Low-Fee Private Schools, the State, and Globalization: A Market Analysis within the Political Sociology of Education and Development

This study investigates the emergence and supply-demand dynamics of a market for low-fee private schools (LFPS) at the level of early childhood care and education (ECCE) in a slum of Lusaka, Zambia. Based on data collection over 1.5 years, the study reveals that, despite a government policy to support ECCE, over 90% of ECCE centers are private; that school operators tend to be former teachers, businessmen/women, and religious leaders; and that LFPSs charge, on average, 2.5 times as much as government ECCE centers for tuition, not including additional indirect costs. The paper discusses how teachers in LFPSs are caught in the middle, making less than the average income earned by others in the surrounding slum, and are unable to afford LFPS fees themselves. Importantly, the paper highlights that lower income quintiles spend a greater percentage of their income on ECCE, and that a majority of families in the study must make tradeoffs between ECCE, food, housing, and other basic expenditures in order to afford private ECCE, which is a necessity given the inadequate supply of government ECCE centers. In addition to addressing school strategies for keeping costs down, this study reports on parental decision-making when it comes to school selection. Finally, beyond a straight market analysis of LFPSs at the ECCE level in Zambia, this article also comments on how this market fits into the dialectical nature of local and global contexts. That is, it draws attention to the workings of the Zambian state and its precarious position in the global capitalist economy.


Introduction
One major trend in education around the world in recent decades is the increase in private schools directed at the poor and relatively less affluent (Srivastava, 2013a;Verger, Fontdevila, Zancajo, 2016). On one hand, there is clear evidence that, in the context of globalization, many international actors are promoting these types of schools not only as a means to make profit but also as a way to meet the excess demand for education that governments are unable to meet. Key actors in this group include corporations, international aid agencies, development banks, philanthropists, and edu-entrepreneurs (Junemann & Ball, 2015;Nambissan & Ball, 2010;Niemerg, 2013;Riep, 2015;Srivastava, 2016). Investors see great potential for profits by serving these consumers, and they have referred to the 4 billion people in middle and lower classes in low-income countries as the "fortune at the bottom of the pyramid" (Prahalad & Hart, 2002). However, on the other hand, it is important to highlight that this trend is not driven solely by the international actors mentioned above; rather, the emergence of low-fee private schools (LFPS), as they are frequently known, is often the result of organic activity at the local level, where enterprising individuals respond to the lack of government service. In the post WWII-context, governments in many low-and middleincome countries have not, for one reason or another, been able to meet the demand placed on public schools that has resulted from growing populations, urban migration, and increased expectations related to the level of education that should be provided for all citizens.
This paper engages with this latter, more organic form, of LFPSs. It does so by presenting insights from a study of LFPSs in a context that has been neglected in the literature-namely, Zambia. Specifically, the paper provides empirical insights along two dimensions. First, it explains the emergence of LFPSs in the capital city of Lusaka. Second, it characterizes the LFPS market in one particular slum of Lusaka known as Mtendere. As further discussed later, many authors have investigated these aspects of LFPSs in one way or another; however, a feature of the present paper that sets it apart from many LFPS studies is that it seeks to connect the findings on privatization to the twin issues of globalization and marginalization (just as do all the papers in this special issue). Thus, our interest is not only to characterize the nature of the LFPS market but also to reflect on the ways in which this form of privatization is enabled by political and economic globalization more generally, in addition to considering the ways that LFPSs combine with local contextual features to exacerbate marginalization. In so doing, we hope to provide insights that may be meaningful when it comes to understanding the ways that the global and the local are connected, and with what implications.
A further feature of this study that sets it apart is that it is focused on the pre-primary level, that is, on the level of early childhood care and education (ECCE). Most studies to this point have focused on the primary and secondary levels. The results shared here are the product of a research project for which data collection spanned 1.5 years (see below for details). The research questions that guided our study were: 1. Why did LFPSs emerge in Lusaka? 2. What is the context and nature of the LFPS market in the study site of Mtendere?
When it comes to the "nature" of the market in question, the focus will be on characterizing supply and demand, as well as their interaction, with the analysis of that interaction being informed by an in-depth understanding of the context. In shedding light on these issues, this study contributes to the growing collection of studies in this area from Africa, Asia, and Latin America. The following sections discuss the relevant literature, our conceptual approach and the methods employed, before turning to a discussion of context, findings, and implications.

Literature Review
The learning outcomes of LFPSs have received considerable attention, and by now advocates and critics each have their preferred studies to support their positions as to whether LFPSs provide a better quality education. However, an extensive review of the literature arrives at the conclusion that the evidence is ambiguous at best because of the inability to control for student background characteristics (Ashley et al., 2014). Controlling for student backgrounds is essential, since even among the poor it is typically the relatively more well-off who are able to afford to send their children to school in LFPSs (Alcott & Rose, 2016;Fennell & Malik, 2012;Harma, 2011;Languille, 2016;Singh & Bangay, 2014). The point here is that, despite the best efforts of researchers, it is very difficult to attain control and treatment samples that are comparable due to the fact that parents with relatively more resources, social capital, etc. self-select into the private schools.
On affordability, most evidence is neutral or negative and indicates that "welfare sacrifices are made and continued attendance is difficult to sustain" (Ashley et al., 2014, p. 2). These sacrifices are often necessary given that LFPSs frequently operate in slum contexts where there are few if any government schools, which tend to be cheaper options. That is, though government schools also charge tuition fees (in the context of insufficient government budgets for education), LFPSs cost significantly more because they have a variety of other fees as well, related to registration, examinations, books, etc. (Harma & Rose, 2012).
Despite high fees in relation to household income, LFPSs have proliferated in low-income countries. A main reason for this is that states are "rather passive when it comes to addressing new educational demands" (Verger, Fontdevila, & Zancajo, 2016, p. 89). More fully, the growth of LFPSs has occurred in the face of rising demand thanks to internal migration, growing populations, settlement in unplanned areas, the removal of fees from public schools, government indifference, and higher expectations with regard to the level of schooling one should attain (Edwards, Klees, & Wildish, 2017;Oketch, Mutisya, Ngware, & Ezeh, 2010;Srivastava, 2008;Stern & Smith, 2016). Additional reasons that have been documented relate to frustration with the quality of government schools, belief in the superiority of private schools, and a preference for English-language or religiously-oriented instruction (Harlech-Jones, Baig, Sajid, & ur-Rahman, 2005;Harma & Rose, 2012;Singh & Bangay, 2014;Stern & Smith, 2016). One tragic consequence for the state of ignoring the quality of public education is that it eventually contributes to its own marginalization in the eyes of its citizens-that is, marginalized populations end up contributing to the marginalization of the public education system when they begin to prefer private schools, even if that means paying more for what is often still a low-quality education where "children may not be achieving basic competencies" (Ashley et al., 2014, p. 1). Such a reaction on the part of the poor makes sense when one considers that government schools in slums and rural contexts are in dire need of resources and support, if they are available at all. While many aspects of the LFPS phenomenon have been explored, there are fewer studies that have attempted to map the supply-demand dynamics of LFPS markets within their structural constraints. Those studies that do look at market dynamics tend look at how parents and students choose among schools and exercise their "voice" (Ohba, 2013;Srivastava, 2007), how LFPSs and parents are engaged in a delicate relationship where they depend on each other (Edwards, Klees, & Wildish, 2017), and how students navigate the social worlds of LFPSs (Sucharita, 2014). These studies provide valuable insights on how markets work in very low-income settings, and this is an important area for further investigation. However, in the present study, we seek to engage in a different kind of mapping. In a particular low-income settlement in Lusaka, Zambia, we seek to characterize the supply and demand sides broadly and then to provide insight into the global-local structural dimensions that constrain and enable the operation of LFPSs and the ability of families to participate in them. As noted earlier, we are, finally, interested to reflect on the implications of the structure and dynamics of this LFPS market for marginalization (of families, of teachers, of public schools, etc.) more generally. In pursuing this work, we follow the example set by similar studies (Harma, 2016;Moschetti, 2015;Srivastava & Noronha, 2016;Tooley & Dixon, 2005). The market portrait and analysis presented here is particularly important given that research is only just beginning to unpack the high preschool enrollment rates across Africa that have been observede.g., 93% of children 3-6 in Accra, Ghana-and which are largely dependent on private schools (UBS Optimus Foundation, 2014).

Conceptual Approach
In accordance with the purposes of this study set out above, we employ a two-level conceptual approach. As will be seen, the first level of our conceptual apparatus corresponds with our goal of characterizing the supply and demand sides of the market for LFPSs, as well as their interaction. This conceptual level-focused on market dynamics-corresponds with the empirical aspect of our study, in that we investigated these dynamics directly through our data collection. The second conceptual level of our approach corresponds with our goal of viewing LFPSs in relation to political and economic globalization more generally. The theoretical perspective on which we draw here-focused on the relationships between education, the state, and globalization-informs both our initial characterization of the context of the study as well as the analysis that is presented at the end of the paper, where we step back and assess how larger structural relationships and constraints connect with the emergence and operation of LFPSs in Zambia.

Market Analysis
This article focuses primarily on documenting and analyzing the functioning of the LFPS market at the ECCE level in Lusaka, Zambia. This first level of analysis is informed by the basic concepts of market analysis from microeconomics, grounded as they are in the theory of the utilitymaximizing consumer and the profit-maximizing firm (or, in this case, private school). There are four dimensions to such a market analysis. In describing these dimensions, we adapt the tenets described in readings of the CFA Institute on microeconomic analysis (CFA Institute, 2011), though it should be emphasized that our approach is common to any basic text on market analysis from the perspective of neoclassical economics.
With that in mind, the first dimension of interest entails an analysis of market structures. Examples include the type and concentration of firms in addition to relevant governmental regulations. The second dimension is the supply-side analysis. Since supply is affected by the prices that can be charged for a good and the costs that are required to offer that good, we document the fees that schools charge and we are attentive to key factors of production (e.g., teacher salaries, the differential cost of students with disabilities) and how schools respond to them. Finally, for this second dimension, we are attentive to school revenues (to the extent possible, given the difficulty of gathering this information). The third dimension of the market analysis is dedicated to the demand side. Here, we begin by characterizing who the consumers are. In our case, since the spending decision is affected by family context (i.e., since 'the consumer' is not the child nor an individual acting in isolation), we document the household context of the children in our sample. As in typical microeconomic analysis of markets, we then focus on both the consumers' ability to purchase a good (i.e., access to LFPS) based on income as well as consumer preferences. This is followed by a consideration of what the consumer would do in various circumstances. In our case, this means analyzing the spending decisions and tradeoffs that families make at different income levels.
In the fourth step of market analysis, one is attentive to supply-demand interaction within market structures. For example, this means analyzing the ways that parents and schools engage in the market, the strategies they use in doing so (e.g., to choose schools, to enable enrollment by their children, or, in the case of schools, to attract families), and the ways that each are dependent on the other-with both being dependent on the larger political-economic context. This fourth step in the market analysis is the critical component for our study. We characterize the previous dimensions in order to be able to analyze how the supply and demand sides interact (that is, to understand the nature of the LFPS market), and with what implications for marginalization, with both of these goals being central to our study and to this special issue of EPAA.

Political Sociology of Education and Development
Beyond a straight market analysis of LFPSs at the ECCE level in Zambia, we are also interested to comment on how this market fits into the relational and dialectical nature of local and global contexts. In the words of Dale and Robertson (2012), "This means placing education policies into a series of contexts … and [emphasizing] that these contexts are themselves understood, not as neutral backdrops …, but as co-constitutive" (p. 35). Towards this aim, we draw on a theoretical perspective that has been labeled the political sociology of education and development (Arnove, Torres, Franz, & Morse, 1996). 2 This theoretical perspective starts with the dynamics of the global (or "world") capitalist system, especially the relationship between core and peripheral countries, where, historically, resources from the latter have been directed to the former, producing relationships of dependency-both in terms of market dependency and dependency on international aid (Arnove et al., 1996;Ginsburg, Cooper, Raghu, & Zegarra, 1990;Wallerstein, 2010). The point here is that this perspective is preoccupied with the structural constraints associated with the evolution of the global capitalist system and the ways that these constraints affect education.
In their own words, the conditions of education reform refer to "historical forces, institutional contradictions, and contextual factors, both national and international, that shape the possibilities of education and social change occurring" (Arnove et al., 1996, p. 140). The focus on institutional contradictions is key, since the education system is caught in the awkward situation of legitimating the inequality that inherently results from being part of a capitalist system (since education holds the promise of social mobility) while also contributing to the reproduction of that unequal system (since education simultaneously socializes and sorts students into different career tracks; Arnove et al., 1996). Importantly, to retain their legitimacy, "conditioned" capitalist states (i.e., capitalist states that are dependent on their relations with the global capitalist economy) attempt to mitigate the fundamental contradictions of capitalism by increasing or enhancing the institutional functions of the education system, such as by offering ECCE (Arnove et al., 1996;Ginsburg et al., 1990). However, the fact that the state is also a site of contestation that is built on alliances among key economic and political actors (e.g., state officials, multinational corporations, international development organizations, and the elite factions of the national bourgeoisie) means that the interests of these actors often prevail as they angle for influence (via the state) in order to benefit themselves, with the implication being that "the state is often unwilling or unable to mobilize enough resources to make public education … generally available" (Arnove et al., 1996, p. 142). Moreover, as Arnove et al. (1996) state, to employ this perspective (on the political sociology of education and development) means to be attentive to whether or not educational and social change "benefits the least privileged members of a society" (p. 140). This perspective is thus appropriate for the present study, given the previously stated focus on marginalization.
To align with the conceptual approach delineated here, before presenting the market analysis, we begin by characterizing the political-economic constraints that, historically, have influenced the context of the local LFPS market at the ECCE level in Lusaka. First, however, the following section addresses methods of data collection.

Methods
Using mixed methods, this case study of market dynamics in one settlement of Lusaka, Zambia, provides insights related to the empirical questions of interest that have emerged from five data collection phases over 1.5 years (February 2016-July 2017). To provide a brief overview of these phases, we mention here that phase one was characterized by review of various kinds of literature and documents (to understand context), phase two was characterized by a survey to all LFPSs with ECCE sections (n=65) in the settlement (to get a picture of their characteristics), phases three and four entailed school visits (with 12 different schools) and interviews (n=40) with parents and school actors (teachers, head teachers, school directors, owners) in order to understand the interaction and strategies of these actors, while phase five completed the picture we sought to construct by surveying parents (n=163) to understand the demand side of the market. Because the majority of the insights shared in this article come from data collection phases one, two, and five, we describe these first in what follows, before returning to phases three and four.
The first phase, which has been ongoing, has focused on answering the question of why LFPSs emerged in Lusaka. To answer this question, we gathered and analyzed academic, governmental, organizational, and news-media resources related to the political-economic context of Zambia generally and Lusaka specifically. Our focus has been to understand the key economic and political developments in recent years that have, for example, affected: economic opportunities across the country (which have encouraged internal migration), budgets for public services (resulting, e.g., from economic crises and structural adjustments to the economy), and the provision of education at the ECCE level by the education ministry (a recent phenomenon in Zambia). 3 The second phase of data collection, in February 2016, focused on the second question of interest, which relates to the nature of the LFPS market at the ECCE level in the study site of Mtendere, located in the capital city of Lusaka. Before saying more about this round of data collection, we should note that this slum was chosen for multiple reasons: because it is one of the largest (with a population of ~109,000 as of 2016) and oldest (thus enhancing the likelihood that market dynamics have had a chance to develop), and because it was considered a relatively safe choice (in terms of safety for the researchers). (Note: The context of Mtendere is further discussed in a later section.) With that in mind, during this data collection phase, the research team sought out LFPSs that offer ECCE by canvassing the slum on foot, asking inhabitants where such schools were located. This was necessary because no central listing of ECCE centers is available. In all ECCEs encountered (n=65), principals or head teachers were surveyed to get a detailed portrait of school characteristics.
The fifth and final data collection phase also related to the question of the nature of the LFPS market for ECCE. (We will describe phases three and four immediately below). While the survey given in February 2016 focused on characterizing the schools (i.e., the supply side of the market), a separate survey was given in July 2017 that focused specifically on parents. We did this in order to better understand the demand side as well. Again, the research team combed the slum, but this time looked for parents with at least one child enrolled in ECCE. The resultant sample (n=163) is not claimed to be representative; it should instead be thought of as indicative of the characteristics of parents participating in this market. Questions in the survey focused on household composition, income and resources, education levels, and the reasons for selecting ECCE centers. The surveys from rounds two and five allow us to portray both the schools and parents in general terms and on the basis of basic quantitative indicators and descriptive statistics.
The majority of the findings reported in this paper come from the data sources described above. Where possible, we incorporate insights from phases three (in November 2016) and four (in January 2017), which were qualitative, in contrast. For the first of these phases, we interviewed three groups of actors. The first group included governmental officials (e.g., who were working on the policy for early childhood education, who sat on the Lusaka City Council) as well as representatives of international agencies (e.g., UNICEF) in order to deepen our understanding of the context, including relevant governmental policies. The second and third groups interviewed were school principals, teachers and parents. During school visits for interviews, we also observed classes in session and made notes on the physical setting and the learning materials available. These interviews focused on logics of engagement. Phase four again sought out school directors, teachers, and parents.
In both phases three and four, our goal was to include schools that would allow us to gain insight into the operation of the market. This meant seeking out schools: (a) that were located in the most densely population section of the slum (where schools are more likely to respond to market pressures), (b) that represented the different kinds of options available to parents (meaning government schools, LFPSs, and community schools, the latter of which are described later in this section), and (c) that reflected a range of price points within the general category of "low-fee." Once we identified these schools, we performed site visits and conducted interviews that were very valuable in that they permitted us to further investigate the ways that key actors at the local level perceive and characterize LFPSs and the market dynamics (or lack thereof) that guide the engagement of schools, teachers, and parents. 4 In all, interviews were conducted with 23 parents, 11 teachers, and 6 school directors or owners from 12 different schools, the locations for which are depicted in Figure 1. As can be seen in Figure 1, our study included one government ECCE center (i.e., the only government center in the entire slum to offer ECCE), two community schools, and nine LFPSs (all listed with pseudonyms). For analysis, we repeatedly reviewed the transcripts of the interviews to look for details and themes with regard to the strategies, forms of engagement, and dynamics individually and collectively represented by the schools and parents in our study. Given space constraints, we cannot fully report or exploit the insights from phase three and four interviews. While a future paper will engage with these interviews at-length, here we draw on insights gained to inform our commentary at various points. Definitionally, our study follows the work of Srivastava (2013b) and Harber (2014) when it comes to the criteria that distinguish the low-fee private schools of interest. First, a school is defined as private if it is established, administered, and controlled by a private entity; where the state plays no or little part in the management and governance of the school; and where the major source of school income is from user fees. Second, for our purposes, private schools are "low-fee" if they serve economically and socially disadvantaged groups while also charging fee that are substantially lower than those paid at elite private schools. 5 As with other research on LFPSs, this definition is necessarily context-dependent and relational, since no absolute or universally applicable definition exists for identifying private schools as "low-fee." In our study, we do not include "community schools" in LFPSs; rather, we distinguish the two (community schools and LFPSs). The Education Act of Zambia (2011) offers four different categories of schools: government, community, private, and grant-aided (with the latter being traditional church-run schools from the colonial times). In Zambia, community schools are established and governed by the parents and local community and are required to be registered with the Ministry of General Education through each District Education Board (unlike private schools). In addition, once meeting the several criteria stipulated in the guidelines, community schools are entitled to receive the same grant from the government as the government schools, while private schools are not. 6 Throughout, we mention characteristics of community schools as necessary for the purpose of distinguishing them from LFPSs; for a detailed treatment of community schools, see Onsomu, Mungai, Oulai, Sankale and Mujidi (2004).

Political-Economic Dimensions, LFPS Emergence, and ECCE Policy
Zambia is a young country, having achieved independence from Britain in 1964. Despite its independence, however, Zambia has continued to be subject to foreign influence economically and politically, just as it always had. Following a period of "company rule" in the late 1800s and early 1900s by the British South African Company, the area now known as Zambia was run by a colonial government that did not interfere with the concessions granted to monopoly capital (related, e.g., to mineral rights). Although the newly independent government nationalized key industries in the late 1960s, profits were not used very well for national development as "national aspiration was sacrificed on the altar of personal enrichment and careerism" (Mudenda, 1984, p. 139). Since the 1970s, economic crises have prompted the involvement of the International Monetary Fund (IMF) and the World Bank, which have pushed for the enactment of structural adjustments to the economy that have been associated with negative consequences for Zambians. As Hansen (2008) writes, since the early 1990s, "The two decades of IMF/World Bank initiated [structural adjustment programs] were marked by growing inequality, unemployment, deteriorating health including a high 5 To give an indication of the discrepancy between what school charge in this study and what elite private schools charge, consider that the most expensive school in our sample charged K1500 ($155) per term, while Kaoma (2016) reports that private schools for high class families charge K7400 ($766) per term or more (p. 65). Separately, our own observations indicate that elite private schools can be divided into two categories. The first category is comprised of schools that are "local" high class private schools (meaning that they are not targeted to international families living in Zambia); they charge between K4500 ($450) and K4800 ($480) per term. The second category contains international private schools, and they charge between K4800 ($480) and K8000 ($800) per term. 6 Although it was reported that many community schools do not receive such a grant regularly nor in full (Okitsu & Edwards, 2017).
HIV/AIDS prevalence rate, declining access to education, and an urban infrastructure that did not keep pace with population growth" (Hansen, 2008, p. 216).
Liberalization of the economy since the mid-1990s has gone hand-in-hand with increased informalization and precariousness for the average citizen. Decreases in jobs with the state together with competition for local companies from imports meant that formal employment declined-from 17% of the labor force in 1992 to 11% in 1999 (Hansen, 2008, p. 217). Despite the intentions of the Zambian government to diversify the economy, mining is still, by far, the dominant sector of the economy, in terms of foreign investment and exports (Sikamo, Mwanza, & Mweemba, 2016). Yet, on the whole mining employs a minority of the population (particularly in urban areas, where it employs 4% of those 12 years or older), with the majority engaged in the informal economy (80% of the population generally and 60% of those in urban areas; Central Statistical Office, 2016). 7 And as Hansen (2008) points out, "Neither the one-party state nor the state in the neo-liberal era has been keen on the informal economy. The general approach to urban growth and along with it, the informal economy, has been neglect if not outright hostility" (p. 221).
The struggles in Lusaka have reflected the trends in the country more generally. At the same time that Lusaka saw exponential growth in its population (jumping from approximately 1 million inhabitants in 1990 to just over 2 million in 2010), the city "experienced a drastic decline in employment opportunities, forcing many urban residents to adopt a variety of coping strategies to ensure household food security and general well-being" (Simatele & Binns, 2008, p. 11). Though the employment rate grew at 11% per year during 2008-2014 (Ravillard, 2017), government figures from 2015 show that only 38.5% of those older than 12 are employed in the formal sector (Central Statistical Office, 2016). Moreover, "because housing markets have been privatized and no low-cost government housing has been constructed since the 1970s, the vast majority of Lusaka's population lives in informal housing in the peri-urban areas" (Hansen, 2008, p. 216). To that end, 12% percent of the population lives in high-cost residential areas, 20% in medium-low cost areas, with the remaining 68% of the city's population inhabiting poor unplanned and squatter settlements (Simatele & Binns, 2008, p. 2). Perhaps unsurprisingly, given the government's history of neglect when it comes to public services in the slums, these areas are where LFPSs have emerged in large numbers in recent years. It is to be expected that private entities would step in to provide schooling in spaces where government schools are either unavailable or over-crowded, a fact that studies from other country contexts have pointed out (Srivastava, 2013a).
Beyond the general political-economic dynamics highlighted above-related to the lack of funding for the public sector and migration to cities-recent research on the education sector points to further reasons for the emergence of LFPSs. A report by the Open Society Foundation from 2013 highlighted some structural issues that afflict the formal education system, 8 including insufficient funding, low absorptive capacity to effectively and efficiently utilize resources, lack of monitoring of the implementation of policies/plans, the poor functioning of designed governance structures, and the ever-present effect of the HIV/AIDS epidemic" (Beyani, 2013, p. 2). One indicator of the shortage of resources available to the Ministry of General Education (MoGE) is that international donors have, at times, contributed over 30% of the national education budget (Beyani, 2012, p. 16). Research from the school level has confirmed what one would expect from the situation described above-namely, that teachers lack basic materials (such as teacher handbooks), schools have an insufficient number of classrooms, and students are in overcrowded classes (Thomas & Thomas, 2009).
Although the above combination of factors has undoubtedly stimulated the growth of private schools at the primary and secondary levels, the factors encouraging their growth are even stronger at the ECCE level. MoGE formally made ECCE part of its mandate in 2004 (UNESCO, 2006), but not much happened in the following decade. 9 In June 2015, the government reported that over the previous 10 years, MoGE had "developed a curriculum, hired 1,000 trained teachers, and allocated 0.05 percent of the 2015 budget to [ECCE]" (MESVTEE, 2015, p. 12). 10 There is no evidence of significant financing to ECCE in previous years, and thus it was only in 2015 that the government began to speak of ECCE centers that would be built in future years, with plans being declared to build 60 new schools (MESVTEE, 2015). What few publications do comment on LFPSs at the ECCE level simply highlight that these minimal efforts by the government have allowed "home-based pre-schools" to "mushroom" (Thomas & Thomas, 2009, p. 7). Similar comments are made by Hamusunga (2012) and Beyani (2013), who point out that there has been a proliferation of private institutions serving pre-primary students and that the government, by default, has left the provision of services in this sub-sector to private schools, NGOs, and churches. Notably, these authors also underscore that the learning environments are not conducive to learning, that the curriculum is inappropriate (due to its focus on academic preparation and not holistic learning and play), and that the schools hire many unqualified preschool teachers (Beyani, 2013;Hamusunga, 2012;Thomas & Thomas, 2009).
While the government acknowledges that ECCE covers children from 0-6 years of age, there is a division of responsibility. MoGE policy only pertains to students 3-6 years old while the Ministry of Community Development, Mother and Child Health (MCDMCH) is tasked with providing for 0-3 year olds (MESVTEE, 2015, p. 12). The vision of the national policy framework on ECCE is that ECCE will allow all Zambian children "to achieve their full developmental potential" through the "provision of high quality ECCE services" including "infant and child stimulation in all areas of development" (MESVTEE,n.d.,p. 16). Yet, as of 2013, only 14.8% of grade one students had attended some form of organized ECCE (MESVTEE, 2015, p. 12). There is thus a large gap between the aspirations communicated in policy documents and the realities of the ECCE sectorrealities which remain poorly documented and understood.

The Context of the Private Preschool Market in Mtendere
As mentioned, we seek to address the lack of knowledge on the ECCE sector by characterizing the context and nature of the LFPS market in one large slum-Mtendere. Our goal in 9 While the shift from a sector run predominantly by private providers, churches and non-governmental organizations (with oversight from Ministry of Local Government) to significant involvement of the MoGE occurred in 2004, the results are still modest (MESVTEE, 2015, p. 12). Historically, ECCE has not been a major responsibility of government. This has been the case for both preand post-independence governments. In colonial times, preschool education was offered for one year and included learning to write letters of the alphabet, on the ground, for African children. This was the closest the system then came to offering ECCE. Later, the colonial government came up with Day Nursery Act of 1957 to benefit local children. After independence, the government established nurseries and pre-schools through the Ministry of Local Government and Housing. These were mainly located in welfare halls. The level of participation, though, remained low, and by the middle 1980s this provision had fizzled out (MESVTEE, 2015, p. 12) 10 This percentage of the budget is equivalent to about USD$300,000 (MESVTEE, 2015). this and the next sections is to describe key features of the location as well as the schools in the market and the parents that participate in said market, just as a handful of other studies have done in other locations (Fennell & Malik, 2012;Harma, 2016;Moschetti, 2015;Ohba, 2013;Oketch, Mutisya, Ngware, & Ezeh, 2010;Singh & Bangay, 2014;Srivastava & Noronha, 2016). Not only will these two sections help to answer basic questions about market participants at the ECCE level, but they will also help to interpret the market dynamics that are reported. Beyond characterizing the supply (schools) and demand (households) sides of the market, our commentary here is likewise concerned with observed connections between privatization and marginalization. Explicitly, for marginalization, we are concerned with the ways that participation in LFPSs at the ECCE level may undermine the stability and survival, let alone, flourishing, of students and their families. 11 Mtendere is one of Lusaka's oldest slums (one which would later become a somewhat planned "settlement"), dating to the 1960s. Residents began locating here from another slum (Kalingalinga) upon being forced to leave because that area was being demolished for the expansion of the University of Zambia. As with other settlements, houses were initially (and still are, in some areas) makeshift in nature, being "made of either poles and grass or mud bricks with grass thatched roofs" (Nkole, 2018). However, by 1971, the Lusaka City Council (LCC) had graded the unpaved roads, demarcated the area into sections, and subdivided land into plots, which were then given out for the purpose of housing construction by individuals. Over time, though the LCC was not able to realize its hopes of turning Mtendere into high end housing, many residents have built houses made of concrete blocks and metal roofs.
Many challenges remain, despite the settlement's longevity. For example, as of 2018, it was reported by Nkole (2018) that there is no police station (leading to high crime rates) and that there is erratic water supply (with residents unsure that the situation will be fixed). One of the newer sections of the settlement, known as Mtendere East has 50,000 inhabitants and no school. In this area, there are reports of students as old as 10 or 11 being enrolled in grade one, likely a result of the fact that parents do not want their children to travel long distances alone. In the map in Figure 2, one can see Mtendere ward outlined in black. Mtendere East is in the central part of the outlined area, while our data collection focused on the older and more built-up part of Mtendere, on the lefthand side of the outlined area. In terms of employment, most of the settlement's population works in the informal sector (more on this below), a result of the fact that the area is home to few formal businesses and is located far from the Central Business District. According to the local health clinic in Mtendere, there was a population 109,500 as of 2016, 20% of which (or approximately 21,000 children) was between 0 and 59 months of age. 12 We identified 65 ECCE centers operating at the time of data collection that served Mtendere settlement. Out of these 65 ECCE centers, 61 are private, 3 community and only 1 public, which indicates that private ECCE centers account for over 90% (93.8%) of all the ECCE centers offered in the slum (Table 1). Furthermore, it is revealed that the private ECCE centers have drastically increased since 2005. As Table 2 shows, more than 86% of the private ECCE centers have been established since 2005, while 60% of them have been established since 2010. This trend indicates that the rise of ECCE centers has coincided with population growth in Lusaka and with the establishment of ECCE as a priority of the government, albeit one to which the government has dedicated insufficient resources. 13 The school survey and the interviews with the owners of the schools indicate that the private ECCE centers have been primarily established by individuals (Table 3). Interestingly, the school survey reveals that more than half of those individuals (33 out of 61 centers) were former and/or current government school teachers (Table 3). Although it is unclear why such a concentration of these individuals have decided to start ECCE centers, what is certain is that the government is not striving to avoid the marginalization of the government's ECCE center in the eyes of parents, given that it provides no funding to the school apart from teacher salaries (which explains the school fees discussed in the next section). Put differently, parents are likely to view the public ECCE unfavorably due to the government's lack of investment in it. Following teachers, the most common previous occupations of the founders of the private ECCE centers were businessmen/women (7 out of 61) and religious leaders (4 out of 61). The interviews with the founders indicate these individuals opened their schools by responding to the market signals they saw in the compounds-poor parents' high demand for early education for their children prior to entering primary school in the context of the absence of government provision in this area. The total number of LFPSs is given to change, as new centers open or close. That said, the majority of ECCE centers (69%) in our study are more stable because they are attached to primary schools, with these primary schools existing first, before the ECCE section was established, with the implication that ECCE likely benefits from the schools reputation and existing family networks. To that end, in our sample of 11 schools from the qualitative portion of this study, the only one to close was the one that was a stand-alone ECCE that provided services in the living room of the house where the proprietor lived. Generally, though, our observations indicate that in the majority of ECCE centers the environments are not conducive to learning: some schools are over-crowded, there is poor sanitation, hygiene is a problem, and there is a shortage of teaching-learning resources (inadequate textbooks for pupils, lack of teaching aids and toys for children at ECCE level, absence of a playground).

Regulation
When it comes to regulation, LFPSs are required to be registered with the LCC in order to operate. It is the "health" and the "commercial" sections of the LCC that are responsible for the registration of these schools. Each institution needs to pay the registration fee of K88.58 (approx. USD10) to the LCC, in addition to property taxes. The LCC only looks at health/sanitation issues of each school when approving the opening of such private schools. Education quality is not considered at all, as these schools are registered with the LCC as "commercial enterprises." While many studies of low cost private schools in sub-Saharan Africa report that LFPSs are unregistered, our study in Mtendere found that most (all but 25% of the LFPSs) of the schools are registered. 14 In fact, the schools that we visited were very sensitive about disclosing their sanitation/health conditions to us, as they were afraid of their school being closed own by the LCC. The officers working for LCC Mtendere office reported that at one time all the pre-schools were suspended by the LCC, due to the outbreak of cholera. On the other hand, these same officers said that apart from that one occasion, they were not aware of any case of disclosure/suspension of these schools by the LCC.
MoGE plays absolutely no role in registering, inspecting, training or equipping LFPSs. The director of Early Childhood Education within MoGE, whom we interviewed, did not seem to be willing to take any active role in the regulation of such low cost private schools from an educational stand point, saying that "these are just profit[-seeking] organizations." However, in light of the fee levels charged by LFPSs (discussed immediately below), it is unclear that all or even most of them are able to make a profit. Table 4 provides information on the fees charged by LFPSs, community schools, and the government school on a per term basis, with each term lasting four months. There are three terms per year (term 1: January-April; term 2: May-August; term 3: September-December). As can be seen, LFPSs, on average, charge 2.5 times as much as the government school, which has fees associated with basic expenditures since government funding only covers teacher salaries. In LFPS preschools, there are additional fees related to indirect costs such as lunch, uniform, registration, and school trips. Despite the higher fees charged by private ECCEs, all parents interviewed reported that they have enrolled at least one of their children in a LFPS in the past, with the implication that LFPSs contribute to increased access for the poor in general, although certainly at a burdensome cost. As will be discussed in the next section, on the demand side, the fees and indirect costs often prevent the poorest and most vulnerable from sending their children to the preschools they want or from sending children to any preschool at all. As with other studies of LFPSs and school choice (Edwards, Klees, & Wildish, 2017;Verger, Fontdevila, & Zancajo, 2016), our survey results suggest that LFPSs in Mtendere manage their costs by excluding students with disabilities and by keeping teachers' salaries down. In terms of disabled students, only 16% of LFPSs reported enrolling any students who would fall in this category. One school director confirmed the tendency to avoid special needs students: "For kids with special needs, we refer them to schools who provide special needs education" (School K, DIRECT2). Likewise, where possible, parents of children with disabilities likely avoid low quality LFPSs (and community schools) because they are not equipped to handle such students. LFPSs, rather, focus on students without special needs while also allowing for higher student-teacher ratios.

Fees
Our interviews indicate that LFPSs permit student-teacher ratios up to 1:45 while the government school reports a ratio of 1:21, though, to be fair, on average, the LFPSs have student-teacher ratios that are identical to that of the public school. 15

Teachers
Turning now to the teachers, most in LFPSs have either a preschool teaching certificate or a diploma (86%, compared with 100% of public school teachers and 20% of those in community schools). 16 The key differences between teachers in public and LFPS schools relate to their age, pay, and certifications. While teachers across all schools surveyed tend to be female (greater than 95%), those in LFPSs tend to be younger (29 years old on average, compared with an average of 39 years old in the public school), and tend to have obtained their certification from private colleges, which have grown exponentially in recent years, and which are frequently unaccredited. In contrast, teachers in the government school have certifications recognized by the Zambia Preschools Association. 17 Crucially, LFPS teachers at the ECCE level earn one third as much as government teachers, on average. Whereas public school teachers reported earning between K2700-4000 ($280-415) monthly, their LFPS counterparts in our survey earn a mean of K807 ($93) per month and a high of K1700 ($176). Clearly, teachers in LFPSs, due to their low pay and insecure employment, are made to suffer in these arrangements, where they earn significantly less than the average income for Lusaka (K1800, or $186, monthly). Yet many are willing to endure low pay and longer hours (typically 7am-4pm, as opposed to the 4-5 hours worked daily by government teachers) in hopes of one day gaining employment with a public school, either at the ECCE or primary level. Notably, there are many teachers in this precarious situation, thanks to the emergence of numerous low-cost, private, and unaccredited teacher training colleges that have attracted and graduated an over-supply of teachers who cannot find employment, due in part to the fact that the government itself was not even hiring ECCE teachers until recently, and has, according to available documentation, only hired 1000 such teachers. As the next section makes clear, many parents are in a similarly precarious situation.
Lest we give the wrong impression, we should clarify that the LFPSs in Mtendere are not immune from precarity. That is, the strategies enumerated above related to teacher salary and disabled students, for example, do not ensure that schools are stable or profitable. One issue is that parents have many options (discussed further below), but the larger issue is that there is a limited market, in the sense that only a subset of households from Mtendere can afford to send their children to an LFPS, unless they are of extremely low quality in terms of infrastructure and available materials. Informally, in terms of learning resources and the schooling environment, our observations indicated that only those schools that charge at least 3-4 times as much as the 15 Based on the 13 LFPSs for which we have student-teacher ratios. 16 Although the MoGE now requires all teachers to have a diploma, at the time of data collection either a certificate (two years of training) or a diploma (three years of training) was acceptable. 17 The Zambia Preschool Association (ZPA) was formed as a result of the Day Nurseries Act of 1957, in the colonial period. From its inception, the association has had close relationship with the Ministry of Local Government and with the Lusaka City Council, who were mandated to run preschools. In the 1970s and 1980s, all pre-schools were registered with the district or city councils. ZPA played the role of creating an affiliation among preschools in order to give technical advice on how to run such institutions. ZPA also trained pre-school teachers and organized workshops for them. Currently, however, ECCE training is done by the government itself in the government-run teacher training colleges and universities. With this recent change, the role of the ZPA is not clear. government showed some seriousness in offering a good education. 18 The other schools may rely on marketing (e.g., in the form of a new paint job to attract parents) but are in dire need of teaching materials, or, in the worst cases, are offered in informal settings (e.g., a private residence) and incomplete buildings without lighting. In such struggling LFPSs, teachers are paid significantly below the average mentioned above (as little as K300 [$31] per month) and do not even have certificates from unaccredited private colleges. In these situations, the schools and teachers are undoubtedly faced with decisions about how to survive, in the face of parents who cannot pay regularly or in full, a point to which we return later. Immediately below, we shift to a characterization of the demand.

The Nature of the Private Preschool Market in Mtendere: The Demand Side
Although there is much that could be said based on the parental surveys, we focus here on some aspects that will help us to understand the ability of households in Mtendere to participate in the LFPSs. We direct our attention to characterizing the families in our sample, situating them in the context of Lusaka (to ascertain their relative marginalization), assessing the extent to which LFPSs are within reach vis-à-vis their financial and social characteristics, and reflecting on the implications of these insights for those families who cannot afford to enroll their children in an LFPS.

Household Characterization
A basic characterization of the sample would note that a minority of houses have one adult (4.9%), while most (~75%) have either two (54.6%) or three (18.4%). 19 Most children live in a household headed by their father (66.3%) or by a grandparent (20.3%). About half of the households have two earners (48.5%), with the second earner being predominantly female (76.7%). With their earnings, the great majority of families are able to afford at least one TV, radio, and mobile phone. It is less common, though, for the head of household to have completed a very high level of education: 9.3% completed grade 7 or less, 25.9% completed graded 9 or less, and 74.1% percent completed grade 12 or less. Perhaps surprisingly, 12.3% and 4.3% report completing college and university, respectively. As for children, 30% of households have at least one child in the 0-2 age range, while 88% have at least one child between of the ages of 3-6. Interestingly, most of the households (75%) also have at least one child aged 7-18, which is important to note because of what it implies for household schooling expenditures. Many of these statistics are summarized in Table 5. 18 While there is debate in the research literature on the connection between education quality and school resources, infrastructure, and teacher credentials, our comments here are based on the extremely dire situations that characterize those LFPSs that charge less than 3-4 times what the government school charges for fees. 19 Small percentages have 4 (9.8%) or 5 (7.4%) adults.
Notes: *Defined as members over 18 years of age. **Defined as members under 18 years of age who have no income. ***Defined as number of members earning income of any kind.
It might bode well as an indicator of stability that 78.5% of students in the sample live in a household with married parents. At the same time, however, our survey indicates that the families in our study are very susceptible to life events that can jeopardize their ability to earn and therefore to send their children to school. To that end, 47.2% report experiencing at least one injury, illness or other shock during the year of the survey, where shocks included loss of job, death, divorce, major theft, fire, and arrest.
One key way to characterize the marginalization of a household is to look at the kind of employment that earners are able to obtain. In Table 6, we see that 40% of heads of households have formal employment as either high or low-status wage earners. High-status wage employment includes such positions as government official, teacher, NGO officer, car driver, executive, health workers, car driver, etc. Low-status wage employment includes such positions as guard, cleaner, and restaurant worker. These statistics are consistent both with Lusaka more generally, where 42.6% of those 12 years of age or older have formal, paid employment (Central Statistical Office, 2016) and with the statistic mentioned earlier that 60% of those in urban areas are in the informal economy. Household earners that fall into the other employment classifications listed in Table 6 would be labeled as unemployed or economically inactive by the government. We get a more nuanced sense of what non-wage earners in our sample likely do for a living by noting that, according to government statistics, Zambians who are generating income but are labeled as unemployed or economically inactive typically engage in the following kinds of activity: 24% of them are engaged in petty vending at home (selling vegetables, bread, etc.), 22% in other petty vending/hawking, 5% in baking, 4% in brewing beer, 7% in hairdressing, and 20% in piece work (gardening, digging pits, etc.; Central Statistical Office, 2016, p. 66). More specifically, we seek to put sample households into financial perspective, to understand the place of education within household expenditures and the tradeoffs that are necessary, and at what literal and metaphorical costs. Surprisingly, in terms of average income. our sample seems to be a reflection of Lusaka more generally, as opposed to being worse off, as shown in Table 7. The fact that the average income in our sample is on par with Lusaka overall is surprising because a study by USAID from 2017 found the average monthly income to be much lower than that for our study ($193 vs. $302, respectively). 20 The average income for our study could be drastically different because it was a requirement for those in the sample to have children in ECCE (hence excluding the families who are extremely poor and have no access to preschool) and because our sample is taken from the central part of Mtendere (not the fringes, which are newer and more poor).
Stepping back, one sees that most income figures in Zambia reflect some level of poverty. Consider that 70% of Zambians live below the national poverty line while 57.5% live on less than $1.90/day (World Bank, 2019). 21 More tangibly, in the context of our study, we can look at the tradeoff between food and schooling. Based on figures from 2017, it is estimated that, for a family of five in Lusaka, the monthly food expenditure for the Basic Needs Basket is K1420 ($157.77) (Posiana, 2017). If the families in Mtendere spent this amount on food, the expense would consume over 77% of their income; and for the average household in our sample, it would represent half of their monthly earnings. The fact that Mtendere households spend significantly less on food than the amount mentioned above means that families are malnourished due to simple diets. 22 Moreover,20 Note that, throughout this section, dollar values should be interpreted as approximations, since the figures compared have been converted from Kwacha and come from years in which there has been significant fluctuation in the exchange rate. 21 According to Chibuye (2014), there are two poverty lines in Zambia: "'extreme' poverty is the inability of a household to meet basic nutritional requirements even if their consumption basket is defined to include food alone; 'overall' poverty, on the other hand, is the inability of an individual/ household to meet non-food needs, while meeting basic minimum food requirements (p. 236). 22 This is not surprising given that 15% of children in Zambia are underweight and 40% are stunted (USAID, 2018). according to the definition given by Chibuye (2014), 25% of the families in our sample would be considered extremely poor because they cannot afford the Basic Needs Basket (let alone housing). As Table 8 shows, the highest salary for the first quartile of monthly incomes is K1400 ($155). Even at the level of the median monthly income, the Basic Needs Basket still represents well over half a month's earnings. However, the fact that these families are in our sample means that they are making sacrifices in order to be able to send their child to school at an LFPS. What adds to the precarity of many households is the fact that roughly a third of their income is spent on nourishment, making families vulnerable to food price volatility (USAID, 2017, p. 20), in addition to their vulnerability to the other shocks discussed above (job loss, illness, etc.).

LFPS Participation
Despite the constraints on their resources, families in Mtendere demonstrate the value they place on education by making it the third largest expenditure, after food and housing (see table 7; USAID, 2017). With regard to our sample, despite financial difficulties, families are able to send at least one-and many times, more than one-child to ECCE. As Table 9 shows, in 70% of cases families also have a child in grades 1-12, for whom there are undoubtedly additional school fees to be paid. Tables 9 and 10 summarize a number of specifics that allow us to better understand the characteristics of the children in our sample who participate in ECCE. Table 9 includes information on each of the ECCE children in the household (allowing for up to three different children). The most salient characteristics to highlight here are that (a) the overwhelming majority of children (approx. 85%) enroll in ECCE between the ages of 2-4, (b) private schools account for 93% of enrollment, and (c) there is a somewhat even distribution of children across the three types of ECCE classes offered, with these being labeled "baby class" (for children up to two years old), "middle class" (for children 2-4 years of age), and "reception class" (for those 5-6 years old). Note: Primary school is grades 1-7, lower secondary is grades 8-9, and upper secondary is grades 10-12. Note: Children typically start grade 1 at around 7 years of age.
While the above information helps to paint a picture of LFPS participation, the larger question is the connection between LFPS participation and household finances, because that can shed light on the issue of sacrifice and marginalization. To speak to this issue, we must first look to fee levels reported by parents. 23 Table 11 reports school fees per term (with each term lasting four months) as well as other indirect costs (lunch, parent-teacher association, exam fees, uniform cost). On a monthly basis, if we average the school fees and indirect costs reported, we arrive at a total amount of K169 ($18.77), equivalent to 6.2% of the average monthly income for our sample. 24 In our sample, the amount spent and the percentage schooling represents as a fraction of monthly income are on par with the rest of Lusaka (K164, 5.7%) and other urban areas (K180, 6.7%), as can be seen from Table 7. It is unknown why the figures reported by USAID for educational expenses in Mtendere are drastically different, since the report does not clarify which fees and costs are included in their monthly educational expense of K283 (or 12% of income).
In the context of earnings, what reported fee levels mean is that, on average, households (which tend to have 1-3 earners) are making ~$10/day (assuming they work 30 days/month), only to then spend ~$1/day on school. Importantly, though, fees are not typically collected on a daily basis. Schools collect their fees at different intervals, and even when schools collect for a month of enrollment, parents are often unable to pay in full. In Mtendere, as elsewhere, schools take what parents can offer, with the expectation that parents will pay the rest later, though parents may switch schools to avoid ever paying (Edwards, Klees, & Wildish, 2017;Srivastava, 2007). A more significant and consistent challenge is what to eat for lunch. Parents in our sample reported an average daily cost of K19 ($2.13) when lunch is purchased. If students were to purchase lunch daily (Monday-Friday), the average monthly cost of schooling would jump to K552 ($92), representing 30% of household earnings. When put into this perspective, it is no surprise that many Zambian children struggle with basic nourishment, as noted previously. It should be a surprise that parents are able to send their children to preschool at all, especially when one considers, first, that most households have more than one school-age child, and, second, that the monthly income figures reported are by no means firm since 60% of the sample is self-employed in the informal sector, hoping to sell their services, produce (e.g., vegetables), or hand-made products at local markets, along the side of the road, or as they go from house to house. A more fine-grained analysis underscores this point. Referring to Table 12, we see that the upper income quartile spends only 1.7% of their income on term fees, while the lowest quartile spends 6.2%-and these figures do not include indirect costs (for parent-teacher association dues, uniforms, exam fees, lunch, etc.). The implication here is that the poorest parents spend a larger portion of their earnings on schoolingwhen these earnings should arguably go to food, housing, healthcare, clothing, etc.-only to send their children to ECCEs that are more lacking in terms of resources, teacher qualifications, and the quality of the physical space. Surprisingly, 89.6% of parents report that they cover school costs from earnings. Almost no-one reports getting a loan (only one household) or having their children cover the cost (one household). And no-one gets a scholarship to cover school fees.

LFPS Selection
It may come as a surprise, but the affordability of ECCEs did not rank as either the first or second most frequently cited reason for ECCE selection. Table 14 shows that perceptions of good teaching and proximity weighed more heavily in decision-making. Interviews and the parental survey confirm that parents were aware of a number of ECCEs, and thus they placed a higher priority in their decisions on geographic location. Since the children enrolled are very young, safety and distance are of primary concern, a finding that has unsurprisingly emerged elsewhere as well (Balarin, 2015).
Interviews suggest that the desire for "good teaching" manifests in how quality education is viewed. Parents have a high demand for early English education and pre-mathematics that can be offered by private preschools in order for their children to gain academic "readiness" before entering grade 1. Put differently: parents have a strong desire to equip their children with "academic" and English skills at early age. For many parents in urban slums of Zambia, English competence is a decisive skill in the formal sector, one that gives them a better chance at consistent income. In addition, many parents interviewed reported that they believe that equipping their children with English and mathematics at early stage would enhance their academic performance once they start primary school, so they can proceed to secondary and then college level by successfully passing the examination. 25 As Phillipson (2008) points out in their study, parents intuitively believe that an early start to learning English will enhance their children's future. They seem to believe that earlier is better-and many even send their children from the age of 1.5 years old to such heavily academic-oriented private preschools. 26 In contrast, the current Zambian ECCE curriculum emphasizes a whole-child approach that is play-based and offered in the mother tongue.
As a final note on school selection, it needs to be highlighted that parents perceive the education offered at the government school to be low quality, which leads to their strong desire to send their children to private preschool and private primary school up to grade 7. However, the interviews with parents revealed that making a choice based on these beliefs is not unproblematic. Based on interview data, hardly anyone actually compared the quality of the government preschools and low cost private preschools before making a choice for the latter. That is, few parents have visited the government preschool or have first-hand knowledge of its offerings. Sometimes parents chose low cost private preschool because they distrust the quality of government preschools, based on their experiences in the past in government primary schools, which they stated were overcrowded. As noted earlier, however, the LFPSs in this study actually allow higher teacher-pupil ratios than the government preschool.

Interaction and Implications of Supply and Demand
While previous studies on Zambia made claims about the rise of the private sector at the ECCE level, our research has documented its presence in one area of Lusaka. In so doing, we answered the question of who provides private ECCE, who participates, and at what cost. As was shown, over 90% of ECCE centers in the settlement of Mtendere are private, with 86% being established since 2005, a time during which Lusaka's population has grown drastically and the government has highlighted the importance of ECCE, though without dedicating significant resources to it. Private ECCE centers have been established most commonly by former publicschool teachers, as well as businessmen/women and religious leaders. While private ECCEs charge term fees that are, on average, 2.5 times as much as the government school, these LFPSs keep costs down by avoiding students with disabilities, offering very low teacher salaries, and generally allowing higher student-teacher ratios-in addition to providing very few educational resources in the school itself.
In the context of Mtendere, where only 40% of the population has wage-based employment, it is perhaps surprising that dozens of private ECCE centers would be found. Put differently, when 60% of the population is self-employed in petty vending and piece work, and when the average monthly income is $300, one would not expect that families are able to afford private schools for their children. Yet the fact that many extremely poor families are in our sample means that they are making a tradeoff between nourishment and education. As noted, for half of our sample, the monthly cost of food (as measured by the Basic Needs Basket) for a family of 5 would represent half or more of their monthly income. In order to send their children to school, then, the quality of the family diet suffers. The severity of the choices facing many families in Mtendere is highlighted by the fact that the basic costs associated with attending an ECCE center (tuition, uniform, lunch) represent 20% of the average monthly income (or over half of the monthly income of the poorest quartile). Comparatively speaking, in making the necessary sacrifices to pay tuition fees, households in the poorest quartile spend a percentage of their income (6.2%) that is four times greater than what is spent by households in the top income quartile (1.7%), in addition to other fees. Moreover, on a daily basis, 47% of families in our Mtendere sample have had to face difficult decisions about survival in the face of the crises that they frequently experience, with these crises being related to illness, injury or other shocks (e.g., loss of job, theft, fire, divorce, etc.).
Despite such challenges, a few trends drive families to send their children to private ECCE centers. One trend is increasing employment opportunities in informal areas for women (in retail shops, hair salons, and houses, working as a maid, etc). The tendency of women to participate in informal employment, thanks to the economic boom in urban Zambia-since about 2004, driven by the global copper price hike-has necessitated that their children be looked after by someone else while at work (Randolph, 2016). Relatedly, there have been increased short-term employment opportunities in the urban service sector (thus increased cash income for some households in Mtendere, however unsustainable it may be) that enables some of them to pay for LFPSs. The copper boom over the last 10 years or so increased the informal jobs in construction and domestic work (cleaning, washing clothes etc.) in Lusaka. Not surprisingly, parents in our study reported that it is cheaper and safer to send their kids to low-cost private preschools than hiring a nanny.
The combination of low fees together with daily challenges experienced by households means that many LFPSs are in a precarious position. As shown in Table 15, this precariousness can be characterized by noting, first, that a quarter of the schools in the sample enroll 25 students or fewer (with one having only two pupils, in comparison with the 74 enrolled by the government ECCE) and, second, that smaller schools raise less money through tuition fees (since the fees are set lower). The challenge for schools, additionally, is that their enrollment figures shift as parents withdraw their children once they have accumulated school debts they cannot pay. The financial limitations of parents are well-known, and almost all schools give parents leeway when it comes to making payments. According to the school survey, LFPSs allow parents a grace period that ranges from 10 days to three months. Despite this flexibility, one quarter of the schools surveyed reported an annual profit of $20 or less in 2014 (calculated as the difference between claimed income and expenditures), as Table 16 shows. While this figure underscores the slim margins that some schools must manage, at least eight schools in the sample (from 2014) were not able to cover all their costs in that they claimed to have zero profit or to be operating at a loss. The reliance of poor schools on poor families has been highlighted in studies from other country contexts and has been characterized as a form of mutual dependence wherein students are at the mercy of the school director if they cannot afford the fees (Edwards, Klees, & Wildish, 2017). Although table 16 shows that some schools (e.g., in the top quartile for profit) are not at risk of losing money, the majority of schools in our sample report incomes that are quite low. To put school profits into perspective, recall that the families in our sample reported an average monthly income of K2721 ($302). Thus, mutual dependence between low-income LFPSs and low-income families is the status quo, given that too few spaces are available in the government school, and given that, as mentioned earlier, parents interviewed expressed their preference for private over public schools, as they believed that government institutions offer poor quality service (at least up to Grade 7). 27 The perception prevails that unless you pay you will not be able to receive quality service. In other words, parents generally believe that they can receive good quality education in a low cost private preschool simply because they pay them, though we should recall that parents interviewed admitted that they had not actually gone to visit or observe the government's ECCE center. As Bajaj (2010) has observed in the context of Zambia, even though "neoliberalized schools," such as LFPSs, "operate amid larger structural constraints vis-à-vis macro-level economic and labor market policies," they still benefit from "historical meanings about a perceived linear progression from schooling to opportunity that mask the limited chances for employment, let alone social mobility, and instead place the burden for academic and professional failure on the individual" (p. 187).
When it comes to quality, we should also recall that parents looked to language of instruction, with English being preferred, and whether or schools have an academic orientation, as opposed to being play-based, as in the government school. Parents explicitly stated their perspective that the government school is "too playful," though it should be noted that research actually supports play-based and learner-centered pedagogies at the ECCE level that do not look like the traditional academic learning parents may expect (see, e.g., Lynch, 2015). Similarly, research has shown that learning in one's mother tongue is better for effective learning, though, as noted, parents prefer English-language instruction (UNESCO, 2008). Additionally, and crucially, appearances factor significantly into parental perceptions, with such things as the school's paint job and uniforms attracting parents. In the words of one parent, "I just love that school-the way it looks, their uniforms" (PARENT 6). Schools may thus be low-fee and low-quality, but they can attract parents nevertheless if they are attentive to optics.
As Harma (2010) has noted elsewhere, public education has become a "ghettoized option of last resort for the poorest and most marginalized in society" (p. 38). To that end, as Table 12 showed, the families with their children enrolled in the government ECCE center were found to have lower incomes than those with their children in community schools or LFPSs. While this trend may be appropriate from an equity perspective, since lower fees are being paid by families with less money, these families' lives may reflect what Fennell and Malik (2012) observed in the context of Pakistan: "a sense of powerlessness experienced by parents and children in these poorest households-feeling condemned to the worst schooling type," or at least what is perceived as such (p. 258).
Returning to LFPSs, what is tragic is that the poorer families are making sacrifices in order to send their kids to schools that are questionable in terms of their quality. Many of the schools within reach of these families, at the lower end of the fee spectrum, operate in small rented premises converted from someone's home, and are typically not conducive to learning. The classrooms we observed were extremely overcrowded, children were squeezing each other for space with little room for them to move around, and the hygiene was of concern. The observation of classes in action also revealed the highly academic orientation of their operation, putting emphasis on English 27 The interviews with parents suggest that their demand for private education is not high at the secondary level. The parents interviewed revealed that they wanted their children to go to private school up to primary but they want their children to proceed to government secondary school afterwards. For they perceived the quality of government services at primary are low, while those at the secondary level are not necessarily low. This phenomenon has been observed elsewhere as well (Ohba, 2013). proficiency from an early age. 28 In all classes, including baby class, teachers stood in front, facing children, mimicking the instruction style of primary schools in Zambia. There was very little singing and play, and very few storybooks or toys were seen (save for a few in the highest fee school visited). Moreover, the schools operate from 08:30am-4pm, with the school schedule filled with academic subjects (English, pre-mathematics, environmental science etc). The reception class (targeting 5-6 year-olds) has the strongest focus on academic/English skills. We did not encounter any scene where children asked questions of teachers.
In contrast, the public preschool appears to offer more age appropriate education, in that it is focused more on learning through play, as the national ECCE curriculum stipulates. At the time of our visit, teachers tried more action-oriented approaches and group work in the classrooms, which were more spacious; they also fill in daily lesson plans and a daily assessment sheet for each child, which was not observed in low cost private preschools. Additionally, public school teachers have the opportunity to get their skills and knowledge updated by participating in the refresher course offered by the Zambia Preschool Association, where they also receive training on the government's play-focused curriculum. On the other hand, we encountered no evidence that the teachers in the LFPSs receive any in-service training opportunities. Indeed, these schools have never had any MoGE official visit them for supervision/monitoring.
As a final point on teachers, we note that they are, in many ways, caught in the middle of the tensions generated by supply and demand. There is a surplus of individuals with ECCE training, mostly from the mushrooming private colleges, many of which are unaccredited, and they are desperate for jobs. Our study found that most of the preschool teachers are female, with quite a few of them being either single or divorced with children. By and large, however, the only positions available are with LFPSs, since the government has hired only 1,000 ECCE teachers nationwide. In our sample, the LFPS teachers earned, on average, one third of the monthly salary received by their government ECCE counterparts. An interesting, and in some ways cruel, twist is that half of the owners of the LFPSs surveyed were previous public school teachers-with the implication being that former teachers are now profiting (or attempting to profit) off the backs of other teachers, who are often paid erratically, depending on the payment patterns of families. And since the average monthly earnings of LFPS teachers in Mtendere (K807/$93) is less than that earned by the families in the poorest income quartile in our sample (K1400/$155), the teachers themselves are living on poverty wages and would not be able to send their own children to such a school without also making extreme sacrifices in terms of other basic expenditures. Other studies have similarly shown that LFPSs attempt to remain financially viable by compromising "teachers' qualifications and labor conditions" (Verger, Fontdevila, & Zancajo, 2016, p. 92).
The preceding discussion of the interaction of supply and demand-with its implications for family sacrifice, school precarity, and low quality-raises concerns and important questions about LFPSs in the ECCE sector, particularly when these dynamics are placed into the larger context of globalization. The concluding section offers comments in these regards.

Low-Fee Private Schools, the State, and Globalization
It is safe to say that LFPS centers contribute to ECCE access of poor children, who would not have the opportunity to expose their children to ECCE without such schools, given the fact that there is only one government preschool in the entire settlement of Mtendere. However, as has been shown, for many families it is only possible to send their children to ECCE if they make significant sacrifices. As was noted, the food cost for a family of five consumes over half of monthly income for families in our sample, and all (or more) of the monthly income for families in lowest income quartile, assuming that families eat a basic but balanced diet. In this context, affording an additional $18.77 per month, on average, for direct and indirect school fees becomes a major challenge, a challenge which grows if children purchase lunch. Indeed, even for the average family in our sample, school-related expenditures jump to $92 (or 30% of monthly income) if the average lunch cost is included. There are clearly financial tradeoffs with significant consequences when it comes to paying for food, shelter, and education, especially since most families have more than one school-age child.
In this context, MoGE has plans to expand ECCE access by making all public primary schools attach preschool sections. The government's promise to expand the public provision of ECCE is a welcome move from the perspective of equity. However, the feasibility of such a plan is rather questionable, given that many primary schools in urban slum areas are already over-crowded due to the country's free primary school policy (though, as noted, in practice there are still fees, although much lower than the average LFPS). Public schools simply do not have any classrooms available for preschool classes. Presumably, ECCE construction-together with ECCE teacher salaries-is what the government has in mind when it comes to the public education budget that it began dedicating to the ECCE level starting in 2014 (World Bank, 2015).
While these actions are promising, the question is whether the government can offer ECCE with quality. The answer to this question is uncertain, however, since Zambia is in the middle of a public finance crisis (discussed immediately below). In the short-term, the government may do well to focus on quality assurance in regards to the LFPS preschools that are mushrooming around Lusaka, in addition to step-by-step expansion of public provision with quality. That said, research has shown the limited ability of governments to adequately oversee the private school sector and to provide these kinds of support (Baum, Cooper, & Lusk-Stover., 2018;Harma, 2019). At the moment, these low cost private preschools function without any monitoring or support from the MoGE. Given that these schools operate to satisfy the market needs of poor parents who are not aware of appropriate pedagogical practices for early age children and lack ability to assess the quality of education on offer, there is a strong argument for enhanced governmental oversight. That said, as Srivastava (2008) has pointed out, governmental officials may be purposefully allowing LFPSs to operate, despite quality concerns, because they know that the government does not have the financial resources necessary to provide ECCE. To that end, from the perspective of the "conditioned" state, we can see ECCE policy to this point as an attempt to enhance "the institutional functions of the education system" (Arnove et al., 1996, p. 141) in order draw attention away from the fundamental contradictions of capitalism and to redirect it towards the idea that the public school system can provide a pathway to a better quality of life. That is, from this perspective, the government's ECCE policy to this point can be seen as playing a legitimizing function, a conclusion which is strengthened by the fact that the government has tended to highlight the policy without dedicating significant resources to support it in practice, as described earlier.
In order to fully interpret the government's constrained position, we must return to the political sociology of education and development and the way that it looks at LFPSs in relation to the dynamics of economic globalization. Here, the point is that Zambia, and its education system, are subject to the effects of a global capitalist economy. As far as the education budget is concerned, it has historically been the largest governmental expenditure, and has since at least 2006 fluctuated between 15-20% of the government's budget (Mutwale, 2018;World Bank, 2015). Yet the fact that 85% of the budget is consumed by personnel leaves few resources for other needs, like ECCE, particularly when the system is already severely lacking in materials (as described earlier; Mutwale, 2018). While international aid was for many years a source of additional financing-with cooperating partners providing 18% of the education budget in 2009, for example-the government reduced the proportion of international aid to 3% by 2014, following 10 years of strong economic growth, thanks to high copper prices (World Bank, 2015). And even when both donor contributions and the education budget were higher, ECCE was marginalized as a priority due to the attention that that primary education received in the context of the Educational For All movement and the Millennium Development Goals (Bajaj, 2010), which focused on access to primary education. Although donor focus now includes ECCE under the Sustainable Development Goals (adopted globally in 2015), ECCE was for many years marginalized at both local and international levels, in part because of the preferences and consensus of the international community.
While Zambia has reduced its dependence on one set of global actors (i.e., international donors and multilateral banks), the country is still very much subject to the whims of economic globalization (Fraser & Larmer, 2010). Thanks to an overall downward trend in global copper prices since 2011, combined with low tax collection, government revenues have decreased at the same time that public sector expenditures-in the form of infrastructure, hospitals, and increased wages for civil servants-have increased under the presidential administrations of the Patriotic Front, since 2011 (The Economist, 2018). As a consequence, debt service payments now consume 25% of the government's revenues (The Zambian Observer, 2018). Making matters worse, two-thirds of Zambia's debt is denominated in foreign currency, meaning that if the Zambian Kwacha loses value (e.g., as a result of declining copper prices), the real cost of external debt rises. Of course, as debt payments rise, which they have continued to do in recent years, there is less funding available for social services. In view of the foregoing, the need to raise tax revenue seems apparent. Indeed, as has been pointed out, it will not be possible to meet any of the agreed-upon international education goals without addressing tax justice (Balsera, Klees, & Archer, 2018). Yet this suggestion seems unlikely when the leaders of the various political parties have been described as "local agents of foreign capital" (Latzera, 2017), or what Mudenda (1984) has labeled the "comprador bourgeoisie" in that they are the local representatives of global capital, or what has historically been colonial interests (Gould, 2008). Such circumstances led a Zambian newspaper to remark: "Zambia is the tip of the tail of the global dog. When the dog is happy we find ourselves merrily flicking from side to side; when the dog is miserable, we find ourselves coiled up in a dark and smelly place" (quoted in Fraser & Larmer, 2010, p. 1). The difference for parents in Zambia is that, even in good times, many of them are faced with daily tradeoffs between basic needs and education for their children.
implementation. Geographically, these areas of focus have led to research projects on education in many countries across Latin America, Southeast Asia, and Africa. His recent books include The Trajectory of Global Education Policy: Community-based Management in El Salvador and the Global Reform Agenda and Global Education Policy, Impact Evaluations, and Alternatives: The Political Economy of Knowledge Production (both with Palgrave MacMillan). He also has a forthcoming co-edited special issue of Educational Policy entitled "School choice policy and politics around the globe: Sociological contributions." Readers are free to copy, display, distribute, and adapt this article, as long as the work is attributed to the author(s) and Education Policy Analysis Archives, the changes are identified, and the same license applies to the derivative work. More details of this Creative Commons license are available at https://creativecommons. Please send errata notes to Audrey Amrein-Beardsley at audrey.beardsley@asu.edu Join EPAA's Facebook community at https://www.facebook.com/EPAAAAPE and Twitter feed @epaa_aape.