JHD052P004 FIRST PROOF 26-05-04 15:31:08 AccComputing

Journal of Human Development Vol. 5, No. 2, July 2004

Escaping Poverty and Becoming Poor in 20 Kenyan Villages

ANIRUDH KRISHNA, PATTI KRISTJANSON, MAREN RADENY and WILSON NINDO
Anirudh Krishna is Assistant Professor of Public Policy and Political Science at Duke University, Durham, North Carolina, USA; Patti Kristjanson is Project Leader of the Targeting Opportunities Research Theme at the International Livestock Research Institute (ILRI) in Nairobi, Kenya; Maren Radeny is an economist and Research Assistant at ILRI, and Wilson Nindo is a facilitator at Community-based Investigations in Kisumu, Kenya

Abstract Three hundred and sixteen households in 20 western Kenyan villages 19% of all households in these villages managed successfully to escape from poverty in the past 25 years. However, another 325 households (i.e. 19% of all households of these villages) fell into abiding poverty in the same period. Different causes are associated with households falling into poverty and those overcoming poverty. Separate policies will be required consequently to prevent descent and to promote escape in future. Results from these 20 Kenyan villages are compared with results obtained earlier from a similar inquiry conducted in 35 villages of Rajasthan, India. Some remarkable similarities are found, but also several important differences. Key words: Chronic poverty, Causes of poverty, Anti-poverty policies, Comparative poverty analysis

Introduction
By various estimates, people living in poverty constitute more than one-half of Kenyas population (Narayan and Nyamwaya, 1996; Government of Kenya, 1 1998; 2002).1 Economic growth has been inconsistent in the period from 1978 to the present, and while population growth has declined from 4% per annum in the 1970s to just below 3% in the 1990s, the Gross Domestic Product per capita has not improved much over the past three decades. Simultaneously, other social welfare indicators have also continued declining or remained mostly static, including life expectancy, child mortality, and primary school enrollments (OBrien and Ryan, 2001). However, while overall poverty has remained static in Kenya, one should not infer that those who were poor 25 years ago are still poor today or that those who are poor today will continue to remain poor in future. Aggregate head-counts can
ISSN 1464-9888 print/ISSN 1469-9516 online/04/020211-16 2004 United Nations Development Programme DOI: 10.1080/1464988042000225131

JHD052P004 FIRST PROOF 26-05-04 15:31:08 AccComputing

A. Krishna et al. conceal a much more dynamic picture of poverty one where there is a substantial flow of households into and out of poverty even as the net numbers remain stable or grow slightly at the national or regional levels. For example, it was found that between 1978 and 2003, as the national economy remained nearly stagnant and as the quality of social services and governance deteriorated in Kenya,2 316 households in 20 western Kenyan villages (19%) managed successfully to escape from chronic poverty. Chronic poverty exists among people who remain poor for much of their life course, and who may pass on their poverty to subsequent generations (Hulme and Shepherd, 2003, p. 6). Although these households were quite poor 25 years ago and regarded as such by others in their community, they are not poor now, and their success is widely acknowledged by their fellow community members. There is nothing particularly encouraging about these villages they are situated within one of the poorest areas of Kenya, Western Province, where 60% of the population (2.4 million people) fall below the official 1 poverty line (Government of Kenya, 2003. Nor is there anything exceptional or idiosyncratic about these particular households: there were no windfall gains or sudden bequests, nor were any particular patronage benefits directed their way. There are, however, some common threads running through most of these successful households experiences, and they help illustrate the pathways that other households could take to escape from poverty in this region. Learning from these experiences, suitable public policies can be designed to facilitate other poor households transitions out of poverty. Different policies will be required, however, to deal with the equally large number of transitions occurring (for different reasons) in the reverse direction. Nineteen percent of all households in these villages, including some better-off households at the start of this period, have made a reverse transition, falling into abiding poverty over the past 25 years. Overall poverty has declined by relatively little; in fact, it has increased marginally, by 0.6%, over the preceding generation. Households that declined into poverty face situations that collectively constitute some fundamental reasons for deepening poverty in this region. Addressing these causes will be fundamental to reducing future poverty in this region. In other words, two different sets of assistance programs will be required: one set to promote escape from poverty, and another set to prevent decline of households into poverty. Together these two sets of programs could make a considerable impact upon poverty, which persistently afflicts over 50% of households in this region. Localized investigations need to be carried out and more precise knowledge generated about factors associated with escaping poverty and entering it in different regions as a variety of agro-ecological and market conditions are associated with different regions and subregions. Diverse factors are likely to be associated with households escaping from poverty and households falling into poverty in a given area, and this information could be integrated and used to increase the effectiveness of policy interventions that could accelerate progress in poverty reduction. This paper addresses both the generation of knowledge of poverty-related 212

JHD052P004 FIRST PROOF 26-05-04 15:31:08 AccComputing

Escaping Poverty and Becoming Poor in Kenya factors and the necessary policy interventions, with a particular and detailed focus on the methodology underlying the study.

Methodology Background to a community-based methodology for assessing household poverty dynamics

Few studies have examined poverty in a dynamic context (see, for example, Walker and Ryan, 1990; Grootaert and Kanbur, 1995; Gaiha and Kulkarni, 1998; Glewwe and Hall, 1998; Carter and May, 2001; Dercon, 2001; Baulch and McCulloch, 2002; Christaensen et al., 2002; Deininger and Okidi, 2002; Haddad and Ahmed, 2003; Sen, 2003). Relying mostly on panel data collected for the same sets of households over multiple periods of time, such studies are expensive to conduct and, unless some pre-existing data are available for an earlier period, one needs to wait for a long period before comparisons can be generated and results obtained. Careful community-based inquiries can provide a viable alternative method for this purpose. Knowledge about changes in the situation of particular households is widely shared among members of close-knit communities. People who have lived together over reasonably long periods of time tend to know who among them has progressed or not and what events were associated with the rise and decline of different households: someone got a job in the city, while someone else succumbed to an expensive and ultimately untreatable illness; someone benefited from an irrigation scheme and now grows two crops instead of one, but at the same time another person lost their land to erosion or in a lawsuit. Eliciting this information carefully and systematically from community members and complementing and verifying it with information gained independently from individual households can assist in re-constructing the sequence of events associated with mobility (or stasis) in each particular case. An approach to assist with such comparative community-based inquiries was developed recently and utilized in detailed field inquiries conducted in two regions of India (Krishna, 2004). With the help of a pilot USAID-funded project, this methodology was piloted and adapted for use in five rural communities in Western and Nyanza Provinces in Kenya during October and November 2002. The results of this pilot study indicated that the methodology could be used with confidence to generate significant and reliable policy-relevant learning. This paper is based on the application of this Stages of Progress method to 1706 households located in 20 villages (i.e. 10 villages each of the Vihiga and Siaya Districts).3

Methodological considerations
Three aspects of the approach need to be discussed before presenting the steps in the methodology. First, defining poverty carefully and in a contextually relevant manner was critical for implementing the Stages of Progress 213

JHD052P004 FIRST PROOF 26-05-04 15:31:08 AccComputing

A. Krishna et al. methodology. In this regard, no pre-articulated definitions of poverty were imposed upon the villagers; instead, the understanding of poverty was elicited from the villagers themselves. Second, the villages were chosen making use of new high-resolution poverty maps for Kenya that show poverty incidence at the Location level.4 In order to compile a diverse group of villages, locations with higher versus lower incidence of poverty and concentrations of poor were identified, and one village was randomly selected from each identified location. The locations and villages and their corresponding poverty levels are shown in Figure 1. This paper reports the results of these investigations, carried out between May and September 2003, in 20 villages of two of Kenyas poorest districts, Vihiga and Siaya, located near Lake Victoria in western Kenya. Considerable geographic, economic and other differences exist among these villages: some are located close to major roads whereas others are more remotely situated; some are more commercialized and others are less affected by commerce and markets; some are quite large while others are much smaller; and so on. However, the reasons that explain escape from poverty or descent into abiding poverty were similar. Since the same method and questions were recently examined in Rajasthan, India, interesting and informative inter-country comparisons of both similarities and differences could be made between these two sets of studies. Third, a period of 25 years was selected for this study because it corresponds roughly to one generation in time. Periods of time much shorter than this are not very reliable for examining trends in household mobility. For instance, Walker and Ryan (1990, p. 99), referring to their stated inability to draw any conclusions from their 9-year panel data for households, claim that Nine years is too short a time to analyze issues that pertain to income mobility, which ultimately can only be addressed with intergenerational data.

Stages of Progress methodology

The Stages of Progress methodology, involving the following steps, was implemented separately in each of the 20 selected villages. Considerable care needs to be taken at each step, and the implementing team needs to undergo significant training at the start. Master trainers who had implemented this approach earlier in India led the 15-day training session in Kenya. Project implementing teams trained initially in the classroom for 5 days before going out into two villages each, where they learned to implement these methods in practice. Following training, the following steps were implemented within each of the 20 selected villages. Step 1: meeting with a representative community group. In order to elicit shared collective memories about the conditions of different households today and 25 years ago, it is necessary to first meet a diverse group representative of different segments of a community. All villagers are invited to these meetings. It is particularly important that older members from each 214

JHD052P004 FIRST PROOF 26-05-04 15:31:08 AccComputing

Escaping Poverty and Becoming Poor in Kenya

FIGURE 1. Study sites and poverty incidence for Nyanza and western provinces of Kenya. Source: CBS (2003).

community segment attend the meeting, as they can speak knowledgeably about the situations of different households 25 years ago and in the intervening period. In order to assure maximum possible attendance, an advance party was sent to each of the selected villages. This advance party met the village elder and village chief, explained the objectives of the project, determined the venue and time of the village meeting, and extended an invitation for all villagers to attend. 215

JHD052P004 FIRST PROOF 26-05-04 15:31:08 AccComputing

A. Krishna et al. Step 2: clear presentation of objectives. The objectives and procedures of the study were clearly explained, and it was emphasized that no tangible benefits would accrue to anyone from participating (or not) in the study. Quite often, visitors have gone into villages with the express intention of flagging off some new development program, and community members might have an incentive to deliberately misrepresent their material condition in order to gain the benefits of such programs. It is therefore very important to clarify the objectives of the study team, and to stress that there are no benefits to be had nor any losses to be incurred from participating in these particular community meetings. Step 3: collective definition of poverty. It is critically important to discuss and define a common understanding of poverty and of what it means for some household in the community to be regarded as poor. The rest of the exercise hinges upon this step. We asked the assembled community group: What does an extremely poor household do with the first bit of money that it acquires? Which expenses are usually the very first to be incurred? As a little more money flows in, what does this household do in the second stage? What does it do afterwards, in the third stage, in the fourth stage, and so on? Discussions on these questions elicited very active participation and debate from the assembled villagers. There was general consensus in each village at the end of these discussions, however, on what constituted the successive stages of households progress from acute poverty to a progressively better state. Most interestingly, there was broad agreement across nearly all villages on the sequence of these stages. Figure 2 represents a typical sequence of these Stages of Progress as recorded in the 10 Luhya-speaking villages of Vihiga District and the 10 Luospeaking villages of Siaya District. Not all households go through each of these stages in exactly the same linear sequence from first to 10th, but most households do traverse at least the first six stages in the order defined.5 Households progress upward out of poverty by first acquiring food, then clothes, then basic shelter, then money to pay for their childrens primary school costs, and then acquiring small animals, including chickens, sheep and goats. Once households have reached and crossed this particular stage, they are no longer regarded as poor within villages in this region. The first horizontal line in Figure 2 represents the poverty cut-off as it is constructed and perceived socially by residents of these 20 villages. Households that have not been successful in progressing beyond Stage 6 consider themselves to be poor in these localities and they are commonly regarded as such by other villagers in these locations. Households that are poor strive to progress further and to overcome poverty. The strategies that they adopt are related directly to how they understand this condition. Adopting the local understanding of poverty is thus useful not only to identify who is poor and who is not, but also helps us to learn what people are doing to deal with poverty as they see it. 216

JHD052P004 FIRST PROOF 26-05-04 15:31:08 AccComputing

Escaping Poverty and Becoming Poor in Kenya

FIGURE 2. Stages of Progress.

Step 4: definition of 25 years ago a well-known signifying event. While it is important to investigate household mobility over a 25-year period that corresponds roughly to one generation, merely saying 25 years ago can give rise to considerable confusion among members of the assembled community group. It helps to have a commonly known signifying event that fixes the date for the earlier period. While conducting these inquiries in India, reference was made to the national emergency of 19751977, an event that is clearly remembered by all older villagers. In Kenya, the time of President Jomo Kenyattas death (in 1978) was referred to, which nearly all older villagers remembered quite vividly. Step 5: current households as units of analysis with reference to stages of progress today and earlier. All households in the village were listed in this step. Referring continuously to the shared understanding of poverty developed in earlier Step 3, the community group was asked to describe each households status today and for the earlier period. Ranking each households position in terms of the successive Stages of Progress helped verify who was indeed poor in each period, and to assess relatively how 217

JHD052P004 FIRST PROOF 26-05-04 15:31:08 AccComputing

A. Krishna et al. poor they were in each period. For example, could they afford food but not shelter or clothes or primary school fees? Did they progress through each of these stages, but were unable to go on to the next stage, possessing first chickens and then sheep and goats? The assembled community members were quite forthcoming in these discussions, and it was possible to proceed quite expeditiously down the list of households, ascertaining each households poverty status today, 25 years ago, and also in the intervening period. There were relatively few disagreements regarding these classifications, and those that arose were resolved through discussion and debate among the assembled participants. Step 6: categorization of households. After ascertaining each households situation on the Stages of Progress at present and for the earlier period, all households were classified within one of the following four categories: Category A poor 25 years ago and poor now (remained poor); Category B poor then and not poor now (escaped poverty); Category C not poor 25 years ago and poor now (became poor); and Category D not poor 25 years ago and not poor now (remained not poor). Step 7: ascertaining reasons for change in a random sample of households from each category. A random sample of households was drawn from each of the four categories. At least 40% of households from each category were included within this sample. The assembled community group was asked about the circumstances associated with each such households trajectory over the past 25 years (e.g. How was Household X able to move out of poverty in this time? What were the major factors related with its escape from poverty?). Such in-depth inquiries about reasons were conducted for a total of 816 households in these five villages. In each case, it was important to dig deeper to ascertain the true combination of reasons and to adopt a comparative perspective while conducting these inquiries. Step 8: further verification of reasons for change or stability. Since many factors may not be known outside the particular household, it is important to consult another independent source in order to cross-check and verify what community members have stated. After learning about the factors associated with a households trajectory from the assembled community group, individual members of these households were interviewed separately. It typically took us 2 days to complete this inquiry in a small village (composed of less than 100 households) and it took from 3 to 4 days to complete the inquiry in larger villages (more than 100 households). Participating villagers were assured that everything they said at the meeting would be held in strict confidentiality in so far as individual households were concerned.

Results The dynamics of poverty within 20 Kenyan villages

A remarkably similar understanding of poverty appears to exist within these different villages and across the two different ethnic groups. This 218

JHD052P004 FIRST PROOF 26-05-04 15:31:08 AccComputing

Escaping Poverty and Becoming Poor in Kenya

TABLE 1. Percentage of households in four different categories 25 years ago At present Poor Not poor Poor 14.2% (A) Remained poor 18.5% (B) Escaped poverty Not poor 19.1% (C) Became poor 48.2% (D) Remained non-poor

Notes: Percentage poor 25 years ago (AB)32.7% Percentage poor today (AC)33.3%

understanding of poverty guides households strategies in this region. Understanding poverty in terms of these local understandings is effective for identifying successful and unsuccessful household strategies, while more theoretical definitions of poverty may not capture the specific nature of poverty and household strategies in a particular area. The data in Table 1 present the overall trends, providing the complete breakdown for all 1706 households in these 20 Kenyan villages. Members of 325 households among all 1706 households in these villages (i.e. 14% of all present-day households) have remained poor continuously over the past 25 years. On the other hand, members of 316 households (19%) that were poor 25 years ago have managed to overcome poverty in the intervening period. This achievement is compensated, however, by the experiences of another 325 households (19%) who have fallen into poverty during the same time period.6

Comparison between Kenya and India

Table 2 compares these trends in these 20 Kenyan villages with similar trends observed using the same methodology in 35 villages of the state of Rajasthan in India. It is important to note that the Stages of Progress and the poverty cutoff are not identical in India and Kenya. In Rajasthan, the poverty cut-off was located by village residents immediately after the first four stages, which
TABLE 2. Households in Four Categories in Rajasthan, India and Western Kenya Percentage of households that over the past 25 years: Remained poor (Category A) Rajasthan, India (35 villages) Western Kenya (20 villages) 17.8 14.2 Escaped poverty (Category B) 11.1 18.5 Remained Net improvement over not poor 25 years (Category B (Category D) minus Category C) 63.2 48.2 3.2% 0.6%

Became poor (Category C) 7.9 19.1

219

JHD052P004 FIRST PROOF 26-05-04 15:31:08 AccComputing

A. Krishna et al. included food, clothing and primary education (as in Kenya), but also repayment of private debts (which was not recounted as a distinct stage by Kenyan villagers). On the other hand, repairs to houses and purchase of sheep or goats were included in Kenya as separate stages within the poverty cut-off. It is contestable whether the differences that arise here are smaller or larger than other differences that arise on account of differently operationalizing a common monetary standard in different contexts (e.g. $1 per 2 day) (Schelzig, 2001; Reddy and Pogge, 2002; Wade, 2002). It is useful, however, to explore how households in each different locales performed in terms of their own particular understanding of poverty. Nineteen percent of households in the Kenyan study escaped from poverty, compared with only 11% in Rajasthan. However, a much higher proportion of Kenyan households also fell into poverty at the same time 19% compared with only 8% in Rajasthan. Consequently, progress in poverty reduction was higher over this period in Rajasthan (3.2%) compared with Western Kenya (0.6%). Quite similar reasons for falling into poverty were found in Western Kenya and in Rajasthan, India, as will be discussed later, but the extent to which these factors influenced poverty dynamics differs between the countries. Not all of these Kenyan villages have fared equally well or equally badly. The 20 villages that we studied are all located within two adjacent provinces of Kenya, but there are very considerable differences among them in terms of households escaping poverty and households becoming poor. Table 3 provides some illustrations of these inter-village differences. While the percentage of households in poverty declined overall by as much as 53% in Ebwikabula Village (Vihiga District), with a large number of households escaping poverty and fewer households becoming poor, this net percentage increased over the same time period by 36% in Usonga Village (Siaya District). Such large inter-village differences are noted also by Narayan and Nyamyawa (1996), who found households in poverty ranging from 33%
TABLE 3. Highest and Lowest Rates of Poverty Reduction Net reduction Percentage in poverty poor 25 over 25 years ago years 66.4 62.5 37.8 31.6 17.7 28.2 32.7 53.1 41.7 27.6 27.6 31.0 35.9 0.6

Remained poor (Category (A)

Escaped poverty (Category (B)

Became poor (Category (C)

Remained not poor (Category (D) 30.1 31.9 59.2 34.2 49.6 30.8 48.2

Percentage poor today 13.3 20.8 10.2 59.2 48.7 64.1 33.3

Villages with the highest rates of poverty reduction Ebwikabula 9.7 56.6 3.5 Buronya 15.3 47.2 5.6 Mweywe 7.1 30.6 3.1 Villages with the lowest rates of poverty reduction Ujwinya 25.0 6.6 34.2 Asere B 15.9 1.8 32.7 Usonga 23.1 5.1 41.0 Average (20 villages) 14.2 18.5 19.1

220

JHD052P004 FIRST PROOF 26-05-04 15:31:08 AccComputing

Escaping Poverty and Becoming Poor in Kenya to 91% in different villages. In neither their study nor the present one, however, does any particular village characteristic or any set of characteristics help to explain these inter-village differences. Factors such as village population, distance from market, and extent of commercialization do not associate significantly with the overall change in poverty in these villages. No single reason appears to account for escape or for descent into poverty; rather, different sets of reasons are associated with transitions in either direction. The same sets of reasons are broadly shared among all 20 villages. However, their incidence is quite different among these villages.

Reasons for escape and descent Reasons for descent into poverty
No single reason is responsible, in most cases, for households decline into poverty. Rather, a combination of reasons is responsible for plunging households into abiding poverty. Poor health and health-related expenses constitute the most often stated reason for households declining into poverty. Nearly 73% of households that have fallen into poverty (i.e. 149 out of 205 households of Category C) mentioned sickness, poor health, and heavy healthcare expenses as a principal reason for their decline into abiding poverty. It is not just HIV/AIDS that is responsible for these households decline into poverty. Poor health and high healthcare expenses had been ravaging these households economies long before AIDS emerged as a major scourge in sub-Saharan Africa. AIDS is the crushing blow, however, that devastates households already weakened by long-term illnesses and ensuing poverty. However, it was beyond the scope of this study to determine how many households were seriously affected by AIDS. In addition, death of the major earner on account of illness was mentioned as another principal reason for falling into poverty. The resulting dependence of survivors, including orphans, upon other households, thereby increasing the burden on these households, was a contributory factor for descent in another 32% of cases.7 It appears that improving healthcare provision constitutes the single most important aspect for policy intervention. However, three other reasons combined with health to influence decline into poverty in these villages. First, heavy expenses related to funerals (including slaughter of the familys livestock) were mentioned in 63% of cases. Second, large family size was reported to be a contributing factor in 38% of cases, and third, small landholdings, together with uneconomic subdivision of land, was also thought to increase the chance of declining into poverty by 62% of households. Large family size and land subdivision are often closely related, and households seem to increasingly recognize these to be risk factors for deepening poverty, resulting in a growing acceptance of family planning within these communities. The provision of family planning services is, however, uniformly poor and often non-existent in this area. Making such 221

JHD052P004 FIRST PROOF 26-05-04 15:31:08 AccComputing

A. Krishna et al. services available to the relatively receptive populations in these villages should also be a priority. Finally, two other factors were examined that are frequently mentioned by urban elite as causes for enduring poverty among the rural poor: drunkenness and laziness. These factors were not found to be very significant in these villages. In less than 12% of all cases of descent did either drunkenness or laziness figure among the principal reasons. Poor people do not become poor principally on account of drunkenness or laziness. Neither do they remain poor for these reasons. Drunkenness and laziness figured as contributing causes for less than 15% of Category A households. For the remaining 85% of Category A households, remaining poor had nothing at all to do with drink or with indolence. Rather, their reasons for remaining poor were similar to those observed earlier for becoming poor in this region namely, ill health and health-related expenses (66%), heavy funeral-related expenses (60%), and small or unproductive landholdings coupled with large family sizes (88%). The principal reasons for descent into poverty recorded in these Kenyan villages are quite similar to the reasons recorded earlier in 35 villages of Rajasthan, India. Ill health and high healthcare expenses constitute the single most important reason for descent in both contexts. Expenses on death feasts are also an important reason for descent in western Kenya and Rajasthan. However, in India two additional poverty inducing factors emerged: high expenses on marriages and high interest private debt. These factors were not as significant in the Kenyan context. Drunkenness and laziness appeared equally insignificant in both contexts.

Households escaping from poverty

Diversification of incomes by establishing links with the urban economy is critical for the majority of households who have escaped from poverty in these Kenyan villages. Some people have found jobs in the formal or informal sectors, while others have established themselves in some petty trade. Seventy-three percent of households that have successfully escaped poverty reported a household member obtaining a job. More than 80% of these jobs, accounting for 61% of successful escapes, were held within the private sector, while government jobs were found in relatively few cases (13%). In 42% of cases, one or more household members diversified their incomes by setting up a trade or a craft in the city. Another 36% of escapes from poverty were associated with successfully establishing a small business in the vicinity of the native village.8 It can be concluded that diversification of income sources accounts collectively for the vast majority of all cases of successful poverty escapes in western Kenya, just as it did in Rajasthan, India. However, diversification is not an option that is open to all villagers who are able and willing to travel and work hard. In order to diversify successfully, these household members must be able to overcome entry barriers defined by skills, contacts and capital access (Barrett et al., 2001). Furthermore, while they may be successful in 222

JHD052P004 FIRST PROOF 26-05-04 15:31:08 AccComputing

Escaping Poverty and Becoming Poor in Kenya acquiring the required skills by themselves, capital and particularly contacts are not equally available to all poor households (World Agroforestry Center, 2002). Households that have successfully made an entry into the urban economy, whether in the formal or the informal sector, have almost invariably possessed a privileged connection, an uncle or cousin or some other willing patron, who has taken the new entrant under their wing and provided assistance with establishing economic and social connections. In other words, it appears that education, intelligence and hard work are not enough per se to escape from poverty. For example, although a majority of households of Category B (69%) mentioned hard work and personal initiative among the principal reasons for escape, members of households belonging to Categories A and C also reported hard work to a similar extent. Because the required access points and opportunities are lacking for the second set of households, however, hard work in their cases does not translate into escape from poverty. Similarly, education is not enough by itself to ensure escape. Many educated villagers remain poor on account of lack of access to opportunities. Furthermore, in an economy quite sharply segmented by non-market barriers of information and access, high transaction costs limit the efficient use of human capabilities and individual resources. Access to opportunities in the city is not an option available equally to all able and willing household members. Instead, these households could benefit from improved harvest and yields. Making this option available to a larger number of poor households will require improving the rural infrastructure, expanding the accessibility of extension services, and reducing the costs incurred by small farmers. These approaches depend on prevailing national and international conditions, and subdivision as well as increasing landlessness might further limit the scope of this option.

Conclusion
Statistically, poverty levels have remained virtually unchanged in Kenya, and they also seem virtually unchanged when one looks only at the aggregate picture for the 20 villages of this study. While this might, at first glance, suggest that poverty in Kenya has barely been affected by policy changes and economic conditions over time, a dynamic micro-level examination of poverty reveals just the opposite conclusion: 19% of households have escaped from poverty in these villages, while another 19% have fallen into poverty during the same time. This paper has demonstrated that there are active pathways both into and out of poverty. Identifying these pathways can help generate more precise knowledge about reasons for escape from and descent into poverty in specific contexts. Programs and projects can be developed that block pathways leading into poverty while reinforcing those that lead out of poverty.9 It is helpful to adopt the definition of poverty used by local residents as their strategies for breaking out of poverty are intimately related to how they define and understand this condition. The Stages of Progress 223

JHD052P004 FIRST PROOF 26-05-04 15:31:08 AccComputing

A. Krishna et al. methodology helps to facilitate this understanding and assists in identifying the successful and unsuccessful strategies as well as key obstacles to coping with poverty in any given context. Since different factors are associated with escaping poverty and falling into poverty, a more comprehensive policy for poverty reduction will need to take both sets of factors into account and deal with them differentially. For example, the striking importance of health and disease-related problems and expenses in both contexts, Rajasthan, India and western Kenya, implies that government assistance in the form of loans and subsidies will not suffice to keep people from becoming impoverished. Until health constraints are overcome, decline into poverty cannot be feasibly arrested. Policies promoting diversification of income sources, on the other hand, should assist households in escaping from poverty. Education is important, but it cannot constitute the key element in poverty alleviation. Unless opportunities are available additionally to obtain jobs or to set up small trades or businesses (or improve farm incomes reliably), poor households will quite probably continue to remain poor. Similar findings reported from Rajasthan and western Kenya and also from other parts of the developing world (Eder, 1999; Ellis, 2000) indicate that it would be beneficial to explore more widely the utility of diversification as a strategy for escaping poverty. The Stages of Progress methodology is a relatively fast, inexpensive, reliable and participatory approach that can be utilized by community residents, and also by researchers and policy-makers. Community residents can be empowered through training to apply this methodology by themselves, tracking poverty and identifying reasons for escape and reasons for descent. Linked with other research approaches, such as detailed householdlevel surveys, and addressing additional aspects, including intra-household differences, this method is a useful tool for future studies. It is indeed important to continue research of this kind to monitor Kenyas poverty dynamics over years to come, and to fine-tune the understanding of poverty inducing and poverty relieving factors within these differentiated and quite volatile contexts.

Acknowledgements
The authors would like to thank Milton Esman, Alan Fowler, Ruth MeinzenDick, Frank Place, and Brent Swallow for helpful comments and advice. Sharad Pathak played a key role as master trainer. Chad Hazlett provided excellent research assistance. The authors remain entirely responsible, however, for all errors and omissions.

Notes
3

1 Depending on which source one consults, recent measures of poverty in Kenya range from 42% to 54% these numbers are 52% in the Welfare Monitoring Survey (1997), 42% in World Bank (2002), and at least 54% in each district examined by Narayan and Namwaya (1996).

224

JHD052P004 FIRST PROOF 26-05-04 15:31:08 AccComputing

Escaping Poverty and Becoming Poor in Kenya

2 Per-capita Gross Domestic Product in Kenya declined at a rate of 0.5% per annum between 1990 and 2000. The growth rate over the longer period 19752000 was a bare 0.4% per annum (UNDP, 2002). 3 Funding for this larger study was made available from the Food and Agriculture Organizations Pro-Poor Livestock Policy Facility. 4 A Location is an administrative unit in Kenya. Each District is divided into multiple Locations. Poverty incidence is defined in these maps as the percentage of the population living below the official rural poverty line in each Location. 5 After reaching this level of relative prosperity, households tend to diverge somewhat in the later stages (e.g. some among them rent land before purchasing cattle). Until they have advanced past Stage 6, however, households mostly tend to follow similar stages in their pathways out of poverty. 6 A note of caution is in order while interpreting these figures. Households today and households 25 years ago are not strictly comparable. Some households that exist today did not exist 25 years ago, and some households that existed 25 years ago do not exist today. By regarding households of today as the unit of analysis, as mentioned earlier, what we are, in fact, comparing is inherited versus acquired status: did a person who was born to poverty remain poor, or did they escape from poverty in the past 25 years? Is another person who was part of a non-poor household 25 years ago still non-poor, or have they fallen into poverty in this time? 7 Multiple reasons are associated usually with any households transition in the downward (or upward) direction, so these percentages do not add up to 100. 8 Once again, multiple reasons are associated with households escaping poverty. 9 Some among the successful 19% of households could possibly slide back into poverty, and some proportion of the unsuccessful households would manage to escape poverty in the future. This instability of the movement in and out of poverty has to be considered by policy-makers as well. It would help to conduct similar Stages of Progress inquiries at regular intervals, say, at 5-year intervals, in order to generate more precise knowledge in this regard.

References
Barrett, C. B., Bezuneh, M. and Aboud, A. (2001) Income diversification, poverty traps and policy shocks in Cote dIvoire and Kenya, Food Policy, 26, 367384.. Baulch, B. and McCulloch, N. (2002) Being poor and becoming poor: poverty status and poverty transitions in rural Pakistan, Journal of Asian and African Studies, 37(2), 168185. Carter, M. R. and May, J. (2001) One kind of freedom: poverty dynamics in post-apartheid South Africa, World Development, 29(1), 19872006. Christiaensen, L., Demery, L. and Paternostro, S. (2002) Growth, Distribution and Poverty in Africa: Messages From the 1990s, World Bank, Washington, DC. Deininger, K. and Okidi, J. (2002) Growth and Poverty Reduction in Uganda, 19922000: Panel Data Evidence, World Bank, Washington, DC. Dercon, S. (2001) The Impact of Economic Reform on Households in Rural Ethiopia, 1989 1995, Center for the Study of African Economies, Oxford. Eder, J. (1999) A Generation Later: Household Strategies and Economic Change in the Rural Philippines, University of Hawaii Press, Honolulu, HI. Ellis, F. (2000) Rural Livelihoods and Diversity in Developing Countries, Oxford University Press, New York. Gaiha, R. and Kulkarni, V. (1998) Is growth central to poverty alleviation in India?, Journal of International Affairs, 52(1), 145180. Glewwe, P. and Hall, G. (1998) Who is Most Vulnerable to Macroeconomic Shocks? Hypothesis Tests Using Panel Data from Peru, World Bank, Washington, DC. 1 Government of Kenya (1998) Second Report on Poverty in Kenya. Volume 1: Incidence and Depth of Poverty, Ministry of Finance and Planning, Nairobi, 89 pp.

225

JHD052P004 FIRST PROOF 26-05-04 15:31:08 AccComputing

A. Krishna et al.
1 Government of Kenya (2003) Kenya: Interim Poverty Reduction Strategy Paper 20002003

(http://poverty.worldbank.org/files/Kenya%20IPRSP.pdf). Grootaert, C. and Kanbur, R. (1995) The lucky few amidst economic decline: distributional change in Cote dIvoire as seen through panel data sets, Journal of Development Studies, 31(4), 603619. Haddad, L. and Ahmed, A. (2003) Chronic and transitory poverty: evidence from Egypt, 199799, World Development, 31(1), 7185. Hulme, D. and Shepherd, A. (2003) Conceptualizing chronic poverty, World Development, 31(3), 403423. Krishna, A. (2004) Escaping poverty and becoming poor: who gains, who loses, and why? Accounting for stability and change in 35 north Indian villages, World Development, 32(1), 121136. Narayan, D. and Nyamwaya, D. (1996) Learning from the Poor: A Participatory Poverty Assessment in Kenya (http://www4.worldbank.org/afr/poverty/databank/docnav/show_ doc.cfm?ID3280). OBrien, F. S. and Ryan, T. C. I. (2001) Kenya, in S. Devarajan, D. R. Dollar and T. Holmgren (Eds), Aid and Reform in Africa: A Report from Ten Countries, World Bank, Washington, DC. Reddy, S. G. and Pogge, T. W. (2002) How Not To Count the Poor (www.socialanalysis.org). Schelzig, K. (2001) Escaping poverty: behind the numbers, Public Administration and Development, 21, 259269. Sen, B. (2003) Drivers of escape and descent: changing household fortunes in rural Bangladesh, World Development, 31(3), 513534. UNDP (2002) Programme Support Document, Programme of the Government of Kenya (http://www.undp.org/sl/Documents/Country_specific/Kenya_PSD/Kenya_psd.htm). Wade, R. (2002) Globalization, Poverty and Income Distribution: Does the Liberal Argument Hold? (www.lse.ac.uk/depts/destin). Walker, T. S. and Ryan, J. G. (1990) Village and Household Economies in Indias Semi-Arid Tropics, Johns Hopkins University Press, Baltimore, MD. World Agroforestry Centre (2002) Voices of Poor Livestock Keepers in the Lake Victoria Basin, report commissioned by NRI for the UK Department of International Development, World Agroforestry Centre, Nairobi, 60 pp.

226