Middle-Income Groups in Kenya. Conflicting Realities Between Upward Mobility and Uncertainty

For more than a decade scholars mostly from economy and development studies have described the rise of a newly emerging ‘middle class’ in the Global South including Africa. This has led to a ‘middle class narrative’ with the ‘middle class’ as the backbone of economic and democratic development. Especially with regard to the stability of the position of the people in the ‘middle’, empirical social science studies challenge the ‘middle class narrative’ and at their uncertainty and insecurity. This tension between upward mobility at the one hand uncertainty and instability at the other hand (the vulnerability-security nexus) and the options to cope with this challenge under the condition of limited provision of formal social security is the focus of this case study on Kenya. Instead of an analysis of inequality based on income, it is more helpful to start from the welfare mix and the role of social networks as main elements of provision of social security. Against this background, we identify different strategies of coping that go together with different sets of values and lifestyles, conceptualised as milieus, that are not determined by the socio-economic situation.

Global South, we observe the opposite. For more than a decade, scholars, mostly from economics and development studies, have described the rise of a newly emerging 'middle class' and a slow but visible decrease in the number of people living in poverty (Banerjee/Duflo 2008;Birdsall/Graham/Pettinato 2000;Heiman/Freeman/Liechty 2012;Kharas 2010;Milanovic/Yitzhaki 2002;Ravallion 2010). This observation nurtures the idea that the rise of the 'middle class' is an indication of the success of economic development and the underlying, or corresponding, development policy. This has led to a 'middle class narrative' (Neubert 2016;Neubert/Stoll 2018). In this narrative, the 'middle class' is presented as the backbone of economic development (Easterly 2001), as the advocate of a "just and sustainable economy" (Wiemann 2015b), or as being "good for good governance" (Birdsall 2015; see also Wiemann 2015a). This follows the classic notion of the 'middle class' as a carrier of democracy (Lipset 1959;Moore 1967). 2 It not only underlines high expectations, but also includes the assumption that there is a real 'middle class' that is homogeneous and stable ('the' middle class).
At least in respect of Africa, the idea of a rising 'middle class' as promoted by economists has been under scrutiny in the social sciences. In particular, critics express doubt that people who have just escaped poverty constitute a real 'middle class'. Some critics also doubt that the conventional class concept helps us to understand the social-economic and socio-cultural differentiation of this growing 'middle' in society. They call for new analytical approaches, whether with or without the concept of class (e.g. Darbon 2014;Melber 2017;Neubert 2019). Against the background of the conceptional problems, definitions of this growing group in the 'middle' of society are often very cautious and refer to 'inbetweenness' (Fichtmüller 2016;Lentz 2016: 43;Nallet 2014: 77;Phadi/Ceruti 2013), a situation in which people are neither rich nor poor (Darbon 2014: 14;Kroeker/O'Kane/Scharrer 2018: 9), without any clear class position. The 'middle class' in Africa, including empirical findings on socio-economic situation and conduct of life, has been intensively debated in several edited volumes Darbon/Toulabor 2014;Kroeker/O'Kane/Scharrer 2018;Melber 2016). A very useful short summary of the debate has been provided by Roger Southall (2018).
The question whether this growing group in the 'middle' of society in Africa is a class in the strict sociological understanding of the term or not is not the topic of this article. The focus here is on the stability or instability of the socio-economic situation of this rising group labelled as 'middle class'. Therefore, I prefer the more neutral and descriptive term 'middleincome group' for my analysis. This connects to the socio-economic data that triggered the current debate on the 'middle-class'. The definition of this 'middle class' centres around per capita income per day. Depending on the authors, the lower threshold is set between 2 and 10 US$ per capita and day calculated according to purchasing power parity (PPP). 3 For Africa a report of the African Development Bank is the most influential source (AfDB 2011). The AfDB defines the middle-income group as those with an income between 2 and 20 US$, and refers to the group with 2 to 4 US$ per capita and day as the unstable 'floating class'. The growth of the middle-income group results from upward mobility out of poverty beyond the 2 However, Lipset warned that we should not overstress this correlation (1959: 72). 3 For an overview of thresholds, see Neubert/Stoll (2018, 61). Some authors include in their definition secondary education, formal sector employment or adequate housing with access to basic infrastructure (Thurlow/Resnick/Ubogu 2015: 589f). statistical poverty line of 2 US$ per capita and day. The AfDB report makes a difference between the 'middle class' in general (including the 'floating class') and a more nuanced presentation in its tables, where it differentiates between the 'floating class', on the one hand, and those who cross the 4 US$ line into the 'lower middle class' and 'upper middle class', on the other hand. This implies that the latter have reached a fairly stable position. At the same time, this 'middle class narrative' implies a certain homogeneity of conditions of life and attitudes (implied by the term class). Especially with regard to the stability of people's position, empirical social science studies challenge the 'middle class narrative' and point to the uncertainty and insecurity of the middle-income group. 4 This uncertainty and instability of the middleincome group and the options to cope with this challenge are the focus of this article.
The analysis is based on an empirical study carried out in Kenya. Thus, it starts with some background information on the middle-income group in Kenya and on the empirical study. The majority of members of the middle-income group lives in uncertainty. The second part of the paper is a discussion of the consequences of this lack of stability. They work hard to keep their position. We will see that sustaining one's social status is closely linked to the particular welfare-mix (on this term, see Esping-Andersen 1990;Evers 1990). According to their welfare-mix, people have different opportunities to cope with uncertainty. This leads to the third part of the paper. Against this background, we need to analyse social inequality not just via income or 'class' but also with regard to the welfare-mix. However, how the options of the welfare-mix are used depends on the values and decisions of the middle-income earners. Finally, I ask whether these findings also apply to other African countries, and, in the light of the analysis, I compare the challenges facing the European middle-income group with the situation in Kenya, showing that there are some basic similarities but also clear differences.

The Middle-Income Group in Kenya
Kenya is one of the countries where, according to the data of the African Development Bank, we find a middle-income group of a considerable size. Thus, it is an interesting case for a deeper look. If we include the so-called 'floating class' (the group from two to four dollars), Kenya has one of the largest non-poor groups in Africa south of the Sahara, representing 44.9% of the whole population (AfDB 2011: 5). In only five countries (Gabon, Botswana, Namibia, Ghana, Cap Verde) is this percentage larger, while even in South Africa it is slightly lower, with 41.1%. Despite partly diverging data on the size of the middle-income group, 5 the existence of this group in Kenya dates back to colonial times. At the time of colonisation, most Kenyan pre-colonial societies were acephalous societies with limited stratification, ruled by 4 For Benin, see Alber (2016), Heilbrunn (2014); Ghana: Hamidu (2014: 187); South Africa: Kriege (2015: 112). 5 According to the AfDB data, this size is due to the comparatively large percentage of the population with an income above the 20 US$ a day line (13.4%) who cannot all be simply seen as 'upper class'. However, we have to consider that this data has some weaknesses. The results of World Bank Data (www.povcal.net, accessed 7.8.2015) and Thurlow/Resnick/Ubogu (2015: 597) show less people living above the 10 or 20 US$ line (per capita and day). However, the consumer analysis data suggests the existence of a considerable middle class in Kenya (Euromonitor International 2010). The 'Kenya economic report 2013' by the Kenya Institute for Public Policy Research and Analysis (2013: 209) places the percentage of poor people between 46.1% and 50.8% according to a multi-dimensional poverty index for the years 2006-2012. the 'old men ' (e.g. Kenyatta 1953;Ocholla-Ayayo 1976). Only a few coastal towns had stratified societies. With the introduction of a modern state administration and the advance of capitalism, Kenyan society underwent a process of social differentiation. Some people started as blue-collar workers (working, for instance, in the harbour). Especially the small group of educated people experienced upward social mobility. They took lower and middle ranks in the administration. In the beginning, primary education was sufficient, but later secondary education was needed. A few Kenyans with university degrees took higher positions. Those in middle and higher positions were described as the 'African elite', although they were in fact part of a middle-income group. 6 The size of this group grew, especially after the Second World War and after independence. It started with 10.1% of the population in 1950 and increased to 19.2% in 1970(Berg-Schlosser 1979. 7 During this process, a typical pattern of economic activity developed. Urban income was invested in the home farm and was a main factor for the development of commercial agriculture. Sometimes additional land was acquired. This was possible because in the 1950s a process of land registration started with an option for individual ownership. In addition, wage income was also invested in trade. Successful small entrepreneurs used surplus income again for investment in agriculture. This led to an combination of income from different sources (Kitching 1980;Kongstad/Mönsted 1980). Thus, the members of this middle-income group were at the same time peasants, white-collar workers, small entrepreneurs, and sometimes also blue-collar workers in the formal sector. All were closely linked with their rural homes. A more recent study by Githinji (2000) confirms this pattern of combined sources of income still for present based on data for rural households. He states that 44% of rural households combine farm production and off-farm income. The majority of households have members who perform off-farm wage labour, while a small number employ labour. The rest involves a combination of off-farm business and family farm without employees. 8 However, it is remarkable that the question of the combination of sources of income has not been statistically researched. Most statistics draw a clear line between urban and rural households, despite the observation that many families combine rural and urban income. In addition to the combination of income there are links inside the extended family between urban and rural areas. Better-off family members, usually in town, send regular remittances to poorer family members in rural areas. This underlines the frequent socioeconomic heterogeneity of extended families (Ferraro 1971: 119-124;Githinji 2000: 53-56;Kroeker 2018: 281). Interestingly, the combination of sources of income was an important topic in a general development discussion that took place in Germany in the 1990s in respect of the so-called 'Bielefeld approach' (Elwert/Evers/Wilkens 1983). However, the focus was on the poor. Despite the evidence that the concept of combining income sources is a general feature of societies in Africa south of the Sahara, it is still widely ignored.
With accelerated economic growth, especially after the political changes at the end of the Moi era in 2002, the middle-income group rose to approximately 45% of the population. 9 The following table shows a calculation of the monthly income in real US$ of a family of six in the middle-income group.  Table 1 shows the wide range of income for the middle-income group labelled as 'middle class'. Studies on the middle-income group that include also OECD countries for the analysis use different thresholds, e.g. between 10 and 100 US $ (per head/day) (Kharas 2010, 12). This shows that with regard to Africa the AfDB uses much lower thresholds. In the case of Kenya, we include for our analysis also a large number of people above the twenty-dollar threshold in the middle-income group. As I will argue below, due to the insecurity and volatility that many people above the twenty-dollar threshold experience, they do not belong to the more stable group but to a much smaller top income-group. This underlines the above-mentioned critique in the social science debate on the term 'middle class' and the somewhat arbitrary choice of thresholds (see for example Darbon 2014;Melber 2017). In general, we should be careful not to put too much emphasis on income data alone. Income data helps as an illustration, but, as we will see, income does not describe the real socio-economic situation. Thus, lower and higher middle-income positions are just approximate terms. The middle-income group in Kenya was the focus of a research project in Bayreuth (Germany) which provided empirical data for the following sections of this article. As part of the larger project 'Future Africa -visions in time' at the University of Bayreuth (2012-2018), funded by the German Ministry of Education and Science, a sub-project 'Middle class on the rise' was carried out. 11 Its aim was to investigate the social, economic and political contexts 10  and characteristics that shape this middle-income group. Topics with regard to Kenya were socio-cultural differentiation, social mobility and (in)security, and strategies of small-scale entrepreneurs. The 'Middle class on the rise' sub-project was carried out by an anthropological research team comprising Erdmute Alber (project leader), Lena Kroeker and Maike Voigt, and a sociological team comprising Dieter Neubert (project leader) and Florian Stoll. Both teams conducted qualitative field research with long-term stays in the field based mainly on interviews, and participant observation. The empirical data given here for Kenya stems from this project and the ideas presented here are the result of close co-operation with Florian Stoll 12 and intensive discussion with the other project members. A first result of this study was that the combination of sources of income identified in older studies still applies to the Kenyan households of the middle-income group. Especially the work of Kroeker and biographical information collected by Stoll and Voigt underline the uncertainty of the socio-economic position of many middle-income earners which constitutes the background for this article (Kroeker 2018;Voigt 2018). A large part of the interviewed middle-income group had moved up from poverty or from lower socio-economic positions. 13 At the same time, they often reported that they had experienced downward mobility as a result of shocks (such as unemployment, sickness and health-care costs, death of a breadwinner, business failure, etc.). Some even fell back into poverty. Even when they remained in a lower middle-income position, they were always aware of the risk of (further) downward mobility, but also of opportunities to climb further up the social ladder (Kroeker 2016). Thus, our research provides evidence that the need for means to cope with risk and uncertainty is a main feature of the middle-income group in Kenya (Kroeker 2018;Neubert 2017).

Means of Coping with Risk and Uncertainty
The uncertainty of the middle-income group is often overlooked. The combination of income sources is a typical strategy to deal with volatile social positions. At the same time, 'middle class' as an 'aspirational category' (Spronk 2018: 333) shows that dealing with uncertainty is important for improving one's position, with risk-taking as a way of attaining upward mobility. This attitude is usually associated with middle-income earners, but it also applies to those project, and for additional grants by the Volkswagen Foundation that allowed us to make an in-depth analysis of our data: a post-doc grant for Florian Stoll for a one-year stay at Yale University and a one-year 'opus magnum' grant for Dieter Neubert. 12 During his field research, Florian Stoll conducted more than 100 interviews, which constitute the background for this analysis, together with the findings of Lena Kroeker and Maike Voigt. The qualitative and ethnographic approach follows Hartmann's recommendation (2012, 83) that research on lifestyles and conduct of life needs a reliable database. We found our first interviewees in the everyday settings of the middle-income group (shopping malls, enterprises with white-collar workers, leisure events, etc.), and also in more open settings such as church communities and neighbourhoods. Further sampling followed the principles of saturation and contrasting. This included a wide range of socio-economic positions. Despite the limitations of income measurement, our sample covered the whole range from people close to poverty (2 US$ per capita and day) to people with high incomes (> 20 US$ per capita and day). And, as we will see, it also covered different lifestyles and milieus. This was facilitated by a long familiarity with Kenya and long periods of field work. For an overview of the project, see Kroeker/Voigt (2017). I will refer to other published results later in the article. 13 Due to the qualitative nature of our data, these statements reflect trends not precise figures. In the case of upward mobility, our findings correspond to the economic debate. qualified as poor. Otherwise, the growth of the middle-income group that triggered the 'middle class debate' would not have happened. 14 This moderate upward mobility and overcoming the two-dollar-a-day threshold is the reason why the 'growth' of the number of middleincome earners is mainly a growth of the so-called floating class (two to four dollars a day). Despite these findings, in the policy debate, instability and uncertainty (often addressed as vulnerability) is still linked with the poor, or with less successful middle-income earners. But we need to consider it as a much wider phenomenon in African societies, with a manifest influence on social positioning and social structures. At the same time, coping with vulnerability and aspiring to upward mobility are crucial elements of individual social positioning, resulting from agency under the given structural conditions. Uncertainty results from the risk of being hit by typical shocks, such as unemployment, long-term sickness (cost of medical treatment, loss of income), old age, failure of business or farming risks. The means of coping are formal sector social security programmes, investment and assets, and semi-formal and informal solidarity networks. In Kenya there are formal social security programmes, especially for people employed in the formal sector. As in other countries, a core element of these formal schemes is provision for old age through pension schemes. In Kenya, the pension system for the formal sector is well developed. The employee and the employer each contribute 6% of the salary. Retirement age is 60. For the pension, "the insured has three options: purchase a life annuity from a registered insurer, take a partial lump sum and purchase an annuity with the balance, or take a portion of the accounts as a lump sum" (SSA -The United States Social Security Administration 2015: 121). People who are self-employed may join the scheme, but they only have the option of a lump sum payment (SSA 2015: 121). 15 Government officers may receive up to 75% of their salary after a long period of service (Künzler 2016). Disability and early retirement pensions, or pensions for widows and underaged children, are part of the pension. However, the payments are usually much lower than the possible top rates. For an assessment of these pension schemes, we need to consider the often unstable employment situation. Even well-educated people may move between different jobs, with periods of unemployment or self-employment, and salaries may go up and down. For this reason, people are often happy to have the minimum membership that qualifies for the basic pension, often with just 20 or 30% of their average salary. In addition, the pension is only based on the salary, and not on additional payments, such as a housing allowance, (and other subsidies paid by the employer), which may form a considerable part of a person's earned income. Finally, the pension schemes cover only about 10% of the population of Kenya (Dau 2003: 30-31). This means that not even all members of the middle-income group have access to a pension scheme. Unemployment, a typical risk for employees, is not covered by any formal benefit system. 14 Economic growth facilitates individual upward mobility. However, we should be aware of the fact that even when the overall number of people with middle-income increases there are also people who experience downward mobility. A growing middle-income group describes a positive balance of upward and downward mobility. 15 A simplified calculation with reference to the year 2009 may help to assess the amount of the pension: with a comparatively high salary of 500 US$ over 30 years, the payment will be more than 20,000 US$. However, with a life expectancy of 70, the pension would be about 170 US$ a month.
Other serious risks are sickness and maternity care. Formally, all employed persons earning more than 1000 Kenyan shillings (KSh) a month, 16 including government employees and self-employed persons, have to be members of the National Health Insurance Fund (NHIF). The insurance includes the employed person and the dependents of that person. People with a lower monthly income may contribute voluntarily. The minimum contribution is 150 KSh per month and rises to 1700 KSh for those earning more than 100,000 KSh per month. The NHIF has about three million members, including the voluntarily insured (Künzler 2016: 70-74). In view of a population of more than 50 million, , because of the low rate of formal employment the coverage is limited, even if we assume that each insured member of the NHIF has four dependents. The national hospital insurance card guarantees access to free basic medical treatment, including hospital treatment for certain illnesses such as tuberculosis, sexually transmitted diseases, and AIDS (SSA 2015: 122f). Those who are able to invest more in their health subscribe to private health insurance schemes, which cover treatment in private health facilities and more than the basic medical interventions. All these health insurance schemes cover or subsidise only the costs of medical treatment but does not compensate salary or in case of self-employed income losses. The Employment Act regulates sick leave. After two months of employment, the employer has to pay 100% of the earnings for two weeks during sick leave. Some trade union agreements extend this period to between one and six months, and sometimes for an additional period with 50% of the earnings. Maternity leave with full pay is granted by employers for three months (SSA 2015: 122). However, except in the case of special trade union agreements, long periods of serious sickness are a great challenge to security, not only for the self-employed but also for employees, even when the cost of medical treatment is covered.
As a kind of safety net for the poor and groups with special needs (such as orphans, vulnerable children, people over 65 living in extreme poverty, poor people with disabilities), there are cash transfer programmes. They offer 2000 KSh per month and thus reach people living at the poverty line. They do not cover the whole of Kenya, but are often directed at special regions with a large number of poor people. In regions with food shortage, a hunger safety net programme (HSNP) exists. In addition, there are various emergency aid activities in cases of serious food shortages (Künzler 2016: 79f). Thus, there is a certain degree of classical social security provision, which might cushion some shocks, but it not does not provide sufficient social security to keep their socio-economic position, even for the formally employed. The beneficiaries still face a considerable loss of income that often goes together with downward social mobility.
We need to remember that the middle-income group usually combines different sources of income, and that social security payments are based on a salary in the formal sector, and sometimes on voluntary contributions to an additional private health insurance scheme. What is not covered in all cases is loss of income during long periods of sickness, or the death of the breadwinner. In addition, typical entrepreneurial risks are not covered, such as business failure or farming risks (pests, plant diseases or volatile prices).
A very important household strategy for security provision is the diversification of risks. As already underlined, the combination of different sources of income not only creates a higher total income, but also has the effect that problems with one source of income can be compensated, or at least cushioned, by other sources. An unemployed person may still have income from a business and/or commercial farming. A regular salary may cushion financial problems in the (small) enterprise or on the farm. Especially investments in agriculture and livestock are seen as a security net that offers a means of subsistence and additional income even in times of crisis. Investment in agriculture is also seen as an important means of old-age provision. This is often combined with investment in a rural home. In one's old age, one can live in the village without paying rent, and benefit from an agricultural income and the lower costs of living in a rural area. For those with the financial means, investment in urban housing (a flat or a house) is also an option, either to save rent or to gain an additional income by renting out the additional flat or house. Investing in a business can also be used as a means of oldage provision. Such investments are assets that offer a certain security and generate additional income. However, they are also bound up with different kinds of risks.
Against this background, it is not surprising that semi-formal and informal networks are extremely important when it comes to coping with shocks (Kroeker 2016(Kroeker , 2018. These structures have been addressed as 'traditional solidarity' (Benda-Beckmann et al. 1988), or 'informal social security' (Benda-Beckmann/Gsänger/Midgley 1997). Although the organisational patterns of 'solidarity' are constantly developing, the basic elements remain the same (for Kenya, see Neubert 1990). The core element is reciprocity in family and kinship networks. In the absence of a reliable old-age security scheme, the children have the responsibility of caring for their old parents. In cases of sickness and unemployment, the family or the extended family are approached for financial help. Typically, these are costs for health treatment, school fees or school uniforms for the children. Even though primary education is free, many middleincome class families prefer to send their children to better quality private schools. Well-off members of the kinship network often help to pay for the children of their siblings and other more or less close relatives. Members of the middle-income group who received support for their own education, and who have been economically successful, often feel obliged to give something back to the family. This can also be a way of gaining social prestige and a better social status. At the same time, these people can expect support from other members of the extended family if they suffer from shocks. This responsibility towards the family can also be extended to the local community. Successful community members may support other community members, or donate money for community activities or projects. People living in the same neighbourhood may found a network. Givers in these networks can achieve the role of a 'big man' or 'respected big woman'. This can lead to patron-client relationships where the giver offers financial support and expects political support and loyalty in exchange. These relationships may reach beyond family or community bonds. In other cases, the church community may support its young members with a scholarship for secondary education or even for a university degree. This also results in an obligation 'to give something back to community', as it is often put by successful members of the middle-income group in our interviews. These networks are based on the assumption of generalised reciprocity (Sahlins 1965: 193-194). This means that help is given in accordance with people's needs, with the knowledge that the givers can expect to receive support should they be in need themselves. However, where the givers succeed in maintaining their status, this potential reciprocity will not be realised. Due to the inhomogeneity of families, these networks cut across different socioeconomic positions.
There are also semi-formal networks, such as rotating savings groups, which create a kind of obligation. These groups may be formed in a common housing area, or by employees of the same company, or may be based on the same occupation. Some savings organisations take the form of Savings and Credit Co-operatives (SACCOs), mainly based on occupational background (Kroeker 2018: 285). They may even offer a kind of pension scheme. A very special form of organisation is burial societies that help their members to manage the often very high costs of burials. Church communities may also support their poorer members (Niechoj 2016). 17 Pentecostal churches play a special role with regard to these networks (Kroeker 2018: 285f). They expect their members to reduce or even cut off their kinship relations and to practise solidarity within the church community instead. The Pentecostal church communities are usually fairly homogeneous in socio-economic terms, and thus offer middle-income members a network of equals. Against this background we need to assume that nearly all Kenyans, the middle-income earners and the poor, rely on semi-formal and informal security networks, together, if possible, with investment in real estate, agriculture, livestock or a business enterprise.
In the understanding of Amartya Sen, formal, semi-formal and informal security structures are entitlements. With respect to coping with famine, he defines 'entitlement' as "the ability of people to command food through the legal means available in the society, including the use of production possibilities, trade opportunities, entitlements vis-à-vis the state, and other methods of acquiring food" (Sen 1981: 54). For Sen, entitlements are "not 'rights' in the normative sense…They represent actual possibilities" (Sen 1977: 34). Transferred to the informal and semi-formal security networks that we have discussed, the first point is that entitlement to support applies only in a case of need or an emergency. There is a moral entitlement, but the potential giver still has the power to decide whether to help or not, and in what way. By contrast, formal security schemes are stronger, and the right to benefits can be asserted in court. Some entitlements may be reflected in data on household spending, such as regular pension payments, remittances by kin, or regular remittances to kin. The majority of entitlements are not visible as regular income or as regular payments, because they are payments that are made only in times of need. Yet these entitlements make a clear difference. A strong informal network can be a protection against downward social mobility, while a weak network can lead to great hardship when minor problems occur, such as a need for surgery or a short phase of unemployment.
At the same time, these semi-formal and informal security networks can be a heavy burden, especially for better-off network members. In large extended family networks, or any other inhomogeneous social networks, there will always be poorer members who are unable to cover their basic needs or to pay the school fees for their children. This never-ending obliga-tion to support the poorer network members can become a serious strain for the better-off members. They are under constant pressure to give.
Socio-economic positioning of households depends on strategic decisions in respect of managing income generation, making economic investments, paying for the children's education, and meeting obligations towards members of the family network. The handy formula of 'security maximisation' (Elwert/Evers/Wilkens 1983) might work for the very poor, without overdoing the rationality argument (Neubert 1986). However, for the better-off, there is room for manoeuvre. Providing for the future by investing in education, a business enterprise, real estate, agriculture or individual security provision is important for the large majority of Africans. People in Africa need to strategise between acquiring entitlements of different kinds and handling obligations to provide for security. 18 This interplay between vulnerability and strategising, the 'vulnerability-security nexus' (Neubert 2019) is a central feature of African societies and has a crucial influence on social positioning. For the middle-income group, this vulnerability-security nexus is played out in a situation of rising expectations in respect of consumption and living a decent and comfortable life.

Welfare-Mix Inequality and Social Inequality
Against the background of the vulnerability-security nexus, and individual and collective entitlements, a person's socio-economic position cannot be defined by simple income measurement. Even in cases where the household members have stable salaries, or stable returns from a business enterprise or farm, it is not easy to calculate the available household income. One important reason for this is the irregular nature of entitlement transfers, and the problems of including them in household statistics. Nevertheless, these payments reduce the available household income, especially the income available for investment and/or individual provision for social security. The available household income depends on the balance of entitlements and can only be measured over long time spans. Even a year might be too short to capture the real entitlement balance, which defines the available income and influences people's ability to safeguard their social position and their options for upward mobility. For the long-term perspective, the particular welfare-mix is crucial. 19 This includes access to formal social security provision, including individual provision, such as additional health insurance or even a pension plan, assets and investments in real estate or business enterprises (including commercial agriculture) that offer a stable return, and investment in the children's education, and their willingness to support their parents. A decisive factor is the economic strength of the networks, because they fundamentally influence the available options. Two simplified scenarios may show the resulting different opportunities and choices and strategic options (Neubert 2017; for more details, see Neubert in print).
The first scenario is a person who has succeeded in improving his or her social position. The person's kin, friends and community members are still poor. This person is in a special 18 I borrow the concept of 'strategising' from Nici Nelson's analysis of female households in a Nairobi squatter area (1977). 19 Lena Kroeker uses the term 'social security mix' (2018: 284). position as somebody who has successfully climbed the social ladder, and perhaps has access to formal social security programmes, but there will be hardly any other network members in this position. According to the norm of generalised reciprocity, this person is obliged to help the poorer members when they are in need. There are two different strategies she or he can adopt. The first strategy is to accept this responsibility and use a part of her or his income to fulfil the requests of the poorer members. The consequence is constant pressure on the giver. Even if giving support is subject to negotiation, a considerable part of the giver's income will be spent on other network members. This gives the person a high status in the community and it may be possible to maintain a certain living standard, but capital for investment or savings will always be short. This restricts the possibility of investing in the education of one's own children or in a social security scheme, and hinders individual accumulation. In this case, the networks are still the main safety net for the giver. If the giver needs help, the resources in the network constituted by poorer members will be limited. They may help to safeguard survival in cases of emergency, but no more. The consequence is that every emergency destabilises the social position of the former giver.
The second strategy is to distance or even dissociate oneself from the network. The ties in the network are kept weak or are completely cut. The main reference group is the nuclear family. The consequence is social isolation and critique from members of the extended family and other relevant networks. Without constant requests for help from network members, the person's income is available for investment in business, land or a house, a social security scheme, such as health insurance, the education of the children, and/or consumption and a better living standard. Even if their income is still limited, these people may now decide how to use their resources without pressure from their relatives. If everything goes well, they have a chance for further advancement and accumulation. However, in the case of trouble, such as serious sickness, unemployment, business problems, a bad harvest, or simply a burglary, then the situation changes. Coping with these shocks is possible if sufficient capital can be accumulated, or with effective health insurance and an adequate pension scheme, but these need to be supplemented by a farm or a small business, and the hope that the children will do well and be willing to support their parents. All this is only possible if there are no serious shocks, which is rarely the case. If these individual provisions do not work, or are not sufficient, the lack of a social support network will be felt, and the person's social position will change radically for the worse, even threatening their survival.
In both strategies, the person's social position remains instable. In the first case, support for the network provides a basic safety net; in the second case even this basic safety net is missing. This highlights the dilemma of having to choose between safeguarding survival and limiting further upward mobility, on the one hand, and striving for further upward mobility while taking the risk of falling back into poverty without a safety net, on the other hand. In practice, people navigate, or strategise, between these two extremes, which limits their chances for upward mobility.
A second and completely different scenario is a setting where the person and the network members are all well off. This applies to a small number of middle-income families whose parents, or in some cases grandparents, have already achieved upward mobility. The siblings are usually all well-educated and have good jobs. The family has assets, like land, or an improved house 20 in the countryside. In addition, they may have a house or flat in town, perhaps a small business enterprise, and some savings. This network is economically strong, and at least some of the network members may have access to formal social security. The network members can afford private education for their children, and have enough money and/or health insurance to cope with sickness. Thus, cases of emergency are rare. The person's income can be spent on further investment, the children's education, and better social security. In cases of shock, several members of the network are able to help, which allows cost sharing and a reduction of the burden. This underlines that family and kinship obligations in Kenya are not only accepted but also held in high esteem. If the person is in need, support is available, and even serious challenges will be mastered due to the network support. In addition, the network can be mobilised for support for investment, and may offer access to new business opportunities or better employment. Such networks offer considerable support. In these cases, there is no economic reason to distance oneself from the network. Cutting such ties does not open up new options, but limits access to the network if support is needed. People in this situation have a very good starting position to safeguard their social position, including options for further advancement. There may even be some rich network members who can help to improve one's position further. Joining a socially homogeneous (Pentecostal) church community creates a situation similar to this scenario. Once the relation to poorer family members is cut or restricted, the church community steps in, which limits the number of requests for help but also the options for support in times of need.
The choices in each scenario involve different risks. On the basis of our qualitative data, we assume that only a small part of the middle-income group is in the privileged position described in scenario two. These people may make use of their networks for investment, not only in order to stabilise their middle-income position, but also as a means of further upward mobility. The network is strong enough to cushion investment risks. However, according to our assessment, the large majority of middle-income earners live in a situation comparable to scenario one. They need to choose between risk-taking entailing the consequence of weakened network ties, and remaining in the network as a safety net to prevent falling back into serious poverty, with resulting obligations towards poorer network members.
Thus, a person's long-term socio-economic position is defined by the interplay between assets, income generated by salaried and entrepreneurial activities, accepted obligations in social networks, and strategies for coping with shocks, such as individual social security provision or cultivating social networks. From the perspective of the vulnerability-security-nexus, social inequality and stratification can be described in terms of different 'social-economic' clusters defined by networks, their assets, entitlements, obligations and income. Such clusters are not as clear-cut as groups defined by income or position within the labour market. For Kenya, we define the following clusters: • The 'poor' without sufficient assets, with very restricted entitlements, and unable to fulfil obligations beyond contributing to the bare survival of others. Differences are constituted by access to means of subsistence and/or access to networks with some resources. • A widely extended 'unstable floating cluster' with limited assets, limited entitlements and extensive potential obligations, resulting in permanent insecurity and vulnerability, and a highly volatile social position, even for people who have an attractive occupational position or a successful business. • A 'stable middle-income cluster' with some assets, sufficient entitlements and medium obligations. • A 'stable upper cluster' with large assets, extended entitlements and low obligations. • Political and economic top elite with extremely large assets, extensive entitlements and minimal obligations. We use the term cluster, and not class or stratum, because, especially in the 'unstable floating cluster', and to a lesser extent in the 'stable middle cluster', we find people with extremely different levels of income. Successful upward mobility from poverty may lead to a very good or even top occupational position with income classified as upper middle-income ('upper middle class' according to the African Development Bank) or even more. But network obligations and lack of security provision lead to a high risk of moving down again. Thus these clusters do not coincide with the assumed 'class structure' based on income as presented by the African Development Bank (see Table 1).
There is a potential rupture between the poor and the unstable floating cluster, on the one hand, and the two stable clusters and the top elite, on the other hand. The members of the latter clusters, who we see as a minority in Kenya, share the privileged situation that they can use their income to invest for the future, including investing in the education of their children and acquiring new assets. Under favourable economic circumstances, this may be a starting point for a hidden process of social closure cutting through the middle-income group.
A 'cluster' does not refer to a 'class'. Members of a cluster do not automatically share similar interests, or the same norms and values, and they do not develop a common political consciousness. And members of the 'unstable floating cluster' may not even realise that their situation is similar in terms of security. Especially in scenario one, people can choose between risk-taking and security, and their choices differ, depending on their norms and values, their hopes and dreams, and the resulting differences in their plans for the future with regard to risk-taking, and in their feeling of responsibility towards their social networks.
In our research we identified various 'milieus' within the clusters and cross-cutting them. 'Milieus' are "subcultural units within a society which group together people with a similar view of life and way of life" (Flaig/Meyer/Ueltzhöffer 1993: 55, eigene Übersetzung). 21 A few examples may illustrate such differences. In scenario one, we identified a group of typical risktakers, often from a poor background, who have a clear career orientation. They keep their network bonds weak, or cut them off, in order to have a chance for upward mobility. We label this 'milieu' as 'social climbers'. A milieu that seems to accept even more risk is the 'young professionals' (already described by Spronk 2012). They are in a very good employment posi-tion with a high salary, they minimise or cut off their network ties, and they have enough income to combine investment and consumption. However, they are comparatively young, usually without children, and may change their risky behaviour when they start a family. In addition, some 'young professionals' originate from the 'stable middle-income cluster' or from the 'stable upper cluster' and minimise their network relations for social reasons, yet still rely on them in times of need. Another strategy is found amongst members of the 'Pentecostal churches' where people try to build up a socially homogeneous network, while cutting off relations with their poorer family members, in order to reduce obligations and gain access to a stronger network. Whether the new religious network is as reliable as a family network will show up in the case of serious problems. At the other end of the spectrum of strategies in scenario one, we find 'risk avoiders' who try to safeguard their position, consume moderately and keep the family networks strong. We label this 'milieu' as 'stability-oriented pragmatics'. According to our assessment, most members of this group live under the conditions of scenario one. But some may have a much stronger network, as in scenario two. There are also other groups ('milieus') that are not clearly characterised by risk-taking or risk avoidance. This may include people for whom scenario one applies, and others for whom scenario two applies. They all maintain their family ties to some degree, and try to navigate between maintaining their networks, career advancement and (moderate) consumption. They mainly differ in their moral and political values.
Our analysis of the vulnerability-security nexus offers a new perspective on social inequality in Kenya. Instead of starting with income, it is more helpful to start from the welfare mix and the role of social networks. The strength of the social network and its entitlements are crucial factors that define the two scenarios for the middle-income group. The options available in the scenarios help to define the different clusters according to the instability or stability of people's social position. These clusters do not coincide with the 'classes' referred to by the African Development Bank and other economic data. Especially instability goes far beyond the four-dollar threshold (given for the 'floating class' in the AfDB data) into higher income levels. People have different strategic options based either on close links with their networks or on cutting off ties with their networks. The different strategies go together with different sets of values and lifestyles. We conceptualise these as milieus that do not systematically coincide with the clusters. Thus, for an analysis of social structure, we need to analyse two different elements. First, we need to look at the socio-economic situation, reflected by the clusters; this was the main purpose of this paper. Second, we need to consider the socio-cultural differences briefly presented here as different milieus (for a full analysis see Neubert in print).

Conclusions
The first result of this analysis of the conflicting realities between upward mobility and uncertainty is that the instable and uncertain situation of the majority of people in the middleincome group can be described in terms of a vulnerability-security nexus. Thus, to assess the socio-economic position of the Kenyan middle-income group, we need to consider not only income but also the welfare mix with all entitlements, including semi-formal and informal networks. At the same time, the way members handle their vulnerability does not just depend on their particular sources of income and their welfare mix, but also on which strategic decisions they make, as described in the two scenarios. Especially members of the middle-income group with weak semi-formal and informal networks and limited assets need to decide how much risk they can take for the option of further upward mobility. Thus, there is room for action and decision according to individual preferences. Even an elaborated and nuanced analysis of people's socio-economic situation that considers income, including a combination of different sources of income, assets and the welfare-mix, will not reveal their strategies for dealing with vulnerability. Or, with reference to a simplified Marxist idea, we may say that being does not determine consciousness.
The entitlements either of kin or other social networks that mark the situation of the middle-income group also need to be considered for the large group of the poor. For them, as for the middle-income group, strong entitlements can cushion shocks and offer a means of survival. Greater entitlements may offer a chance for upward mobility, for example by provision of education or support for investment in a small business. This underlines the importance of understanding the vulnerability-security nexus for analysing socio-economic inequality. With regard to the middle-income group, the statistics show that this group has been growing since the 1950s. In Kenya, the upper middle-income group (10 to 20 US$ per capita/day) and the group of those whose income is above the 20 US$ threshold have reached a visible size, even though they are still a minority. The socio-economic situation of latter (entrepreneurs, higher-level white collar employees, professionals) allows a living standard that comes close to that of the "well-off" in the Global North. In addition, the size of the other middle-income groups (lower-middle and upper-middle) has increased. Classification according to income suggests that the thesis of the 'erosion of the middle class' in the Global North does not hold for Kenya. This leads to questions that need to be discussed. First, whether Kenya is an exceptional case, and second, what differences and commonalities exist between the Kenyan (and other African) middle-income groups and the middle-income group in the Global North.
As already mentioned, Kenya is one of the countries in sub-Saharan Africa with a comparatively large middle-income group. But this does not mean that the middle-income group in Kenya is exceptional. The typical features described for Kenya, such as a combination of different sources of income, overall insecurity and uncertainty, and the crucial role of semiformal and informal networks, apply generally to middle-income groups in other African countries. 22 With regard to formal social security, Kenya is situated somewhere in the middle between countries with a fairly well developed social security system (e.g. South Africa, Mauritius) and others with only very basic formal social security systems with very limited coverage (e.g. Central African Republic, Mali or the D.R. Congo) (SSA 2015;World Bank 2015). Of course, there are differences between Kenya and other countries in Africa south of the Sahara. In Mauritius, the majority of the population and of the middle-income groups work in the formal sector with a strong social security system and people often have just one source of income: either formal employment, commercial agriculture or income from a small or medium enterprise. Their situation is less insecure, and informal and semi-formal networks are less important for survival. This also holds true for the large formal sector in South Africa. At the same time, South Africa is one of the most unequal societies in Africa south of the Sahara. For the poor, and even for parts of the middle-income group, insecurity is ever-present, and semiformal and informal networks are important as a means of survival (Duclos/Verdier-Chouchane 2011). 23 If we compare the situation of the Kenyan middle-income groups and of similar groups in other parts of Africa south of the Sahara with those in the Global North, we can identify a number of striking commonalities. The thesis of the 'erosion of the middle class' in the Global North is based on the threat of downward mobility, and, depending on the country, a more or less significant shrinking of the size of the 'middle class'. This process is highlighted by a growing inequality resulting from the concentration of capital in the hands of the top elite. This means that the position of the 'middle class' in the Global North is becoming worse, at least in relative terms. Insecurity and Inequality is also a reality in all African countries, but with large variations. Nevertheless, nobody doubts that in all African countries there is a rich and powerful top elite. The difference with regard to the Global North is that the middleincome group -defined according to the AfDB -has been growing in most of the countries in Africa south of the Sahara. In the Global North, a typical response to the threat felt by the middle-income group is 'preventive status work' (präventive Statusarbeit) (Groh-Samberg/Mau/Schimank 2014; Schimank/Mau/Groh-Samberg 2014). 24 This term has been used to describe the fact that safeguarding one's position needs strategic investment in social security, education, housing, and so on. Belonging to the middle-income group in the North means having to actively cope with uncertainties. This is also the case in Kenya (and other African countries), although the middle-income group there is situated at a much lower economic level; yet it faces similar risks and relies on a less well developed social security and social welfare system. Navigating between maintaining networks, career advancement and (moderate) consumption, with the aim of creating a stable welfare mix, is nothing other than preventive status work. One important difference between Kenya and Germany, or other Central European welfare states, is that the middle-income groups in Kenya have to rely on themselves and their networks because of the limited capacity of the formal security systems -as is the case in the majority of the states in Africa south of the Sahara. Thus, the options for preventive status work depend on people's assets, their welfare mix, and especially on the economic strength of their networks, and are not just an individual or nuclear family affair. The empirically identified scenarios discussed here show that the majority of middle-income earners who cannot rely on a strong economic network have to choose between a high-risk strategy with possible upward mobility and greater security with no upward mobility. With favourable economic conditions, risk-taking may lead to success and allow people to rise into the middle-income group and people may improve their income as middle-income earners. However, without more developed formal social security, there will always be a large number of people who fail, move downward, and possibly end up in poverty. This high volatility is the crucial difference between Central Europe and Africa. In Africa, there is erosion of the middle-income group and growth of the middle-income group at the same time. 25