Teachers’ needs and agency in times of low-fee private schools: The case of Kenya

Abstract The proliferation of low-fee private schools (LFPSs) in the global South is one manifestation of the marketization of education. LFPS literature on teachers emphasize exploitation, de-professionalization, and higher accountability, but teachers’ own voice and representation has largely been absent. Based on interviews with 35 Kenyan LFPS and public-school teachers in one urban and one rural area, this paper partly fills this gap by investigating teachers’ own experiences with marketization. The different teachers’ needs and sense of professionalism showed great similarities, yet their type of employment shaped their work and agency. LFPSs alleviated public school (hereafter PS) overcrowding, but high pupil mobility in the education market affected teaching and learning in both LFPS and PSs negatively. LFPS teachers adopted strategies (teaching-to-the-test, deviating from pre-scripted lessons) to retain and attract pupils to keep their jobs.


PUBLIC INTEREST STATEMENT
In the debate on the role of so-called low-fee private schools (LFPS) in the provision of education, the images of LFPS and public-school teachers are often quite instrumental and simplified. This study aims to add teachers' voices on how the growth of a private education market has affected the needs and agency of primary-school teachers in Kenya. We found that the teachers had similar work/life needs and sense of professionalism, but the circumstances of where they were at in their careers shaped their ability to navigate challenges and fulfil their needs. Years waiting for public employment in LFPSs with low pay, precarious employment, competition and pupil mobility affected the LFPS teachers' wellbeing, integrity and ability to progress in their work/lives. Still, they adjusted demands in the classroom for their own and pupils' sake. Publicschool teachers believed that LFPSs may have alleviated their still overwhelming enrolment numbers, but, like LFPS teachers, their relations with pupils suffered from high pupil mobility.

Introduction
Low-Fee Private Schools (LFPS) are non-government funded and/or managed schools with relatively low fees, compared to the private schools that cater for a middle-or high-income population. In recent years a "second wave" of LFPSs has emerged, one characterized by corporate-backed for-profit school-chains, targeting low-income families and areas. Educationally under-served lowincome areas, where such schools used to be an ad hoc response from parents, communities, churches, and NGOs, have thus been identified as a market to tap into. The expansion of LFPSs chimes with a phase in which the marketization of public goods is encouraged as global policy (Härmä, 2021;Srivastava, 2010;Verger et al., 2018).
Proponents of LPFSs find them a good or better supplement or alternative to overcrowded and under-resourced PSs (Dixon, 2012). Some studies have shown that LFPSs can have positive effects on exam results, as keeping teachers (e.g., local youths) accountable leads to lower absenteeism and higher effort (Andrabi et al., 2008;Dixon, 2012;Heyneman & Stern, 2014;Tooley, 2009). Those that view LFPSs as problematic, point to issues of equity (fees hampering access) and the mistaken view that quantitative measures (e.g., exam results) are effective proxies for quality (Edwards et al., 2015;Härmä, 2011Härmä, , 2021Lindsjö, 2018;Nambissan, 2010;Riep, 2017Riep, , 2019. Regardless, when it comes to teachers that work in LPFSs, they are predominantly talked about and often in quite instrumental ways (Singh, 2021). It is highlighted that their salaries are very low, that employments are insecure (Day Ashley et al., 2014;Edwards et al., 2015;Heyneman & Stern, 2014) and that the use of uncertified teachers and scripted lessons risk de-professionalising the teaching profession (Riep, 2017;Srivastava, 2017Srivastava, , 2013. Little in-depth focus is placed on teachers in their own right, as Singh (2021) argues for, by talking to teachers about teachers (for exceptions, see, McKay et al., 2018). To better understand the full consequences of LFPSs, we need to study how they affect teachers' experiences and agency, as well as their sense of significance as people and professionals.
In our study, we pay specific attention to the for-profit school-chain Bridge International Academies (hereafter BIA). BIA was established in 2009 in Kenya, where the company operates almost 300 unregistered LFPSs (BIA (Bridge International Academies), n.d.a.). While BIA has been praised for serving under-served areas, they have also been criticized for exploiting teachers and low-income families, having teachers perform non-professional duties, and using uncertified teachers (Riep & Machacek, 2016). BIA has a spatial division of labour where detailed scripted lesson plans are created in part by distant "experts" in Boston, USA, and local staff follow the manuscript in class on e-readers (BIA (Bridge International Academies), n.d.b.; Härmä, 2021;Riep, 2017;Riep & Machacek, 2016). This spatial division of labour in some second wave LFPSs is one that is important from a teacher profession status perspective. It poses questions, such as whether teacher knowledge and knowledgeable teachers are being replaced with technology and lesson plans produced elsewhere, and if so, what does this imply for their profession?
To discuss these questions, we structure the remainder of the paper as follows. First, we briefly go through what previous literature has said about teachers and LFPSs. After this, we elaborate on the concepts of spaces of dependence and engagement as tools to capture teachers' needs and agency. We then give some background on primary schooling and LFPSs in Kenya, and expand on sampling and the method used. After this follows the more extensive combined results and discussion sections, organised under four headings: teachers' changing spaces of dependence over the life course, teachers' relations in the surrounding community, teachers' relations with pupils and parents, and teachers' autonomy and relations with management. The paper ends with conclusions.

Teachers and LFPSs
As teachers' salaries are the largest cost in the provision of education, LFPSs need to cut down such expenses to attract low-income families. LFPS teachers have been found to have very low salary levels compared to PS teachers (Härmä, 2021;McKay et al., 2018;Singh, 2021), often earning on par with earnings in the low-income communities they work in (Edwards et al., 2015). Furthermore, LFPS teachers' employment situations are more precarious, lacking permanency and with few or no benefits (e.g., pension, healthcare; Bhatta & Pherali, 2017;Härmä, 2021;Kamat et al., 2016;McKay et al., 2018), and LFPS teachers have been found to have long workhours (McKay et al., 2018;Riep & Machacek, 2016). Reported pupil-teacher ratios in LFPS vary greatly, but are generally found to be lower than in PSs, which may give LFPS teachers a better chance at building relationships with their pupils (; Härmä, 2021;Unterhalter et al., 2018). Walford (2015) points to an oversupply of teachers in many countries facilitating the growth in LFPS, as the surplus of teachers pushes down pay levels. Another way to cut down on wages is through hiring uncertified teachers, often local youth. Such practices have in some studies been found to have positive impacts on learning outcomes, as LFPS teachers are being held accountable and are "invested" locally (Andrabi et al., 2008). In BIA the use of minutely detailed lesson scripts is aimed to facilitate the use of untrained teachers and achieve a similar standard across schools. Several scholars are however critical of this "dictatorship of the e-reader" (Härmä, 2021, p. 127), as it risks devolving the role of the teacher to that of a "functionary rather than a vocational professional" (Kagan & Gez, 2021: 396), circumscribing teachers' agency in the classroom. The use of unqualified staff, manuscripts and/or precarious employments further risks quality, deprofessionalization and teachers' status, as well as exploitation of youths and teachers (Härmä, 2021;Locatelli, 2018;Singh, 2021). These conditions in LFPS further contribute to a high teacher turnover, as LFPS teachers want to move on to better opportunities or government positions (Härmä, 2021;McKay et al., 2018; see also, Kagan & Gez, 2021, on BIA teachers' geographical and socio-economic mobility aspirations), indicating one way that teachers may exert agency.
In a study on teachers in registered LFPSs in Johannesburg, South Africa, teachers responded to open-ended questions in a questionnaire that they wished for better funding; adequate pay; better infrastructure and learning materials; a lower workload/more teachers; better supervision and management of LFPSs; further education and training; and more motivated pupils. Reasons mentioned, though not expanded on, were that poor working conditions made teaching difficult and demotivated the teachers (McKay et al., 2018), which chimes with findings in a large literature review on teachers' motivation by Han et al. (2016). A qualitative study with teachers in both private and PSs in Kenya seconded the need for resources and more interactive supervision of schools (Abuya & Ngware, 2016). This study also found that inadequate parental support and pupil mobility between schools were challenges hampering teachers' work and pupils' learning. While Abuya and Ngware (2016) did not always distinguish between LFPS and PS teachers' responses, the inclusion of both groups highlights some challenges that teachers more broadly experience in the education market, something we sympathize with and want to build on. Cox (1998) introduced the concepts of spaces of dependence and spaces of engagement to capture the content and form of spatial politics. Spaces of dependence make up the context within which an individual or a group find the social relations they need for their existence and "sense of significance" (Cox, 1998, p. 2), relations they cannot readily substitute elsewhere, e.g., family ties, employment. These relations can be more or less "local" and territorially bounded, and an agent can have several spaces of dependence on different scales (Cox, 1998), such as the national labour market and the rural ancestral home of many Kenyan urban dwellers (see, Onyango et al., 2021). Spaces of dependence can vary between agents, but can also be shared, such as workers' rights. Spaces of engagement are made up by the relations and networks through which the spaces of dependence are defended and secured. This is where shared spaces of dependence can come into play, if there is a common cause to rally around.

Capturing teachers' agency-Spaces of dependency and engagement
Spaces of dependence and spaces of engagement can help explore the socio-spatial reality of teachers, and the ways in which they may act to cope with, rework or resist perceived challenges relating to LFPS and the marketization of education (see, Katz, 2004). In the debate on LFPSs, arguments involving teachers' professionalism or wellbeing are invoked, and teachers are often portrayed as either victims or culprits, rather than as complex social beings. One way of disentangling teachers' rationales and actions is as agents with relations through which they, ideally, can fulfil their needs and wants, such as their employment relations. These relations they may act to secure through engaging with other agents, such as colleagues or parents. Effective spaces of engagement are often, but not necessarily, on a "higher" scale (Nespor, 2008), such as bringing a local labour struggle to a national labour union. However, agency and spaces of engagement are shaped by what options are viable and expected to yield results (Cox, 1998). Not all challenges to a space of dependence lead to joint action. Depending on the perceived feasibility of an action, workers can rather "play along" with and support their employer, if this is the safer option, say, for keeping their employment (Gough, 2010).

The primary education system and LFPSs in Kenya
In 1974 Kenya introduced free primary education. As a result, gross enrolment rose and remained over 100 % for the main part of the 1980s (UIS (United Nations Educational, Scientific, Cultural Organization Institute for Statistics), 2020). During the 1990s, enrolment ratios declined, partly due to demands to impose school fees in a "cost sharing policy", and partly due to economic decline and a perception of education as not leading to expected outcomes (Bold et al., 2010). As the national and international policy winds changed again, in 2003 the government re-introduced free primary education-free from tuition fees but not from costs for school uniforms and learning materials (Abuya et al., 2013; MoE (Ministry of Education), 2012). The increase in enrolment that followed once again put strain on government resources (Oketch et al., 2010; TSC (Teachers Service Commission), n.d.b.; UNESCO (United Nations Educational, Scientific, Cultural Organization), 2017). At present, there is a shortage of around 38,000 teachers in public primary schools (TSC (Teachers Service Commission), n.d.b.).
All teachers in all schools are supposed to be registered with the Teachers Service Commission (TSC): the minimum required to register with and be employed by the TSC as a primary teacher is to have a P1 certification (TSC (Teachers Service Commission), n.d.a.). However, implementation of this policy is difficult, as many LFPSs are unregistered and use uncertified teachers (Zuilkowski et al., 2017). In 2017, there were over 750,000 registered teachers, of which around 40 % were employed by the TSC. The remaining were engaged in private schools, NGOs, and different government agencies (TSC (Teachers Service Commission), 2017). As for the school system, there are two official categories of formal primary schools: public and registered private (RoK (Republic of Kenya), 2013). In line with the marketization of education in Kenya (Edwards et al., 2015;Riep, 2017), the number of private schools has increased rapidly (Diagram 1), as has private school enrolment (Table 1). The number of public primary schools has also increased, but not at the same rate.
Besides formal private and PSs there are non-formal private schools registered under Alternative Provision of Basic Education and Training (APBET). APBET regulations are somewhat lax to facilitate education in hardship areas. In APBET schools only 30 % of teachers need to be certified at the time of registration, with the requirement that they be trained within three years from registration (MoEST (Ministry of Education, Science and Technology), 2015). In addition, there is a large number of nonregistered private schools, often a result from not meeting standards (e.g., land ownership) and a difficult registration process (Stern & Heyneman, 2013). This makes it is hard to estimate the total number of schools.
LFPS is not a homogeneous category (Srivastava, 2007). It includes community schools, which in this case had an open enrolment policy similar to that of PSs; schools run in sole proprietorship; small school-chains; and large corporate-backed school-chains. Some schools are for-profit, some are not, some are registered as private or APBET, some are not, some have external funding, some do not, some schools allow pupils in class despite missing payments, and some refuse. What the LFPSs have in common is that they are relatively low-fee and not owned or run by the government, and subsequently do not have TSC employees.

Method and data
A qualitative method was used to glean more in-depth perspectives from teachers on their own situation, and, importantly, in their own words. Two areas were chosen for this study: an informal settlement in Nairobi and a rural town located three hours' drive from Nairobi. The informal settlement was chosen because this is where a majority of LFPSs are found (Oketch et al., 2010). The rural town was chosen as it is situated in a sparsely populated area in a geographically and culturally different context, although the town lies along a tarmac road and has access to electricity. Both areas had to have PSs and BIA schools. BIA was of interest as "second wave" LFPSs using scripted lessons, but also served as an indicator of there being a LFPS market in the area. This limited the possible locations, as BIA schools are primarily found in more densely populated areas (Härmä, 2016). While the sampling of schools aimed at a variation of types and sizes of LFPSs, no small, independent LFPSs were included there. The two rural PSs, one of the urban PSs and the large LFPS were contacted beforehand about interviews. The remaining schools were visited without prior arrangements, although the urban schools were visited with a well-known local social worker as a link to the managers and/or teachers.
Accessing teachers was based on convenience at the school managers' 1 and teachers' discretion, but with the request for both genders, different ages, levels of education, and type of employment considered. On arrival in the schools, the first author and the two female Kenyan research assistants introduced themselves and the project to the person in charge, asking permission to interview up to five primary school teachers about their work life and the teaching profession. If interviews were granted, the first author and two assistants introduced themselves and the project to the interviewees. The interviewees were guaranteed anonymity and told that they could stop their participation at any time. Permission to record was asked and granted in all cases but one, where notes were taken instead. No names of teachers were included in the recording. The interviews were structured around themes related to a) the teachers' career and work, b) changes and challenges to their work, in part related to marketization, c) agency, and d) general views on education. When heartfelt topics were brought up by the teachers this was encouraged. Recordings were transcribed verbatim, with names of places given a code, to avoid identification of the schools and teachers. All transcripts were then coded using NVivo to facilitate overview and analysis. The codes related themes in previous literature on LFPSs-e.g., pay levels, job security, accountability, to the teachers' needs, relations, challenges and ability to act. Other themes found, such as pupils' mobility, were brought up by the teachers.
All in all, 36 teachers were interviewed. Yet, one interview in a BIA school was excluded, as the area manager insisted on supervising the interview, and no questions on private schools or marketization were allowed. This left 35 interviewees (Table 2). An interview lasted 40 minutes on average. Most interviews were conducted in Nairobi. As for gender division, in Nairobi nine female teachers (five in PSs, four in LFPSs) and 13 men (four in PSs, nine in LFPSs) were interviewed.
In the rural town, seven female teachers (five in PSs, two in a LFPS) and six male teachers (five in PSs, one in a LFPS) were interviewed.
Teachers included in PSs as "non-TSC" are teachers hired by the school board or the parentteacher association and paid by a supporting organization or by the parents. In one urban PS the teachers employed by the supporting group outnumbered the teachers employed by the TSC. Thus, the distinction between a PS and a community/private school is not straightforward: two PSs in Nairobi were run in tandem by the state and religious groups. The teachers interviewed in Nairobi ranged from young to senior, but the majority, 16 out of 22, were relatively young. This is explained by 13 of 22 interviews were carried out in LFPSs, where staff generally is in the early stages of their career. Another reason is that the schools are in an informal settlement with socio-economic challenges, meaning more senior teachers employed by the TSC may have been granted transfers (see, Luschei & Chudgar, 2017). The teachers interviewed in the rural town ranged between young and senior, with most being middle age, though all three LFPS teachers were young. Some teachers lived in the area with their families, some had their families in nearby areas, and some in their home counties.

Changing spaces of dependence over the life course
The teachers' spaces of dependence changed through their career and life. Nearly all teachers wanted to be employed by the TSC. The exception was a senior teacher that ran his own LFPS in the absence of PSs in his area, feeling obligated to stay in his community. The main reasons for wanting state employment were the better and pensionable income, the ability to take out loans, and job security, extrinsic motivations which according to Han et al. (2016) appear to be more important in low-and middle-income countries. This TSC teacher compared private with public employment: Working in a government school is good, because of the security of tenure, because here we have a lot, we have insurance, [you are] sure of your work. Nobody will one day wake up and tell you "hey, no job today!" That was my experience in private schools. Because that's somebody's business, they can tell you, you do something funny they tell you "hey, get out of my premises!", so you're not sure. But here in government [schools] there is a security in the job. And as such also the peace of mind I was talking about. (Young male, degree in education, TSC-employed, Nairobi) These sentiments were expressed by both PS and LFPS teachers and echo findings on the discrepancies between private and public employment, as well as on aspirations for better employment (Härmä, 2021;McKay et al., 2018). The interviews also brought the need for "peace of mind" to our attention, here in relation to stability of employment. Being able to afford a safe and peaceful home, where the teacher could rest, and know that their family was well, created a mental state where they felt they could focus on the learners.
To become employed by the TSC means accepting a posting anywhere in the country by reporting to the school in question within 30 days. Failing to show up means the employment offer lapses (TSC (Teachers Service Commission), 2015). Moving for public employment could be disruptive to the teachers' family relations, but the space of dependence of public employment was prioritised by the teachers. The teachers knew this relocation was expected of them, and they all expressed a willingness to teach pupils anywhere, as "they are all Kenya's children" with a right to education. Thus, all teachers, bar two senior teachers approaching retirement, said they would take a posting anywhere in the country if the TSC sent them. The exception, where only two young, male teachers would go if deployed by the TSC, was the northeast, as "outsider" teachers had been killed by Al-Shabaab there (Makong & Wanzala, 2018).
The more senior teachers had entered state employment immediately after graduation. In later years, however, despite PSs being under-staffed, graduates have had some years of keeping afloat and gaining experience before being employed by the TSC, something also found by Bold et al. (2013). Newly graduated primary teachers in Kenya thus have a choice between finding employment in a private school, as a non-TSC/contract teacher in a PS, and seeking other employment. Private employment as teachers could quite easily be found where the teachers lived, or wished to live, most teachers reported, as there were plenty of LFPSs and high teacher turnover. LFPS employment had no formal job security, however, with salaries predominately well below half in public employment (see also, Edwards et al., 2015). As has been found elsewhere (Locatelli, 2018), the teachers found themselves competing with secondary school leavers, who were more desirable for employers, as they would settle for less pay and bemore willing to accept poorer working conditions.
Several of the LFPSs teachers in Nairobi were working where they had grown up, some even in their old schools. This was not the case for the majority of the BIA staff in either area, as BIA, like the TSC, had a more centralized teacher employment and management structure. This meant that BIA teachers could be deployed where there was a vacancy, but also that they could request transfers, something that Kagan and Gez (2021) found gave BIA schools an edge over other LFPSs. To the teachers interviewed in this study, such requests had however been denied thus far. The idea that LFPSs generally hire locally may thus not apply to all kinds of LFPSs, meaning that gains from hiring locals found in some studies of LFPSs should not be assumed to be universal (Andrabi et al., 2008).
As mentioned above, nearly all young and middle-aged teachers prioritized stable employment over location, but most teachers had an ideal area to work in. This was either Nairobi, the area they now lived and were settled in, or their home county. This depended on when and where they were at in their lives and careers. If they had not yet started a family, moving was viable, and expected. Sometimes moving was even welcomed, as it would bring them to new places and experiences. At a later stage of their career, when already employed by the TSC, a specific place to live and work mattered more, as they had families, or started preparing for retirement. The matter of employment and place is illustrated by a woman teacher in a rural school, who, by request, transferred to the area she was originally from: Of the PS teachers in the rural area several were older and had requested transfer there, thus being quite content with their place. In Nairobi, more teachers (both public and private) were at earlier stages in their careers, thus more open for transfers instigated by the TSC, but not generally discontent with the area they were in as such. In this way the teachers' spaces of dependence for employment and securing a career and decent life shifted with time and feasibility. This shift was generally from the local and finding a private employment, to the national in accepting a posting anywhere to secure a state employment, to being in an area of choice, where living would be cheaper and family closer, towards the end of the career.
It is not as if place did not matter in the years before retirement, but rather that teachers' main consideration was being able to sustain themselves. A LFPS employment was generally considered better than an employment outside of education, but a PS job would give higher income and security. To achieve this security, teachers had to play by the TSC rules. It was a way of acting on what appeared a feasible option (Gough, 2010), rather than individually protesting the system of deployment and transfer, thus risking the employment relation so long sought. The fight for teachers not to be "delocalized" and separated from their families was rather pursued through the Kenya National Teachers' Union (KNUT), where all TSC-employed teachers interviewed were members (NTV Kenya, 2018).

Relations in the surrounding community
To many of the teachers teaching was a calling, but they still needed their own lives to function and a need to feel safe. Thus TSC and BIA teachers that worked in and/or had moved to an area they were not originally from expressed a need for security outside of the school, and a fear of difficult relations with the surrounding community. Several teachers in the rural area expressed that some members of the community did not value education. For some teachers, this made them feel like they were in a difficult situation in doing their job while maintaining good relations. As this female teacher put it: [I just need] at least encouragement from the community members, and unity among me and the people around the school, or the community. Because right now, some parents, or the community around, do not understand why the children should be in school. It's like we are forcing them. (Middle age female, P1, TSC-employed, rural area) Still, this teacher did not feel that the frail relations with the community were a threat to her, but rather something she had to navigate to be able to do her job well and make sure pupils attended school. The situation was worse for the rural BIA-teachers who felt like they were outsiders and struggled to get on with the locals, especially when doing marketing, as they did not speak the local language, and even if the language barrier was overcome they were ignored. As the BIAteachers already felt insecure in the area, interactions with the local community felt perilous. That the school owner was unknown and not present in the community did not help the BIA-teachers' endeavours to try to attract more pupils: This stands in contrast to Kagan and Gez's (2021) findings of BIA schools having an international allure because of their white owners. While none of the BIA teachers spoke of direct repercussions if they failed to attract pupils during marketing, not attracting pupils affected the teachers emotionally, as their classes were nearly empty. As the teachers, and the owners, were not wellknown in the area, the teachers had few relations to draw on to engage the parents. The only other teacher assigned to do marketing was an uncertified young male working in a small LFPS in Nairobi. While he was not a local, his school had a very engaged head teacher from the local area with whom he did the marketing, which may have helped him connect with the community. In contrast to Riep and Machacek (2016) findings, the teachers reported getting a sum for the extra time spent doing such tasks.
In the urban area, teachers did not feel out of place in the same way as the rural BIA teachers. The majority of urban LFPS teachers were well at home in the community as locals. The downside to this for some young male teachers was that they felt they were expected to live up to a certain material standard and behave in certain ways to be role models. While this was difficult on their meagre earnings, they connected their profession with their past and with the ability to give back to their community, something serving to motivate them (see, Andrabi et al., 2008;Han et al., 2016). These relations outside of school mattered for teachers' ability to do their job, and affected their sense of importance and wellbeing. Of most direct importance for teachers' ability to do their job well were however their relations in school.

Relations with pupils and parents
When asked what the teachers needed in school to be able to do their job well all teachers stressed their relations to pupils and parents. Generally, the smaller LFPSs had relatively few pupils per grade, ranging from zero to 30. In PSs teachers expressed frustration or resignation over the very large number of pupils per class, sometimes as high as 100, and the "constant" admission of new pupils. Such large numbers negatively affected teacher-student relations, as too little time could be spent with individual pupils, and correcting homework. This meant that the existence of LFPSs was partly positively viewed by PS teachers as it alleviated the pressure on them. While relations with pupils and pupils' progression were not always easy and rewarding, for most of the teachers they were their primary motivators in work (see, Han et al., 2016). Parents had a potential for both positive and negative impacts on the teachers' work, as an uncertified teacher in BIA put it (a sentiment expressed by all teachers): If the parents become supportive, then you become a good teacher. You know some parents sometimes they make you lose that urge of being a teacher. But if the parents are very much supportive you become like . . . you like the job and you also like what you are doing. (Young female, non-P1, BIA, rural area) To the LFPS teachers, however, parents could shift between allies and threats to their spaces of dependence. While accountability is important, and was welcomed by the teachers, it was clear that exam scores were the main way that schools measured themselves against one another, similar to what Lindsjö (2018) found in Tanzania, and financially able parents used exam scores to choose where to place their children. For LFPS teachers, exam scores and parents' contentment could thus have direct impacts on their employment. Teachers in PSs sometimes referred to schools' test scores too, but did not place as much emphasis on them for the ability of the school to attract pupils as the private school teachers did, as "pupils come anyway". Scholars have pointed out issues of lacking accountability in public education (see, Dixon, 2012;Härmä, 2021), which several of the LFPS teachers also lamented, but LFPSs', LFPS teachers' and parents' focus on scores created something of a too mobile market (see also, Abuya & Ngware, 2016).
Nearly all teachers, in both areas and in both LFPSs and PSs, experienced that there was a high pupil mobility, and it affected relations with pupils negatively. The high mobility of Kenyan primary school pupils between schools made these relations even more fleeting. Even if high mobility pupils would "only" be 5-20 % of pupils (as Maluccio et al., 2018 found), they affected the progression for the remaining pupils, and the ability of teachers to do their work. This was both demotivating, and, if it hampered pupils' progression, could be a danger to the LFPS teachers' employment. It seemed something of a vicious circle.
Absent pupils because of unpaid fees was a problem in the rural BIA school, as well as in BIA and some of the other LFPSs in the Nairobi area, depending on the school policy. Sometimes in a class, the majority of pupils could be sent home for fees. "We call those children 'not allowed'", one BIA teacher reported, as pupils were not allowed back in class until the fees were paid. This has been pointed out elsewhere (Riep & Machacek, 2016;Unterhalter et al., 2018). As many families in both the rural and urban area struggled financially, paying fees could be very hard (see, Edwards et al., 2015). The teachers were aware of this problem. A pupil not allowed in class obviously affected him/her, but also pupils still in school, as well as the teacher, who would be unable to progress with the syllabus as planned: When you're in a class where there are 20 pupils and 15 are sent home, then you remain with the 5. It's terrible. You are used to that big number, then others are there, you go to class, you can't deliver everything to your fullest, because you are knowing you are delivering the thing and half of the class are not there and you have to repeat it when they come. So that is a challenge. (Young male, P1, LFPS, Nairobi).
The issue of fees also created a difficult situation for teachers to maintain good relations with parents while convincing them to send their children back to school, further straining the relation of educational provision between teachers and "clients" (see, Martin, 1998). Allowing pupils in class when not allowed was a type of seizing agency and reworking of the system, trying to correct perceived injustices (see, Katz, 2004), with the risk of falling out with management or being dismissed. On the positive side, this could build alliances with parents. Teachers in LFPSs and PSs in both study areas expressed concerns for the wellbeing of the pupils, difficulty progressing with the syllabus and maintaining good and formative relations because of the high pupil mobility. This threatened teachers' spaces of dependence and made work harder, thus threatening their ability to do their job to their fullest. It also affected LFPS teachers' work relations, as exam results may drop if they were unable to cover all subjects.

Autonomy and relations with management
Except for teachers' needs and agency in relation to their employment options as well as in their relations with pupils and parents, the teachers themselves and their relations with management deserves attention to better grasp teachers' agency and why they acted in certain ways in relation to LFPSs. In the relations making up the spaces of dependence in-/outside of school, teachers' education, experience, professionalism, and commitment mattered to their sense of significance and ability. These traits also mattered for them meeting challenges.
Most of the P1-certified teachers expressed great confidence in their own teacher training. The certified teachers all strongly believed that all teachers in all primary schools should be at least P1certified, as their education had taught them to adhere to the curriculum, had given them confidence, had enabled them to handle situations in class, and had given them a better understanding and knowledge of how to handle pupils in different ages. While the quality of teacher training in several countries is low (Härmä, 2021), the teachers were still proud of and felt their education had been valuable to their work. In contrast, the use of untrained and uncertified teachers was seen as risking the credibility of the profession, as: . . . some parents have this problem of including . . . Just, say if they are commenting on something they'll just say 'those teachers'. They will not specify which teacher. Like, for example, if a teacher is untrained, you know he will not be able to handle a child the way a trained teacher is. (Young female, P1, non-TSC employed, Nairobi) While the teachers' knowledge levels are not the subject of this study, their sense of professionalism was threatened by uncertified teachers. Further, it was to most of the teachers an absolute given that they themselves prepare their lessons as they see fit. Even the teachers that worked at BIA, whether currently or previously, who had to follow a standardized script, reported having, or taking, the liberty of shaping the lessons to various degrees, something that a report by EI (Education International), KNUT (Kenya National Union of Teachers, 2016) also noted. For this teacher, using her own knowledge was in accordance with her sense of professionalism and ability to teach. For others, it was a matter of not trusting the content on the Nook. As one uncertified BIA teacher said: Even man is to error, and those things are made by man, so if you depend on them fully the children may end up astray, so just use your wisdom 100 %. (Young female, uncertified, BIA, rural area) A question that this raises is what evaluations are actually saying about a model strictly following scripts, if it turns out that many of the teachers, trained or not, are not really using the scripts in class? Leaving the script can be seen as a strategy where teachers were not out to challenge the company or its model, but reworked the system by doing as they saw best. The BIA teachers said they went beyond the scripts either to make sure even the slower learners could keep up or to focus more on revision for the end-term tests. Though only five BIA-teachers were interviewed, the teachers' disregard of the Nook for the sake of the pupils is contrary to observations made elsewhere (Härmä, 2021;Riep & Machacek, 2016). While the strategies were different, and not necessarily mutually exclusive-getting a deeper understanding from all, or focusing on the test,both aimed at improving results.
Cheating and teaching-to-the-test in LFPSs was reported as a well-known fact by most teachers in the Nairobi area (and has been reported in other contexts, see, Härmä, 2021), although no one explicitly admitted to such practices themselves or wanted to point out specific schools where this occurred. Spoon-feeding for exams in part leads to a type of de-professionalization, according to one teacher: I think teachers who are in school where you are only told to teach those things which can only be tested in the national examinations so that the school gets a higher grade on the general performance, so that they can in turn get more pupils from other schools to come . . . that is tragic. Because this turns those teachers into people who can't think for themselves. You're just doing things because you've been told to, not because you think they are right or you should do them, you just do them anyway. (Young male, P1, LFPS, Nairobi) Who would choose which strategy is hard to discern, but apart from wanting to do a good job, and to remain employed, some teachers seemed to have a more holistic and caring approach, whereas others had a more pragmatic ditto, acknowledging the heed paid to test results in wider society. Many teachers understood why parents would move their children if they found the school was not performing well in tests. This led to stress and high levels of insecurity, but also to a sense of pride when "their" pupils performed well. Teachers in LFPSs were however very much part in driving the culture of schools being measured by exams, as this is by many perceived as the only feasible option (Gough, 2010). Specifically, they feared that if they did not conform to this mode of accountability they may lose their job.
Good communication and feeling that the management, the head teacher and other superiors "had their backs" was important for teachers, both in LFPSs and in PSs, mainly in relation to government inspections. Teachers in LFPSs to a higher degree reported being supervised by managers and head teachers, although some LFPS-teachers reported large portions of freedom as long as results were good. PS teachers on the other hand reported spending a lot of time on administrative tasks similarly aimed at keeping them accountable, but in a less direct way, and with little consequence. Those under close supervision generally felt this as a stress and something that hampered performance. Teachers being held accountable has however been ascribed as part of why LFPSs perform better (Andrabi et al., 2008). Several LFPSs teachers expressed the desire to work in a PS because of perceived less supervision there. In BIA schools, upper levels of management appeared "duped", as teachers turned to their Nooks when there was an inspection by the company. Similarly, teachers could allow "not allowed" pupils in class if the immediate management was on board with this, even if the owners did not condone this. In this sense, the distance to the management and owners mattered, something Härmä (2021) also notes. BIA teachers generally reported a closer relationship with their immediate management compared to area managers or distant owners. This was true in other LFPSs as well, where the owner was not working in the school as head teacher/manager. In those cases, the relations differed depending on the personal relationships developed and the (financial) pressures put on the school.
The networks of relationships with colleagues were, similar to relations with management, important for the teachers' sense of wellbeing, and being able to influence their work situation (see also, Han et al., 2016). Teachers in PSs are also highly unionized in Kenya, but few of the teachers interviewed were active members, apart from paying their membership and voting in elections of officials. This can be linked to Lindsjö's (2018) findings that few teachers in Tanzania have the time to engage in education politics. Although all certified teachers are supposedly allowed in the KNUT, the union was by all but one senior LFPS teacher perceived as only for public teachers. The feasibility of making one's voice heard in the union was perceived as very small, although some, often more senior, teachers knew the channels to use, if they wanted to. The teachers' spaces of engagement, the "arenas" for their struggles against challenges to their spaces of dependence (see, Cox, 1998), were thus generally on a much more local scale-within the school and community.

Conclusions
The debate and literature on LFPSs are polarized, and within them, the representation of teachers is often simplified. Our findings showed a more nuanced and complex situation. To begin with, Kenyan teachers in both LFPSs and PSs shared very similar needs and views of themselves as professionals, but had to navigate different challenges at different stages throughout their work/ lives. The experiences of the teachers in our study add insight to findings in previous studies regarding risks of de-professionalization, exploitation, and lower status of teachers in the marketization of education. The years waiting for public employment with low pay, precarious employment relations, competition, and high pupil mobility challenged LFPS teachers' mental wellbeing and integrity, and ability to progress in their work/lives. Further, while LFPSs have alleviated pupilteacher ratios in PSs, teachers in the latter were still negatively affected by high pupil mobility and struggled to build relationships with pupils. This study also showed contradictory findings compared to previous studies. While some LFPS tried to steer the teachers work, LFPS teachers' agency and sense of professionalism, connected to their education and training, translated into practices of reworking demands for their own and their pupils' sake.
To sum up, once we go beyond the simplifications and instrumental perceptions surrounding teachers, we find largely similar groups with agency and a strong sense of professionalism. This is something worth building on, rather than trying to quench through scripts and narrow demands for high scores. To further our understanding, future research on teachers needs to address the multitude of intertwined actors in a longer perspective, also including labour unions, technological developments, teacher education, and recurring changes in national education policies. Applying a life-course perspective with teachers having experiences from both private and PSs would arguably give valuable insights.