods. It seems plausible, therefore, that the real issue was uncertainty.

Again part of the problem lay in structural adjustment. Maize price instability followed the liberalisation of the white maize market in 1992. This liberalisation, as well as other destabilising factors in the Zimbabwean economy at the time, made the new series of hybrid seeds less attractive to small ‘high risk’ farmers. In the decade following, the accelerating inflation of the late 1990s government price controls combined with unevenly controlled nominal maize prices to further increase real maize price volatility (see Makamure et al., 2001:24). This amplified the uncertainty faced by farmers who could not be sure, when planting, that the post-harvest maize price would cover the following season's input costs.

In summary, the productivity stagnation of the 1995–2000 period accompanied, and may be explained by, state policies that increased risk and thereby hindered small farmers' adoption of improved seed technologies. Initially the policies were ESAP related: declines in the effectiveness of extension services, reduced short-term credit extension, and increasing price instability. Subsequently the negative impacts imposed from abroad by ESAP were overtaken by the damage done by domestic policies that generated high and unstable inflation rates, uncertain access to agro-chemicals, and uncertain supplies and characteristics of seed. With the first decade of the millennium these issues became increasingly significant.

5. Conclusion

We have argued that an increase in the risk environment faced by small-scale farmers in the late 1990s led to stagnating smallholder productivity. This, combined with rapid population growth, had profound implications for poverty, food security and the viability of land reform in Zimbabwe.

Recent literature on Zimbabwe's maize industry attributes its 1980–86 output surge to a ‘green revolution’ in which small-scale farmers learned to use new seed types, fertiliser and so on. In this paper we have made four principle points:

a.	This was not something new: CSO agricultural census data show that hybrid seed adoption by small-scale farmers had already been under way for decades beforehand. Moreover, willingness to adopt more productive seed types was already a norm: high-yielding imported OPVs such as Hickory King had been rapidly adopted when introduced over 80 years earlier.
b.	The output surge coincided with the end of the bush war and a post-independence era in which socioeconomic uncertainty decreased. We suggest that this correlation was causal. In consequence, when economic problems forced the government to adopt a structural adjustment package and levels of uncertainty facing farmers rose, adoption of new seed technologies first slowed and then reversed. Over this post-1990 period yields first stabilised and then fell.
c.	The development lesson, if we extrapolate from the Zimbabwean experience, is not that farmers need incentives to stimulate them, but that they need security (in particular price security), information and reliable infrastructure.
d.	The demand for hybrid maize in Zimbabwe resulted from residual research and production capacity, wide availability and economic stability. When politics and policy rendered the currency unstable and the economy uncertain, farmers responded rationally by seeking OPVs. With economic stability and money again a store of value, rationality should see a return to HYVs.

Zimbabwe's small-scale farmers have not adopted new seeds at the rates expected, and indeed many have opted to return to OPVs and saved seeds. It is easy to ascribe this to delivery failures. Supply has certainly been a problem; in recent years seeds have often been late on the market and difficult to obtain. Our argument is that there is a further dimension to this outcome. Small-scale farmers are rational and risk averse. New generation HYVs may offer the prospect of higher returns, but when inflation is high, maize prices unstable, fertiliser supplies uncertain and seed quality uneven, these returns also entail increased uncertainty. Reverting to lower yielding OPVs can then be a rational production strategy.

Raising the productivity of Africa's farmland remains one of the unrealised challenges of development economics. Sub-Saharan Africa is a fertile continent and one with low population densities by world standards; yet it remains a net importer of food. The history of Zimbabwean smallholder maize farming shows that while agronomic training and provision of extension information are necessary conditions, they are far from sufficient if land productivity is to be raised. The rational entrepreneur evaluates probabilities and takes considered risks, but seeks to avoid uncertainty. Nature already makes farming in Africa a risky pursuit, but what confounds the farmer is the economic uncertainties endemic to so many of its countries. Like Zimbabwe, they suffer inflation, unstable product prices, uncertain property rights, unreliable input supplies, imperfect markets, unstable infrastructure – each factor increases uncertainty and makes it more rational for farmers to retreat to survivalist practices, saving seed each season, minimising the use of commercial inputs such as fertilisers, and targeting an output that meets their needs rather than maximising their income. In a continent so profoundly dependent on agriculture it is especially important that macroeconomic strategies, such as structural adjustment policies, should not worsen these uncertainties in the name of the free market.

Notes

¹An interesting point was made by an anonymous referee: in the 1990s the seeds available to smallholders in Zimbabwe were produced by commercial farmers located on high quality land. Such seeds may have been failure-prone in drier and less fertile communal areas.

²Throughout this discussion ‘year’ is the year of harvest. The agricultural year for maize runs from July to June.

Appendix

Figure 1A shows the declining trend in Zimbabwean maize land productivity. Some factors are clearly exogenous, and the impacts of severe drought years (such as 1982, 1992 and 2001) on short-term land productivity are clear to see. The data aggregate high productivity large commercial farms and lower yielding small farmer operations, which accounts in part for the decline in yield once land invasions began in 1998. Nonetheless, the sustained low yield of the post-millennium period cannot be attributed to falling yields on large farms alone; though historically they provided the bulk of the marketed output, such farms only accounted for approximately 30% of the area planted under maize. For a detailed analysis see USDA (2009).

Figure 1A: Zimbabwe estimated national mean maize yield per hectare, 1960–2008

Figure 2A shows that the total area under cultivation was dropping as these farms went out of production, suggesting that the productivity of smallholders was also declining. The trend line indicates that overall maize land productivity began falling soon after independence. However, it should be remembered that this was a period when the tobacco industry was reviving.

Figure 2A: Maize production and area harvested: 1980–2008

Figure 3A shows that high productivity commercial farmers were moving from the maize planted during the sanctions period to cultivating tobacco, which was more profitable. That the maize yields remained as high as they did is an indicator of the initial rise in small farm and communal land productivity. The yield volatility shown in Figure 3A is largely rainfall related. The early and mid 1960s, 1982, 1991 and 2001 had poor seasons. However, althrough rainfall was back to normal in the seasons from 2003–2006, yields failed to recover.

Figure 3A: Zimbabwe, area planted to tobacco 1981–2003