Depending on the trickle-down effect? The danger of a growth-led strategy for poverty reduction

After four years implementing poverty-combating policies, we can now conclude whether the assumptions made in Rwanda's PRSP-1 were realistic. Targets for annual growth per capita in the first Rwandan PRSP were set at 4–5% for the next 15 to 20 years. This implies 7–8% per annum overall real growth (GoR 2000). Projections were, however, later reduced to 6–7% for the PRSP-1's 2002–05 implementation period (GoR 2002). In subsequent PRSP progress reports and IMF statistical documents, growth projections were typically set at about 6%. Two important questions then arise. First, have these growth expectations been realistic. Second, to what extent did growth translate into poverty reduction?

These ambitious projections for the first PRSP implementation period seemed to be justified based on Rwanda's solid post-civil-war economic recovery, with average annual growth of 8·6% between 1996 and 2001. There are, however, some critical observations to add. Figure 1 shows how the average period growth rate is influenced by exceptional growth figures in the first years (i.e. the steep slope), which moderate a few years later. From a longer-term perspective, economic performance has still not reached mid-1980s levels. Moreover, Rwanda benefited from the receipt of substantial aid funds, which significantly exceeded the sub-Saharan African average (Ansoms Reference Ansoms2005). The country also benefited from financial transfers out of the DRC during the years of Rwanda's military involvement there (Cassimon & Marysse Reference Cassimon, Marysse, Marysse and Reyntjens2001). It seems likely that these capital inflows fuelled Rwanda's exceptional economic recovery in the immediate post-war period.

Figure 1

Rwanda: evolution of GDP per capita.

Source: World Bank 2006b.

In more recent years (i.e. 2003–06), annual growth did not meet the projected rate of about 6% (IMF 2005b, 2006b). This recent trend illustrates the vulnerability of Rwanda's economy to structural limitations, including overpopulation, resource scarcity and a limited potential for economic diversification. Further, overall growth targets for the coming years were lowered to between 4% and 4·5%. This would dampen per capita growth to a modest 2·1–2·6%,Footnote ³ far below the rate of 4–5% targeted in PRSP-1.

With reference to the above, the first issue to consider is whether growth will be substantial and sustainable enough in the near future, or whether current economic trends will signal the onset of more temperate times. The second principal issue is the degree of economic growth's ‘pro-poorness’. There are two dominant views on what constitutes pro-poor growth (Page Reference Page2006). One definition highlights the importance of reducing inequality by defining pro-poor growth as growth that disproportionably benefits the poor (Kakwani & Pernia Reference Kakwani and Pernia2000; Klasen Reference Klasen2003). Another definition regards economic growth as pro-poor when the living conditions of the poor improve in absolute terms, and poverty thus decreases (Kraay Reference Kraay2006; Ravallion & Chen Reference Ravallion and Chen2003). The main issue then is how much growth is pro-poor. This can be measured in several ways, for example with the country's growth elasticity of poverty index.

Cross-country evidence situates the average growth elasticity of poverty within the interval −2 and −3. This implies that positive (or negative) growth of 1% should lead to a 2–3% decrease (or increase) in the incidence of poverty, as measured by the percentage of people living below the poverty line of US$1 PPP per head (Adams Reference Adams2004; Ravallion Reference Ravallion2001; Ravallion & Chen Reference Ravallion and Chen1997; World Bank 2000). Adams found that this elasticity may differ for individual countries, depending on their initial inequality levels. Countries with higher inequality levels (i.e. Gini >0·4) have lower poverty elasticity rates and vice versa.

Turning to Rwanda, the growth elasticity of poverty for the recent post-conflict period is not very promising in comparison with other developing countries. In the immediate post-genocide period (1994–2000), each percentage point of economic growth led only to a 0·37% decrease in the incidence of poverty, this is an elasticity of −0·37 (Ansoms Reference Ansoms2005). Rwanda's case thus seems to be a clear example of the highly negative impact of inequality upon the pro-poor effect of growth (GoR 2002). The high inequality rate of 2001 (Gini of 0·451) contrasts dramatically with the rate in the mid-1980s when Rwanda qualified as a low-inequality country, with a Gini of 0·289 (Ansoms Reference Ansoms2005).Footnote ⁴

Moreover, inequality has further increased over the PRSP-1 implementation period, with the Gini reaching 0·51 in 2006. For the same period, the incidence of poverty decreased from 60·3% to 56·9% based on the national poverty line, which is different from the US$1 PPP per day poverty line (GoR 2006a).Footnote ⁵ In combination with an average annual growth of 4·6%, this results in a growth elasticity of poverty of −0·40. Although this figure is not comparable with the cross-country average (due to the difference in poverty lines), the result is clearly disappointing. The pro-poor character of Rwandan economic growth is thus extremely weak, despite the implementation of the PRSP policy.

Overall, the disappointing derived effect of post-conflict growth on poverty incidence, in combination with more moderate growth projections for the coming years, tempers the potential of a successful growth-reliant strategy for poverty reduction in Rwanda.

The agricultural sector as a growth engine in PRSP policy

Even more challenging is the PRSP's ambition to transform the rural sector into an engine for growth. Over decades, development theory has been influenced by the presumption of the need for the structural transformation of an economy to achieve modern economic growth. As stated by Kuznets (Reference Kuznets1973: 248),Footnote ⁶ ‘(these) major aspects of structural change include the shift away from agriculture to non-agricultural pursuits … with a corresponding change in the occupation status of labour’, i.e. ‘changes in the distribution of the labour force between agriculture and the non-agricultural production sectors’. Based on Western experience, less-developed countries were pushed to strive for economic emancipation through the modernisation of their own economies with a decreased reliance on the primary sector (i.e. agricultural activities).

However, the African experience has been characterised by decades of unfruitful attempts to shift away from the agricultural sector.Footnote ⁷ More recently, with the ‘fight against poverty’ at the forefront of the international agenda and due to high rural poverty rates, the need for rural-led development and economic growth has resurfaced in popular development theory (see Mwabu & Thorbecke Reference Mwabu and Thorbecke2004). As a result, appreciation of the agriculture sector's importance has returned, though the continued need for households to diversify their incomes by shifting away from pure subsistence agriculture towards other activities, both in the farm and non-farm economy, is still recognised (see, e.g. Abdulai & CroleRees Reference Abdulai and CroleRees2001; Yaro Reference Yaro2006).Footnote ⁸

Indeed Rwanda's PRSP also highlights the crucial importance of the rural sector for the country's economic future. The agriculture and livestock sector are presented as ‘the primary engine of growth’, though the document also stresses the importance of finding other new growth engines (GoR 2002: 30). PRSP-1 projections for agricultural performance were ambitious, as ‘primary growth is predicted to start at 5·2% and accelerate over the period’ (GoR 2002: 75). This estimate corresponded with projections made by Mellor (Reference Mellor2002a), who predicted that 75% of this growth would be due to improved fertiliser use, 16% to more intensive farming and 9% to the swamp reclamation programme.

However, between 2002 and 2004 agricultural activity stagnated and even contracted (Table 1). These statistics probably underestimate the sector's poor performance, given that estimates provided by the Minagri/Food Security Research Project (FSRP – available only for 2000–02) are significantly lower. The FSRP's statistics seem to be more appropriate for measuring food production than national account data (IMF 2004b).Footnote ⁹ The most straightforward explanation for contractions in agricultural activities is the impact of poor weather on food production. As mentioned in the PRSP progress report, growth in agricultural output is largely ‘at the mercy of good weather’ (GoR 2004a: 17). Agricultural growth between 2004 and 2005 again met the PRSP target; however Rwanda was once again affected by bad weather in 2006 and early 2007.

Table 1

Agricultural growth in Rwanda, 2001–05

Source: GoR 2005a: 113. For a recent update, see Ruzindaza Reference Ruzindaza2006.

These figures clearly illustrate the failure of the first PRSP strategy to transform the agricultural sector into a stable engine for growth. The poor performance might also be due, in part, to weak budgetary commitments to the rural economy (see Table 2), where two dominant problems emerge. First, agriculture-related spending represents only 2–5% of the priority budget (actual figures) between 2002 and 2005.Footnote ¹⁰ The share of agriculture-related spending is small in comparison to the financial commitments directed, for example, to tertiary education (about 14% of the 2002–03 priority budget). These figures indicate how the Rwandan government presents spending targeted at the urban elite as pro-poor priority expenditure. A second major problem lies in the low absorptive capacity of the agricultural sector, illustrated by the discrepancy between budgeted and actual spending in 2003 and 2004. Despite some improvements towards 2006, agricultural activity and the broader rural sector in general have not received sufficient budgetary commitments. It is not surprising therefore that the rural economy did not meet PRSP forecasts.

Table 2

Agriculture-related priority expenditures, 2001–07

‘Act’=actual expenditures, ‘Budg’=budgeted expenditures.

Sources: Purcell et al. Reference Purcell2005; GoR 2006b; GoR 2007b.

How pro-poor is agricultural growth?

There seems to be increased awareness in Rwanda's second poverty reduction strategy (EDPRS), now being prepared, of agriculture's importance as ‘the pillar for rural development’ (GoR 2004b: 1). During interviews in early 2006, EDPRS ministerial stakeholders expressed their hopes that the new document would accentuate the rural sector, and transform it into a real engine for growth through agricultural transformation.Footnote ¹¹ The goal then is to realise the ambition of the first PRSP, ‘because Rwanda's growth strategy is based on agriculture, it is specifically designed to be pro-poor’ (GoR 2002: 31).

But is it naïve to assume that growth, per se, even when sought in the rural economy, would be pro-poor? Growth in the agricultural sector does have a greater impact on poverty reduction than growth in other sectors (Christiaensen & Demery Reference Christiaensen and Demery2007; Gallup et al. Reference Gallup, Radelet and Warner1997). Moreover, agricultural growth spills over to other activities in the rural economy (Irz et al. Reference Irz, Lin, Thirtle and Wiggins2001; Thirtle et al. Reference Thirtle, Irz, Lin, McKenzie-Hill and Wiggins2001). But the impact of agricultural growth on poverty reduction depends on the extent to which the poor participate in this growth. One World Bank study (2005) found that the participation of poor rural households in agricultural growth could differ greatly depending upon the local context. They identify several policies that can improve the pro-poor character of agricultural growth, all relating to improving the institutional environment of smaller and poorer producers, such as access to markets, technology and risk-coping mechanisms. Other studies emphasise the importance of a relatively equal distribution of assets, particularly land, to achieve an optimal pro-poor growth effect (Deininger & Squire Reference Deininger and Squire1998; de Janvry & Sadoulet Reference De Janvry and Sadoulet1996; Ravallion & Datt Reference Ravallion and Datt2002). A recent (2006) OECD report highlights the importance of small-scale agriculture, with its potential to create a win-win outcome for economic growth and poverty reduction.

The potential of the economy's sectors to reduce poverty through growth can be measured by comparing the growth elasticity of poverty for each sector. The sector's elasticity multiplied by the sectoral share of GDP gives the participation effect for each sector to overall poverty reduction. Christiaensen & Demery (Reference Christiaensen and Demery2007) have estimated both effects for sub-Sahara African (SSA) low-income countries (Table 3, SSA-low income results) from the following equation:

where π_x is the elasticity of poverty of sector xFootnote ¹², s_x is the share of sector x in GDP, and π_x*s_x is the participation effect.

Table 3

Growth elasticity of poverty and participation effect for different sectors

‘Agr’=agricultural sector; ‘Ind’=industrial sector; ‘Ser’=service sector.

* GDP shares for Rwanda are based on the average between 1994 and 2000, and the average between 2001 and 2005.Footnote ²⁷

** The growth elasticity of poverty and the participation effect for Rwanda 2001–06 are not comparable with the other figures, given that a different poverty line has been used to calculate these figures.

Sources: For SSA data, Christiaensen & Demery Reference Christiaensen and Demery2007; for Rwandan data, Ansoms Reference Ansoms2007.

Similar calculations for Rwanda (Table 3) show that the growth elasticity of poverty for the agricultural sector is low; this means that the correlation between agricultural growth and poverty reduction is limited. For the period 1994–2000, the elasticity is almost three times lower than that of the average SSA low-income country. Nonetheless, Rwandan statistics for this period are consistent with cross-country findings, to the extent that agricultural growth is correlated with considerably more poverty reduction than growth in the secondary and tertiary sectors. However, for the period 2001–06, the situation seems to have worsened. Though these figures cannot be compared with other Rwandan and SSA statistics (as a different poverty line has been used for the calculations), we can nonetheless observe that the elasticity and participation effect of the Rwandan agricultural sector are very low, and are no longer significantly different from those of the other two sectors (Ansoms Reference Ansoms2007). It matters little whether growth is realised in the agricultural or in the other two sectors: the overall correlation with poverty reduction is weak.

Overall, the pro-poor impact of agricultural growth over the PRSP-1 implementation period is thus problematic. This can be explained by a low participation of the poorer rural categories in agricultural growth. As we show later in this paper, there are many institutional constraints that Rwandan small-scale peasants face (e.g. the lack of access to markets, credit and risk-insurance opportunities, and fertilisers). Further, Rwanda is characterised by a high degree of land inequality (Table 4), and small-scale peasants have been increasingly marginalised in terms of land ownership over the past decades. According to Jayne et al. (Reference Jayne, Yamano, Weber, Tschirley, Rui, Chapoto and Zulu2003), average land availability has strongly declined for all quartiles between 1990 and 2000, except for the richest. In line with cross-country evidence (Deininger & Squire Reference Deininger and Squire1998; de Janvry & Sadoulet Reference De Janvry and Sadoulet1996; Ravallion & Datt Reference Ravallion and Datt2002), the Rwandan example thus illustrates how highly unequal distribution of land holdings contributes to the weakening of the linkage of land-poorer groups to agricultural growth.

Table 4

Land distribution

Gini 1 is defined in terms of land per household, Gini 2 in terms of land per capita, and Gini 3 in terms of land per adult.

Source: Jayne et al. Reference Jayne, Yamano, Weber, Tschirley, Rui, Chapoto and Zulu2003: 262.

Agricultural policy

Current Rwandan agricultural policy (the National Agricultural Policy or NAP) was elaborated and then operationalised in the 2004 Strategic Plan for Agricultural Transformation (SPAT). Both documents serve as blueprints for the elaboration of EDPRS policy. The NAP's global objective is ‘to create conditions favourable to sustainable development and promotion of agricultural and livestock produces, in order to ensure national food security, integration of agriculture and livestock in a market-oriented economy and to generate increasing incomes to the producers’ (GoR 2004a: 11). This policy outline has been translated into action plans in the SPAT document, which are to be realised over a year-year period, starting with a pilot phase in 2005 and followed by a three-year implementation period. It aims to transform the agricultural sector from a subsistence production orientation towards a professional, commercial and competitive economic activity. The SPAT document (Table 5) focuses on four priority programmes sub-divided into 17 sub-programmes (GoR 2004b).

Table 5

SPAT strategy and budgetary priorities

Source: Rutagwenda Reference Rutagwenda2006 (an earlier indicative budget was published in GoR 2004b).

On different occasions, SPAT refers to its mission to improve the living conditions of the rural poor by guaranteeing that ‘different categories of agricultural farmers, especially the most vulnerable, benefit from the economic growth that is being advocated’ (GoR 2004b: 7). However, of the seventeen sub-programmes only one (SP16 – food security, management of risks and vulnerability) has a clear pro-poor character. This sub-programme focuses on reducing food and nutritional deficits, reducing vulnerability in food deficit zones and affected population groups, and creating massive employment through labour-intensive works targeting vulnerable categories.

There seems to be a clear ambition to satisfy the population's food needs with national production.Footnote ¹³ However, the approach advocated focuses on the supply side by striving for maximum output growth, instead of concentrating on how vulnerable groups will take part in creating this growth and will thus acquire the necessary purchasing power to access food supplies. Little attention is given to preventing those processes or events that increase the economic vulnerability of these peoples' lives.

Also problematic is the conceptualisation of the term ‘vulnerable groups’. This term appears in both SPAT and the PRSP to relate to female or child-headed households, genocide survivors and demobilised/resettled households (GoR 2002; GoR 2004b). This interpretation, based on gender or war-related identities, is very restrictive and disregards the multidimensionality of vulnerability in the rural context. As a result, there is no effort to identify other vulnerable groups (e.g. nearly landless peasants), their current challenges and needs, and how rural development strategy might impact on or improve their living conditions.

The remainder of the SPAT document focuses on agricultural modernisation, intensification, professionalisation and enterprise development to transform the primary sector into a growth engine. Growth is expected to emerge from two sources, ‘those which are linked to export potential within the commodity chains and those which are related to internal market development’ (GoR 2004b: vii). The commodity chains to be promoted include maize, rice, and traditional export crops such as tea and coffee, exactly the crops where SPAT foresees a major private sector role (ibid.: 20–1). SPAT also puts effort into developing integrated livestock systems and exploiting opportunities for agribusiness (e.g. fruit processing enterprise) (ibid.: 39).

The SPAT strategies seem to be tailormade for larger farmers whose farm structure and risk-coping abilities allow them to invest in new, high-potential production systems. However, access to these modernised and professionalised techniques seems less straightforward for risk-averse small peasants. The SPAT document reflects two somewhat different views of smallholders' capacities to transform their agricultural production systems. Some parts recognise the constraints that small peasants face. The document, for example, refers to the lack of credit,Footnote ¹⁴ and the inability of smallholders to insure themselves against shocks and setbacks. However, it does not develop a strategic plan to ensure access of these small-scale peasants to the modernised production techniques that rural policy-makers promote. Other parts of the document mention the ignorance and resistance of peasants to adopting recommended productivity-enhancing measures that go beyond traditional subsistence farming (e.g. see GoR 2004b: 6, 17). As such, their lack of capacity to embrace ‘modernised’ farming is attributed to a ‘wrong mentality’ problem, a view that disregards the institutional barriers these peasants face.

In fact, Rwanda's rural policy has the ambition to reduce the agriculture-dependent population to 50% by 2020, considerably less than today's 87% (GoR 2004a, 2004b). The plan foresees that a growing primary sector will then ‘create progressive development of secondary and tertiary sectors in rural areas, which could help create employment outside agriculture’ (GoR 2004b: 59). The land policy takes it further: ‘the Rwandan family farm unit is no longer viable. … The re-organization of the available space and technological innovations are necessary in order to ensure food security for a steadily and rapidly increasing population’ (GoR 2004c: 16). This process is described by Alison Des Forges (2006) as the government's ambition to ‘winnow out the chaff’.

SPAT, in terms of budgetary commitments, also mentions how ‘allocation of government financial resources will be done with priority towards most competitive actions and productions’ (GoR 2004b: viii). This objective is translated into budgetary planning (Table 5). The ‘pro-poor’ sub-programme (SP16) represents only 1% of the total SPAT budget, clearly not the first priority in financial terms. Next to one-quarter of the budget allocated to management costs, the top priorities for the coming years are the promotion of fertilisers and mechanisation, and the promotion of research for agriculture and livestock development. The main objective of the fertiliser strategy is to increase chemical fertiliser use from 10 to 42 kg per hectare per annum over the next five years, which could, indeed, also reach small-scale peasants. However, further analysis of constraints on fertiliser use is necessary. By putting the blame on peasants' supposed ignorance regarding the profitability of fertiliser use, other institutional barriers are ignored (e.g. their lack of purchasing power, and the limited or non-availability of access to credit and insurance mechanisms to overcome setbacks).

Land policy and land law

The recently adopted land policy and law show a similar commitment in favour of competitive and commercial farmers. The land law was adopted in 2005 after a long process of drafting and negotiation,Footnote ¹⁵ and seeks to formalise land rights through official titling. During the elaboration phase of the land law, the PRSP stated that ‘the design of the land policy to encourage security of tenure is central’ (GoR 2002: 36). The new land law aims to break with a past of informal land arrangements and transfers built upon customary traditions. Although customary land rights are recognised as a basis for acquiring official rights, land registration is made compulsory; and in the future, land arrangements are to be regulated through formal legal procedures (article 26). The Rwandan government hopes that secure official land titles will encourage increased investment in land conservation and quality improvements (GoR 2004c: 24). However, in line with the new law, official titles can only be acquired through a formal procedure of registration with proof in the form of a certificate. More privileged groups have a clear advantage to use this as an additional tool in their ‘struggle for land’. Studies that largely praise the beneficial effects of official land registration recognise the problem of unequal access to information in the registration process, ‘the introduction of a modern registration system to replace a customary (and typically less formal) system may provide opportunities for “land grabbing” by those who are better informed, are more familiar with formal processes, and have better access to officials and financial means to undertake procedures for registration’ (Feder & Nishio Reference Feder and Nishio1998: 38).

The land law further aims to solve the problems of land fragmentation and unproductive use. Fragmentation of land holdings has since long been conceived as a major problem by Rwandan policy-makers. The previous land policy (dating from March 1976) aimed to counter this by only allowing land transfers (with specific permission) when the seller's property remained at a minimum of 2 hectares, and when the buyer had no more than 2 hectaresFootnote ¹⁶ (Décret-Loi n°09/76, art. 2–3, March 1976). Given that average land holdings in the 1970s were about 1·4 hectares per household, this policy aimed to redistribute land by restricting transfers from the land-poor to the (relatively) land-rich. The question is, however, to what extent this policy was followed. The informal land market which emerged during the period to arrange the transfer of land titles, even with written documents used as validation, almost never followed formal policy prescriptions (Platteau Reference Platteau2000).

The new land policy tackles the problem of land fragmentation in a very different way. Article 20 prohibits dividing land parcels of 1 hectare or less. For the division of plots between 1–5 hectares, the owner has to apply to the land commission for authorisation. However, this rule does not apply to cases where authorities ‘approve the consolidation of small plots of land in order to improve land management and productivity’ (article 20).Footnote ¹⁷ The Rwandan government thus aims to consolidate small parts of land into larger plots, and to consolidate land into the hands of fewer, more efficient farmers. The land law, for example, sets no upper limits on the maximum size of landholdings. A ceiling of 50 hectares foreseen in an earlier version of the new land law did not appear in the final approved version. The objective behind this choice appears clear: create economies of scale. Larger plots would become suitable for more modern intensive techniques. Larger farms could be managed more productively and become professional partners with agribusiness concerns. (In the next section, we will question these assumptions.)

A major question is whether land fragmentation is indeed a problematic issue in the Rwandan context. Pre-war evidence for Rwanda showed that land fragmentation could be advantageous for farmers' risk management. Based on this finding, Blarel et al. (Reference Blarel, Hazell, Place and Quiggin1992) questioned the beneficial impact of consolidation on land productivity, and even concluded that concentration-promoting programmes could have a negative impact on farmers' well-being.Footnote ¹⁸ Likewise, Pottier (Reference Pottier2006) challenges the usefulness of consolidation strategies, which he sees as a potential source for future land conflict.

Taking this into consideration, will the prohibition to divide plots of less than 1 hectare have serious consequences for the majority of Rwandan peasants? The average total land surface occupied by rural households is well below 1 hectare (in 2000 this was 0·71 hectares, spread over an average of 2·44 plots). By 2002, landholdings had become even more fragmented (calculations based on Food Security Research Project agricultural dataset, 2000–02). By the letter of the law, small-scale landholders in times of setbacks would only be allowed to sell their integral plot so as to avoid further land fragmentation. Further, their chances of buying back land would be diminished, as they would have to buy either an adjacent plot or a plot of 1 hectare in total.

The new land law includes another ‘guarantee’ for sound land management by giving the authorities the right to ‘impose sanctions … against the landlord or any other person allowed to lease the land who fails to respect the obligation of efficiently conserving the land and productively exploiting it’ (article 73). Productive land use, appropriate protection and sustainable productivity mean, ‘to protect it (i.e the land) from erosion, safeguard its fertility and ensure its production in a sustainable way’ (article 62), and ‘shall be based on the area's master plan and the general structure on land allocation, organization and use and [the adoption of] specific plants certified by relevant authorities’ (article 63).

The goal would be for each region to specialise in certain specific crops based on agro-bio-climatic conditions and in accordance with market needs. This is a return of the 1970s argument to consider the comparative agro-climatic advantage of each region as a main element in agricultural planning (see Pottier & Nkundabashaka Reference Pottier, Nkundabashaka, Parkin and Croll1992). In accordance with current policy the local authorities, in the name of the local peasants, will determine in which crop(s) the region has a comparative advantage. In SPAT, a pilot exercise with a few districts resulted in the identification of three agricultural products per district. The document mentions the need to guide producers in their choice towards commercial production systems and away from subsistence agriculture (GoR 2004a). This seems to be a first indication that the participation of peasants in the choice of those crops would be limited.

The strategy of specialisation could however indeed result in economies of scale in terms of production, and increase the commercial bargaining position of local farmers in regional markets. On the other hand, the strategy should regard possible variations in soil types and climatic conditions within the local setting. Forcefully restricting farmers to a few crops at an aggregate scale might thus make little sense. An additional concern is whether small-scale, non-commercial peasants will be able to confine themselves (even partly) to the prescribed crops, as they usually opt for a diversification in crop types based on risk-averse considerations. If the new policy does not offer them any additional risk insurance, they will not be inclined to go for crop specialisation. Another important question is whether small-scale peasants will be able to defend their interests on the regional markets, or whether power positions in the bargaining process over food prices will be occupied by intermediary traders.

In some regions, peasants are already obliged by the authorities to abandon and even destroy certain food crops. For example, in early January 2007 the governor of the Eastern Province, Mr. Mutsindashyaka, initiated a ban on sweet potatoes. Although the minister of state for agriculture later withdrew this declaration (New Times 2007), such campaigns can and generally do cause a lot of uncertainty and fear among local peasants.

When land is not effectively conserved and productively used, or in more specific terms when it is degraded or has not been used for three consecutive years, the land law provides for sanctions. These typically take the form of requisitioning the land for a period of three years (article 74). Local authorities are delegated extensive powers over managing, requisitioning and even reallocating land: ‘The requisitioned land may be entrusted to another person who so requests and who demonstrates ability to efficiently conserve the land and productively exploit it’ (article 74). Further, in cases of dispossession, the owner can only request repossession in writing, explaining how he or she will commit him/herself to the productive exploitation of the plot in question. When rejected, the only further option is to appeal to court. The formality of these procedures typically leaves little room for illiterate peasants with limited means to pursue their cases.

This legislation gives local authorities considerable freedom in interpreting specific situations according to their own agendas. Pottier (Reference Pottier2006) points out that this might turn out positively for the population when local government agents use their power and discretion to honour people's rights. However, it may also result in the abuse of local power to favour the rights of well-placed individuals.