Study site description

Mukuyu and Munyuki are located in Likuyani and Lugari divisions respectively of Lugari District, Western Province. Nearly 90% of the district population resides in rural areas, with about 60% of it classified as poor (Andersson & Wambugu Reference Andersson and Wambugu2007).

Average land size has been falling since the mid 1990s, as in Kenya as a whole, but is greater in Mukuyu than Munyuki, with 1·5 and 1·1 hectares per household respectively in 2008. In the latter village the land frontier was already exhausted in 2002 (Karugia Reference Karugia2003). Land distribution was relatively equal in 2002, but inequality has increased in both villages since then, with such increases being higher in Mukuyu.Footnote ³

Population densities are, however lower in the two villages than for most parts of the district. Soil fertility is good and average annual rainfall is relatively abundant, ranging from 1,200 to 1,400 mm annually. Both villages are located in the agro-ecological zone UM4 (Jaetzold & Schmidt Reference Jaetzold and Schmidt1983; Karugia Reference Karugia2003; Kenya 2000). Nonetheless, rainfall was reported to be decreasing over time, with low rainfall already being considered a constraint to agricultural production in 2002 (Karugia Reference Karugia2003).

Maize was grown by 95% of the households, compared with 98% of smallholder households in the country as a whole (Kirimi et al. Reference Kirimi2011). Cropping patterns are based mainly on maize, in combination with beans and sweet potatoes and to a lesser extent vegetables, fruits, coffee and sugar cane. As in the rest of the country, tubers and potatoes constitute an important complement, serving primarily as ‘consumption absorber’ crops in relation to maize (ibid.). Hence maize is the most important crop in the villages, especially in relation to households' own production, but also as an important source of income. In this sense, maize takes on the double role of cash and food crop. Average annual household maize production fell dramatically between 2002 and 2008, from a three-year average of 3,060 kg (2000–2) to 2,237 kg (2006–8). In part, this fall in production may be explained by the post-electoral violence which plagued the countryside in many parts of Kenya in early 2008, but it is also likely to be connected to rising man/land ratios and increases in land fragmentation in Kenya generally, as noted in the literature (Jirström et al. Reference Jirström, Andersson, Djurfeldt and Djurfeldt2010; Kirimi et al. Reference Kirimi2011).

Settlement schemes in both villages were part of the Million Acre Settlement Scheme of the early 1960s. In terms of ethnic composition, traditionally a number of Abaluhya sub-ethnic groups have dominated Lugari District (Curtis Reference Curtis1995; Gwako Reference Gwako and Ensminger2005). Since the mid 1970s, in-migration from the Central and Rift Valley provinces has changed the ethnic structure of the population. Relatively large numbers of the sub-ethnic group of Maragoli Abaluhya have moved from the densely populated neighbouring district of Vihiga. The creation of land-buying companies among in-migrants has enabled land acquisition by migrants in the two villages (Andersson Djurfeldt & Wambugu Reference Andersson Djurfeldt and Wambugu2011). Land is held under private freehold tenure, but the land tenure system in practice discriminates against women in terms of acquisition, ownership and inheritance of land, as custom dictates that land is vested in men (Karugia Reference Karugia2003). Seventeen per cent of the households in Munyuki were headed by women in both 2002 and 2008. The corresponding figure for Mukuyu was 20% in 2002 and 23% in 2008.

Average yearly total household cash income in 2008 was US$502, with a minimal difference between the villages.Footnote ⁴ Five households did not earn any cash income. Munyuki had a more diversified agrarian income base, but on the whole market accessibility was relatively favourable in both villages, with an average distance of 2 km to the closest market outside the village.

Production and market-related seasonality

The Kenyan maize market is highly concentrated. Although liberalised since the mid 1990s with the dismantling of the purchasing monopoly of the state-run National Cereals Produce Board (NCPB), the government continues to exercise substantial influence over markets through trade-related measures and control over storage facilities (Jayne et al. Reference Jayne, Zulu and Nijhoff2006b). Nonetheless, the role of the NCPB in the market has generally become clearer, especially since the change of government in 2002 (Kirimi et al. Reference Kirimi2011).

The NCPB accounts for about 25–30% of maize purchases, most of which is sourced directly from large-scale farmers. Small-scale farmers, in a survey carried out between 1996/7 and 2003/4, sold 96% of their maize either to private traders/brokers or directly to other households. Only 4% of smallholder maize was sold to the NCPB (ibid.: 16–18). While market participation increased from 35% to 49%, the amounts marketed declined, with an average proportion of production sold at about 45% per household (ibid.: 29).

Similar trends can be discerned both in the quantitative data and in the interviews with farmers. In Munyuki, about half (53%) of the households reported selling maize, and in Mukuyu 70% sold maize following the harvest in late 2007. On average, 24% of the maize harvest was sold in Munyuki compared with 31% in Mukuyu, considerably lower than the figures reported by Kirimi et al. (Reference Kirimi2011).

A look at the highest and lowest prices received in the village reveals large differences in price: the highest price received for 100 kg of maize in Munyuki was nearly four times the lowest price (US$44 compared with $12). Prices in Mukuyu were generally lower, but so were the price differences – the highest price was US$33 for 100 kg of maize, compared with a low of $10. The literature suggests that many poor households are forced to sell their produce in the immediate post-harvest season to cover pressing cash needs (Kosura-Olouch & Karugia Reference Kosura-Olouch, Karugia and Djurfeldt2005). For this reason the highest prices in the village may not be available to the poorest sellers – a result of both distress sales and segmented markets.

Kenyan maize prices in normal years are at their lowest in November and December, with peak prices in May and June (Kirimi et al. Reference Kirimi2011; Tschirley & del Castillo Reference Tschirley and del Castillo2007). In Lugari, harvests occur in September, with the maize being treated for storage and sale following the harvest. Cash needs related directly to production are tied to the characteristics of the soil, which prevent the use of the hoe for land preparation. Draught power either through the use of animals or through tractors is necessary, and adds to production costs. Maize planting is labour as well as capital intensive, and the use of inorganic fertiliser raises the cost of production further.

None of the farmers sold maize through the NCPB, despite its location at Kipkarren River, close to Munyuki. Respondents argued that sale through the NCPB required not only selling in bulk, and engaging in cumbersome inspection and weighing processes, but also a variety of bribes to enable sale of maize not meeting the required standards. Most farmers sold small amounts of maize either in the villages by the tin or by the bag through agents/brokers travelling by bicycle. The more commercialised farmers sold larger quantities to assemblers using trucks. As suggested by Kirimi et al. (Reference Kirimi2011), the range of market outlets available to small-scale farmers in villages points to the penetration of the maize market even in relatively remote areas. Seasonal price differences follow national patterns, with the lowest prices occurring in November/December.

Seasonality related to non-farm expenditures

Whereas maize prices are affected by variations in supply and demand caused by the production cycle, cash needs and expenditure patterns outside the production cycle encourage farmers to sell maize towards the end of the year. In this way, aspects of seasonality and price differentiation can be understood both in relation to the production cycle, and through household expenditure patterns outside production. Table 1 outlines the price cycles for the three major crops grown in the villages: maize, sweet potatoes and beans, and relates these to costs of production as well as cash needs outside agriculture, connected in part to local cultural practices. Aspects of consumption-related seasonality discussed in the next section are also included in the table.

Table 1

Seasonality in markets for maize, sweet potatoes and beans in relation to cash needs within and outside agriculture, and household consumption burden

Note: The cash needs have been listed in relative terms, rather than quantified, since actual expenditures vary considerably between households. For reference purposes however at the time of the interviews, one 50 kg bag of DAP fertiliser cost 1,600 ksh (US$22), one 50 bag of CAN fertilizer cost 1,400 ksh (US$19), school fees were 1,500 ksh (US$20) per term. Labour costs were 70 ksh/day (US$1), plus tea and meals, ploughing using oxen was priced at 1,000 ksh/acre (US$14). All US$ values refer to the exchange rate for 1 July 2006.

Source: Adapted from Andersson & Wambugu 2007 on the basis of qualitative interview data, 2006 and own survey data 2008.

Cash needs due to traditional initiation ceremonies and Christmas and New Year celebrations are pronounced, especially at the end of the year. Such needs are compounded by the payment of the first term school fees in January. Seasonal price differences, however, are affected by production-related seasonality, market segmentation, and strategies employed by individual households to exploit or counteract seasonality. Importantly, the table shows a schematic picture of seasonality in the village, where some aspects are generalisable, whereas others are not: for instance, not all village households pay school fees, and not all households experience the consequences of seasonality in the same way.

Seasonal shifts in consumption and their effects on production-based seasonality

In addition to seasonal expenditure patterns, the consumption burden of households also varies over the year. This variance is in turn connected to seasonal return migration – whether of adult household members employed elsewhere on a seasonal basis, or children and relatives returning for school holidays. Return migration occurs mainly for the December holidays and to a lesser extent also in April and August. The scale of seasonal variability in household composition varies among households, as do its consequences, depending on household access to income and production levels.

Transfers of maize and other foodstuffs to relatives outside the villages compound the seasonal effects of the production cycle and household non-farm expenditures. In-kind transfers of maize to family members outside the village were used to smooth seasonality in relation to production systems found in neighbouring districts.

Forty-two per cent of households (N=60) reported sending maize to relatives in the 2008 survey. When such transfers are deducted from the maize used for household consumption, it becomes clear that these households are undermining their own food security in order to feed relatives outside the village, with the maize available for their own consumption falling from 180 kg to 133 kg per consumption unit.Footnote ⁵ Cultural and geographical aspects of production systems are relevant to the way in which seasonality plays out in the local environment, and are also tied to historical aspects of settlement and migration. Transfers occur primarily to rural areas: 63% of the households who sent maize did so to neighbouring villages, and 53% to other rural areas (see Andersson Djurfeldt & Wambugu Reference Andersson Djurfeldt and Wambugu2011 for a detailed discussion of in-kind transfers of maize).

The ties between Mukuyu/Munyuki and the neighbouring Vihiga District date to the decolonisation era, when groups of the Maragoli sub-ethnic group were resettled in the villages as part of the Million Acre Resettlement Scheme. Vihiga is densely populated, with the limited size of landholdings constraining production. Maize shortages are therefore pronounced. Family connections are strengthened by differing production cycles for maize, with harvests occurring earlier in Mukuyu/Munyuki. Accounts of how relatives from Vihiga came to the villages to collect maize immediately following the harvest were commonplace among the respondents. In contrast, respondents whose relatives were living in rural areas elsewhere reported not transferring maize or sending smaller amounts. Table 1 details the seasonal shifts in consumption burdens in relation to the production cycle for maize and the non-farm cash needs of the households. Since exact changes in the consumption burden are very difficult to measure, consumption is simply described as ‘higher’ in months where in-migration or in-kind food transfers are increased. Table 1 provides a schematic illustration of the consumption-related influences on seasonality in the villages. Importantly, the table is based on qualitative data: identifying how many households are involved in in-kind transfers, while also receiving migrants on seasonal basis, would require more detailed quantification, beyond the scope of this article.

December is a particularly problematic month, for both expenditure and consumption. The combination of initiation ceremonies with seasonal return migration that temporarily changes the demography of the household puts pressure on households in terms of cash outlays and subsistence obligations. This is compounded by the need to raise cash for school fees. While April is also problematic (the production cycle exerts demands for labour and other inputs, maize prices are rising, and self-produced food supplies are nearing exhaustion), returning migrants during this month can contribute labour for planting.

Smoothing agricultural seasonality through non-farm incomes

To some extent, exploring the commercial possibilities available to the farmer rests on avoiding the negative consequences of seasonality for the household in relation to both consumption and expenditures. Specifically, the ability to keep maize for consumption and sale beyond January/February enables households to avoid buying maize at elevated prices, while selling more expensively than immediately post-harvest. The key to this ability in turn lies in regular sources of income outside the maize sector. In most cases, such incomes are sourced from the non-farm sector, or more rarely through non-seasonal farm incomes, such as milk production. Regular income or income that is counter-seasonal to the various agrarian production cycles therefore constitutes a dividing line between poorer and wealthier households.

Five households of the sixty covered in the quantitative survey lacked cash income entirely and more than a third of the households had no income from non-farm sources. All of these were in the lowest cash income tercile. The importance of non-farm incomes as a dividing line between wealthy and poorer households, however, comes across most strongly between the second and third terciles. For households in the two lowest terciles, non-farm income constitutes 21% and 11% respectively of total cash income, whereas for households in the third tercile it comprises as much as 38% of total cash income.Footnote ⁶ For the lowest tercile, moreover, the high proportion of non-farm income indicates the extremely low total cash income in this tercile (US$17), rather than viable sources of non-farm income. The different sources of non-farm income are illustrative in this respect: the highest tercile earn non-farm income from relatively regular sources such as salaried employment, pensions and remittances (see Table 2). Hence the qualitative difference noted in the literature between the ex ante risk management of wealthier households and the ex post coping strategies of poorer households also emerges in the data from the villages.

Table 2

Cash income composition (%) by cash income tercile, for households which earned cash income

Note: Of the percentages above, the differences are statistically significant (at the 5% level) for staples sales and remittances as share of income composition between the second and third terciles.

Source: Own survey data, 2008.

In this context, households with regular incomes can counteract the negative consequences of seasonality from depressed prices as sellers of agricultural produce, while benefiting from the possibilities of seasonality through commercial diversification within agriculture. The upper income tercile is also less dependent on staple sales than the lower two terciles, suggesting that commercial opportunities within agriculture are broader and that maize can be used mainly for their own consumption. Here income from animal products (milk) is also an important source of non-seasonal diversification. In turn this enables households to respond to the seasonal complementarities of the different markets for agricultural products in terms of price variations over the year. By contrast, the lack of regular sources of income outside agriculture makes the effects of seasonality especially problematic for the poor. Such aspects are moreover aggravated by the seasonal shifts in the consumption burden of poorer households, whether through temporary return migration or through in-kind transfers of food to family members outside the villages.

Seasonality and intertemporal arbitrage among wealthier households

Households with access to regular non-farm incomes, such as remittances, salaried employment and pensions, or reasonably constant sources of farm income, such as milk sales, can wait for particular market segments to turn. Counter-cyclical non-farm incomes such as brick-making may also enable them to engage in commercial diversification within agriculture, without compromising the food security of the household. Crop diversification outside maize – especially in vegetables such as Kenyan kale (sukuma wiki), sweet potatoes, fruits and beans – provides opportunities for commercial complementarity drawing on the different production and market cycles of the crops in question.

Apart from selling produce directly in response to seasonal price variations, in-kind bartering of goods is also a source of potential profits. As with many in-kind transactions, however, bartering is difficult to quantify. A widespread practice in both villages was the exchange of maize for beans, either for home consumption or for trade. Described by one respondent as ‘black gold’, explanation for the relative profitability of beans can be found in the relationship between the production cycles of beans and maize as well as growing urban demand. Some respondents bought maize explicitly in order to exchange it for beans that could later be sold. Beans are intercropped with maize, require few farm inputs, and are easily stored. The bean season is counter-cyclical to the maize season, as the bean harvest coincides with the maize shortage in July. During the months of maize shortages, beans (traded by volume rather than weight) can be exchanged for the same amount of maize. The beans can be resold three months later at twice the price.

Profiting from exchanges, as with outright sale, is connected to the possibility of waiting for prices to rise. Exploiting the intertemporal differences in the barter terms of trade between maize and beans requires storage facilities, as well as flexibility on the part of the trader to make it possible to respond to fluctuations in supply and demand. Characteristics common to both commodities – resilience to decomposition and a relatively compact nature – enable farmers with storage facilities to stockpile the crops to take advantage of their complementary agricultural seasons. Beans and maize are both relatively easy to transport, enabling access to urban markets.

It is harder to profit from seasonal differences in demand for sweet potatoes, and such trading is carried out predominantly by small-scale women traders. Like beans, sweet potatoes are intercropped with maize, but harvested in February. Although maize prices at this point are rising, local cropping patterns make for an abundance of sweet potatoes at harvest time, which supresses prices. Some respondents reported exchanging sweet potatoes for maize, but the relative perishability and bulkiness of sweet potatoes limit the possibilities for storage and transportation.

Apart from trade-based exchanges in which raising income was an outright goal, using maize and other self-produced goods to pay school fees also contained an indirect element of intertemporal arbitrage. Since 2004 primary schooling has been free in Kenya, but the educational aspirations of Kenyan parents on behalf of their children have resulted in one of the highest secondary enrolment rates in Africa. In this part of Western Kenya specifically, early missionary presence historically contributed to high levels of education (Gwako Reference Gwako and Ensminger2005). School fees are a large expense for rural households, and an expenditure item which parents are especially reluctant to give up. Nonetheless, the ability to send children to secondary school in itself indicates access to higher than average incomes. Schools can on an informal basis decide to accept payment in kind rather than in cash. The decision to grant payments in kind is based on the perceived vulnerability of the household. School fees are converted into a maize equivalent on the basis of a fixed price per bag, which in the case of Mukuyu was nearly 40% higher than the highest price received by the farmers in the village, even during the lean season. Paying school fees in kind rather than in cash indirectly provides the highest producer price for maize in the villages, but requires enough maize to be able to engage in such payments, with the necessary production levels being beyond the reach of poorer households.Footnote ⁷

Seasonality, consumption and poverty

For poor households, the lack of regular incomes in itself aggravates seasonal aspects, and prevents their long-term engagement in agricultural markets. Instead, as suggested by the literature, in the most desperate cases households resort to selling maize immediately following the harvest when prices are at their lowest. While it should be stressed that all the interviewed households sold some maize following the harvest to cover the need for immediate cash, such as school fees, the relative amounts sold during the agricultural cycle varied greatly according to the general income level of the households in question. This was also suggested by the quantitative data, which documents the variability in prices received among households.

Although the phenomenon of what are generally known as ‘distress sales’ is documented in the literature (see Kosura-Olouch & Karugia Reference Kosura-Olouch, Karugia and Djurfeldt2005), such patterns in the case of the two villages also need to be understood in relation to demographic aspects of temporary migration and in-kind transfers of maize outside the villages. In this sense, a more detailed consumption perspective that considers the relationship to family networks outside the co-resident household unit (see Andersson Reference Andersson and Djurfeldt2010, Reference Andersson2011; Andersson Djurfeldt & Wambugu Reference Andersson Djurfeldt and Wambugu2011), as well as seasonal changes in the demographic structure of the household, can shed additional light on how seasonality affects poor households. This intersection between temporal and spatial perspectives illuminates the seasonal dynamics of poverty, and how mechanisms both within and outside agriculture combine to exacerbate the plight of poor households.

In the case of the poorer households, increased household size, whether in situ or across space, was related to two main factors. On the one hand, households are swollen by a seasonal influx of family members (whether adult family members or schoolchildren) in April, August and December, when schools and workplaces are on holiday. For wealthier households with larger land sizes, temporary increases in family size may provide sources of additional farm labour, whereas for poorer, land-constrained households this increase taps the production surplus, limiting their waiting power as sellers in maize markets. Since maize prices are already high in April, maize that could otherwise have been used to sell at a profit is used to feed returning migrants.

To seasonal increases in household size can be added the transfer of maize to relatives outside the villages following the harvest in September, which again carries very different connotations depending on the livelihood situation of the household in question. As outlined above, 42% of households (N=60) reported sending maize to their relatives in the 2008 survey. On average the amount transferred was 240 kg annually, with the number of transferring households being highest in the uppermost tercile (54%), and the size of transfers largest among households in the second tercile (35%) who sent transfers (295 kg annually). By comparison, a third of the households in the lowest tercile transferred maize, which on average amounted to 144 kg yearly.Footnote ⁸ The relative amounts of total maize production transferred were remarkably similar, regardless of the wealth of the household, with the average share transferred being 15% of total production: there are no statistically significant differences between the income terciles with respect to the share of production transferred. In turn, this suggests that the consequences of transfers differ greatly depending on the wealth of the household, since the withdrawal of 15% of total production for the poorest households constitutes a real threat to their food security.

The qualitative interview material indicates that while some wealthier households were able to transfer maize to relatives (sometimes in urban areas, in return for cash remittances or farm inputs), the poorer households transferred maize primarily to relieve food constraints in neighbouring Vihiga District. The distributional consequences of differences in consumption burdens may therefore be quite different, depending on the reciprocity and direction of transfers. In effect, for less well-off households this is an extension of household consumption beyond the production unit, which serves to reduce localised food shortages among members of the extended family network. Although wealthier households also support poorer relatives, the consequences of transfers for these households are different since their incomes and production levels are higher.

Asymmetry in reciprocity in the case of in-kind transfers of food is most pronounced for households that lack linkages to urban areas. In these cases, chronic shortages of maize in the maize-deficit receiving areas and among recipient households prevent reciprocity. Transfers in these cases aggravate shortages in the sending areas. In-kind transfers to relatives outside the village, even in the case of counter-cyclical production patterns, undermine the possibility of counteracting local shortages or benefiting from higher prices later in the season, since the surplus available for sale becomes smaller.

Interest in seasonal variation in agricultural markets has traditionally focused on the interaction of the agricultural production cycle with production-related expenses. The role of non-farm income has in this context been to mitigate seasonality by limiting covariate risks associated with a purely agricultural livelihood. An important distinction made in the literature is the difference between the coping strategies of poor households and the risk management strategies of wealthier households. Despite this interest in the distributional aspects of non-farm sector engagement, the focus has been on incomes (whether within or outside agriculture) and expenditures related to agriculture, rather than non-farm expenditure patterns. Moreover, seasonal variation in consumption patterns connected to the consumption of self-produced crops has rarely been considered. Instead, the common perspective on seasonal labour migration in the literature tends to focus on the effects of the withdrawal of agricultural labour at the peak of the production cycle, rather than shifts in consumption burdens over time.

The analysis above shows how seasonal shifts in expenditures outside agriculture as well as changes in the demography of the households aggravate the effects of seasonality for the poorer households, while enabling wealthier households to benefit from seasonal variations in the market for agricultural products through commercial diversification beyond staple crops. The starting point in this respect appears to be the possibility of smoothing seasonality through regular non-farm incomes, which in turn provide the basis for ensuring the food requirements of the household. Self-produced maize can be used to feed the household, rather than to meet pressing cash needs. Wealthier households can thus counteract seasonal effects on their food security, enabling them to engage in a variety of commercial strategies which take their starting point from seasonal variation.

At a general level, the situation of the poorer households points to the common problems associated with the African smallholder environment. In policy terms, the detrimental effects of seasonality on poor households can be eased by addressing some characteristics of the staple crop production environment in general: subsidising farm inputs may, as suggested by the case of Malawi, release production potential among smallholders, while also possibly dealing with parts of the structural maize deficit characteristic of Kenya. Such measures ease general constraints to production, while also averting the need to sell crops during a specific period of the production cycle to purchase inputs. At the community level, the introduction of storage facilities and warehouse receipt systems could prevent post-harvest losses and to some extent smooth seasonal availability in supply. Similarly, introducing standard weight measurements and institutions capable of monitoring their use could help to smooth seasonally fluctuating prices, both in cash and in kind. Nonetheless, a certain element of seasonality will always be present in markets for commodities that are directly connected to weather-related production cycles, as is the case for agricultural goods. Although the situation for the poorest households is aggravated by seasonal shifts in both consumption burdens and food availability, it is therefore important to remember that seasonality in itself is not the main villain: poverty is.