The financial burden of a fee free primary education on rural livelihoods – a case study from rural Iringa Region, Tanzania

ABSTRACT This article explores the financial dilemma from a household level perspective of paying for primary education in Tanzania. In 2001, the primary education fee was abolished in a serious attempt to achieve free universal primary education. By eliminating what was perceived as the main obstacle for reaching education for all, the enrollment rates were expected to increase. However, the schools themselves are unable to manage with the government’s capitation grant only and in order to cope with increasing enrollment rates, households are asked for contributions to cover the expenses of school supplies, food and administrative costs. Based on fieldwork in Iringa Region, this article shows that education today is the main expenditure item for the majority of rural families. Despite a striking difference in housing standard and welfare between the three villages covered by the study, the data indicate a consistent outcome: education, even ‘free’ primary education is expensive in the rural context and families spend a large proportion of their income in order to secure a future of their children. Unlike previous research, children are not found to be withdrawn from school in order to help households with farming or domestic work.


Introduction
For almost two decades, Tanzanian primary education has been free of charge. In 2001, fees were abolished (URT 2001) to enable children from poor families to enter the school system and thus increase access to schooling (Vavrus and Moshi 2009). Universal primary education (UPE) was by that time addressed as part of the global agenda and declared as one of the eight millennium development goals (MDGs) that were launched at the United Nations Headquarters in September 2000, through the United Nations Millennium Declaration (UN 2000). However, the debate on the right to basic education was far from new; already in 1948 education was declared as a human right through Article 26 in the Universal Declaration of Human Rights. Additionally, the declaration stated that elementary education was to be compulsory and free of charge (Sifuna 2007).
In Tanzania today, however, the idea of free primary school education is far from being achieved. Instead of the previous fee paid once a year, parents are now asked to contribute to various items and services like examination fees, desks, electricity and lunch on an irregular basis throughout the year. These costs come on top of the expenses tied to obligatory school uniforms and necessary school supplies. Unfortunately, without parental contributions the schools are unable to run as expected as only 42% of the government capitation grant of 10 USD/ pupil reaches the school (URT 2015). Meanwhile, contributions are perceived as mandatory since children are dismissed or physically punished when contributions are withheld or delayed. While the outcome of the policy appears to be at odds with its aims: free UPE and increased access to schooling, the specific patterns of exclusion vary spatially as well as socioeconomically.
To learn more about the ground-level consequences of the 2001 policy, this article aims to explore four research questions, namely: (I) what is the extent of the financial burden at the household level and how does it vary among villages and (II) how does it differ by social group, (III) how is it handled at the household level and (IV) why do rural households still invest in education. Three villages with different accessibility and living standards were included to examine possible intra-rural differences in households' educational priorities and financial commitments.
This article draws on both qualitative and quantitative data collected during three periods of fieldwork; the first one carried out in October-November 2013, the second in October 2014 and the third in June 2015.

Background
Tanzania remains to a large extent a rural country with only one quarter of its population living in urban areas (UNICEF 2014). In Iringa, an even larger proportion of the population is rural: 83% (Ministry of Planning, Economy and Empowerment 2007). Currently, Tanzania falls within the low human development category according to the Human Development Index for 2016 and is ranked as number 151 out of 188 countries (UNDP 2016). Similarly, according to the World Bank, almost half of the Tanzanian population lives below the poverty line of USD 1.90/day (World Bank 2015). While regional poverty rate statistics do not exist, based on regional employment and housing standard data, Iringa Region to a large extent resembles the national level, although a somewhat higher rate of the population indicated farming as their main occupation, 70.3% compared to 62.8% at the national level (NBS 2014).
The national net enrollment rate (NER) for primary education currently stands at 76.8% for both sexes, with a large discrepancy between the urban and rural enrollment rates, 90.6% versus 72.2%. The NER is an improvement from the last census in 2002 when the rate was at 69.1% (NBS 2014) and 1999-2000 when the NER was 46.7% (Riddell 2003).
Iringa Region is doing well in terms of the NER and is ranked as the third region in the country, just behind Kilimanjaro and Dar es Salaam (NBS 2014). Kilimanjaro region has had strong historical support for education ever since the colonial era (Vavrus 2005) and as Dar es Salaam is the administrative capital and the urban national center it is also expected to be at the top. Even though an urbanrural gap in enrollment rates is visible also in Iringa Region it is not as striking as the national one. Iringa Region has an urban NER of 95.3% while the rural NER is 89.4% (NBS 2014). As Iringa has one of the highest NERs within the country it is likely that the households who participated in the study have their primary aged children in school. Therefore, the region serves as a relevant setting for exploring the outcome of the 2001 education policy on fee free education.

Conceptual framework
This article focuses on livelihoods in the rural context and how livelihoods relate to children and education. Additionally, possible intra-rural differences are discussed and how these relate to primary education. The well-known definition of sustainable livelihoods from Chambers and Conway (1992) states that: [A] livelihood comprises the capabilities, assets (stores, resources, claims and access) and activities required for a means of living: a livelihood is sustainable which can cope with and recover from stress and shocks, maintain or enhance its capabilities and assets, and provide sustainable livelihood opportunities for the next generation; and which contributes net benefits to other livelihoods at the local and global levels and in the short and long term. (Chambers and Conway 1992, 5) Chambers and Conway (1992) divide assets into tangible and intangible ones. The first category includes resources and stores, for example livestock, land and food stocks, while the second one includes claims and access, for instance demands and opportunities to obtain or use a resource (Chambers and Conway 1992). This view is followed-up by de Haan and Zoomers (2005) who urge for livelihood to be seen as a holistic concept containing both material and non-material well-being including, for example, a person's social network as a potential asset. Not only do social relations influence access to livelihood opportunities but also institutions and organizations, all of which incorporate a dimension of power (de Haan and Zoomers 2005). Furthermore, opportunities and outcomes, Zoomers (1999) argues, are influenced by geographical settings, agroecological zones and distance to markets. Locationrelated factors may have a larger impact on farmers' opportunities and outcomes than their strategic actions. In addition to the location-related factors, the author adds seasonality and household life cycle as determining factors of farmers' success (Zoomers 1999). In the sustainable livelihoods framework developed by DFID (Department for International Development) five types of capital are highlighted as core assets of achieving livelihood outcomes. These five categories are human capital, social capital, natural capital, physical capital and, finally, financial capital. The human capital includes skills, knowledge and good health which all enable people to take on different livelihood strategies. Within human capital, formal education is one source of knowledge-based capital and the framework is concerned with the provision and access to formal education (DFID 1999). Chambers and Conway (1992) identify several determinants of livelihood strategies that they refer to as 'initial determinants'. A person may be born into a certain livelihood or socialized into a certain livelihood. However, livelihoods may also be chosen by an individual or household and education and migration are highlighted as the two main channels of reaching a desired livelihood. Nevertheless, the ability to choose is restricted by economic factors and more privileged groups have a wider range of choices.
Increasingly, livelihoods have been diversified and it has become more common to have multiple sources of income (de Haan and Zoomers 2005;DFID 1999) and poverty itself is highlighted as a force contributing to this diversification. Livelihood diversification is based on voluntary or involuntary decisions. An aspiration to engage in certain income generating activities reflects a voluntary based decision. By contrast, it might be a response to factors within or outside the individual household that forces diversification. In order to survive, poor people are pushed to diversify their livelihoods (Ellis 2000b). For a livelihood to be socially sustainable, future generations' capabilities need to be maintained or even increased. One way to achieve this is for children to move into occupations other than those of previous generations. Therefore, households tend to invest in education. Human capital is described as one of the key capitals for positive development and the investments in human capital are made by households and governments (Cypher and Dietz 2009;World Bank 2011;Patrinos and Psacharopoulos 2011).
Livelihoods are directly linked to children's welfare and education, including aspects of access to food, basic needs and families' needs for extra labor (Inder et al. 2017;Vimefall 2015). A poor livelihood may thus negatively hamper children's education. On the other hand, education is referred to as a livelihood strategy in order to increase future chances of a formal employment (Ellis 2000a). Still, if educated youth are unable to find a job, households are faced with a potential double loss: children have been in school instead of working, and secondly, they may be unable or unwilling to do farm work after completing their education (Hagberg 2002).
The livelihoods perspective places a strong emphasis on people's ability to act on and recover from stress and shocks, while using their assets in the most profitable way (Chambers and Conway 1992). Criticism towards the livelihoods approach recognizes that not all individual decisions are strategic and intentional and, furthermore, that the level of agency among households and individuals is largely influenced by the actors' social positions (Sakdapolrak 2014;de Haan and Zoomers 2005;Van Dijk 2011).
Rural livelihoods are affected by politics and power beyond the local arena. Structural forces, institutions and access to assets are all related to politics and power (Scoones 2009). This article specifically explores the 2001 reform and its focus on fee free education. The decision to abolish previous school fees was taken at the national level and primary education was to become free of charge and increase access to education for poor families. However, as the capitation grants are not fully reaching the schools, the local schools themselves introduced contributions for various services and supplies in order to be able to continue keeping the schools opened and functioning. The local level outcomes of the policy therefore stand in contrast to its national aims, but with varied implications for different social groups and places.

Policy background
In 2001, the first primary education development plan (PEDP) was launched by the government. This was a five-year plan that described the visions of UPE within Tanzania and declared the new policy framework and identified strategies for reaching its aims. By the time of its launch nearly 5 million children were enrolled in primary schools, however, at the same time, nearly three million children aged 7-13 were estimated not to be enrolled. In order not to overwhelm the education system, the aim of UPE was to be reached step by step. The four main components of PEDP were: (I) expanding enrollment, (II) improving the quality of teaching and learning processes, (III) building capacity within the education system and (IV) strengthening the institutional arrangements surrounding planning and delivery of education (URT 2001, v). The PEDP declares that: 'The Government will abolish school fees and all other mandatory parental contributions from January 2002 so that no child may be denied schooling' (URT 2001, 5). As clearly stated in the document the abolishment does not only include the previous school fee but also all other mandatory contributions. Primary school was to be exempt from fees and no financial burdens should be carried at the household level as a result of primary education.

Methodology
This article is based on a combination of quantitative and qualitative methods. A household survey was conducted in the three villages in order to explore the extent of educational expenditures, and possible differences between villages and socioeconomic groups. To be able to explore the rural households' perceptions and strategies individual interviews and focus group discussions were included in the study. An overview of the data collected is presented in Table 1.

Household survey
A household survey was conducted and in total 209 households were surveyed across the three villages. A sampling frame of all households with children between the ages 7-18 was constructed for each village and households were sampled randomly from this list. The sample is therefore statistically representative for these households at the level of each village.
The number of surveyed households within each village ranges from 67 in village A, 69 in village B and 73 in village C. The sampled share of the total village population ranges from 18-22% depending on village ( Table 1). The sample size was restricted by financial constraints.
The questionnaire was carried out by four field assistants that were trained by the author. The data have been processed in IBM SPSS Statistics 22 and analyzed primarily through cross tabulations and comparison of means between groups and variables. The limited number of households included in the survey makes it difficult to establish statistical significance at the sub-village level, for instance between groups of different socioeconomic standing. As livelihood and education are in focus the main variables include cash income, particular sources of cash income and possession of assets. Additionally, variables on housing standards were included to give a clearer picture of possible rural differences. Variables related to education, for example, amount of annual educational expenditure and drop-out rates have been analyzed. Differences between villages and groups were tested through analysis of variance of means.

Interviews and focus group discussions
The household survey is complemented by individual interviews at the household level and focus group discussions with parents or caregivers 1 as well as interviews with primary school teachers. Respondents for the individual interviews at the household level were randomly selected from a village list of households with children. Focus group participants were selected by someone already participating in the individual interview or local gatekeepers. Each focus group included 3-6 participants.
Teachers were included as they are aware of the current status of primary education in the rural schools today. They are also aware of the local context and the possible obstacles households are facing in relation to obligatory schooling. The selection of teachers was done by the head teacher. After interviewing the head teacher, two additional teachers were included at each school. Selection intended to avoid interfering with ongoing classes as rural schools tend to face difficulties in attracting teachers and thus have a shortage of teachers.
The aim of the qualitative data is to contribute to a deeper understanding of the local context and parents'/caregivers' perceptions of education as well as their educational expenses. Interviews and focus group discussions with parents and caregivers centered on the value of education, future expectations on the next generation and possible problems in primary school education. Interviews with teachers focused on the outcome of the 2001 reform, the current status of primary education, parents' and caregivers' possibility to contribute financially to their children's education and possible obstacles to education within the specific rural context.
The treatment of the first two research questions (which explore the extent of financial burden on a household level depending on village and social group) rely primarily on quantitative data from the household survey, while the analysis used to answer the third and fourth research questions (which focuses on perceptions of education) draws on qualitative data.

Discussion
The three villages included in the study were selected based on their differences in accessibility to urban centers, transport network, and the range of services provided to their residents. The selection was done in cooperation with the two districts in which the villages are located. Each village has one public primary school. Table 2 provides a short description of the village characteristics.
During fieldwork, the difference between the three villages in terms of access to urban centers versus remoteness, and, not the least, variances in welfare and living standards became obvious. Village A, with its location along the highway and frequent public transport has easy access to two urban centers. Even though the travel might take up to one hour, the availability of buses and the smooth drive on tarmac roads adds to the sense of easy accessibility. Villages B and C face limitations in regards to public transport as the bus to Iringa town runs only once per day. When overcrowded, the bus does not take additional passengers, so even this limited service is erratic.
Villages B and C without doubt are located in remote rural areas. Poor roads and long distances give a travel time of at least two hours to the nearest urban center. As mentioned earlier, this is accompanied by poor public transport. However, the villages are relatively well endowed with natural resources, for example forest suitable for timber and in village C pear trees grow everywhere. Unfortunately, the time of harvesting coincides with the rainy season when the village is inaccessible and thus a more extensive commercialization of pear is impossible.
A difference in welfare and housing standard should also be noted. In village A, bicycles and motorcycles appear more frequently on a daily basis than in village C. Village C stands out as having considerably poorer housing standard. The perceived differences in welfare and standards are confirmed in the household survey and, while the villages resemble each other in terms of family structure, results from the survey indicate a discrepancy in livelihood patterns, housing standards and ownership of assets, e.g. mobile phones and bicycles indicating a difference in level of welfare among the villages.
Before returning to the research questions, an overview of the household survey will be presented. Tables 3 and 4 give an indication of family structure and the differences in livelihoods, housing standards and asset ownership between villages A, B and C. Table 3 is a descriptive overview of the family characteristics of the participating households between the three villages in the survey. Considering family size and household type, the data suggest that the three villages to a large extent resemble each other. Nuclear families and skipped generation households (SGH) are slightly more common in village B, while female headed households (FHH) are somewhat more frequent in village A. Village C has the lowest rate of both SGHs and FHHs. The average household size in the study is somewhat higher than both the national and regional means. In the last census, 2012, the national rural household mean was 5.0 while the regional was 4.2 (NBS 2014).
As Table 4 indicates, village A generally stands out as the better off village among the three villages. The level of income is considerably higher in village A compared to the other two villages. The difference is most striking between villages A and C. Additionally, in terms of housing standard and asset ownership village A is almost at all times ranked as number one. Village C is the least well off village both in terms of income, housing standard and possession of assets. Village B falls in between, sometimes being closer to village A considering for example roofing materials and ownership of mobile phones, and at other times being closer to village C as in ownership of bicycles. The gap between villages A and C is large; the average income is more than double that of village A, and a clear difference is identified in asset ownership and standard of living. It is also village C which has the lowest share of households who are able to eat lunch during the leaner seasons.
Only five households in total indicated that their main source of income was generated from formal employment. Not surprisingly, agriculture (181 households) is the first and foremost source of income, followed by Kilolo Village B has a few small shops, a weekly market when the main street is crowded with market stands and people bargaining over clothes, fruit and vegetables. A dispensary is available in the village and the electricity network was being extended in 2014, still it is expensive for individual households to pay the connection fee needed to each house. Water needs to be collected from a nearby river and public transport to urban areas is limited to once a day in each direction C Kilolo Village C has a few small shops but no electricity and no boreholes. Transport network is the same as in village B. The poor road network makes the village inaccessible by car or bus during the rainy season. The distance between villages B and C is half an hour's drive, a distance many residents walk to attend the market in the neighboring village or to charge their mobile phones  informal employment (123 households) suggesting that households engage in various forms of informal income generating activities to diversify their livelihoods and increase their income from agriculture. The informal activities in the study sites include, among others, tree planting and logging, preparing and selling local alcohol, the making and selling of bricks and selling of various items such as firewood and charcoal.

Educational expenses
Despite the governmental promise of making primary education in Tanzania free of charge, this study suggests that the reality has become quite the opposite. When households were asked to rank their expenditures by the following categories: transport, food, education, agricultural inputs, health related expenditures and 'other', education was ranked among the top two expenditures in all three villages. The cost of education, which includes primary school contributions, secondary school fees, uniforms and school supplies, is, together with food, ranked as the highest expenditure for households. In villages B and C, it is ranked as the number one expenditure while in village A it is ranked as the second highest, just after food. These two items combined make up for an extensive proportion of the rural households expenditures constituting on average 85% in village A, 87% in village B and 95% in village C, as a share of total household expenditures. As suggested in the conceptual framework, education closely relates to livelihoods as it is perceived as one of the core assets in achieving positive livelihood outcomes. The empirical data suggest rural households place a large proportion of their income in their children's education. Consequently, less cash is available for other kinds of livelihood investments, for instance livelihood diversification or intensification. Even though costs related to education are ranked highly among all three villages, there is a large difference in the average amount spent on primary education. Households in village A spend the highest average amount of 120,398 Tsh, followed by village B with 91,241 Tsh. Village C has the lowest spending of 88,714 Tsh. On the other hand, when the spending of primary education is seen in relation to average monthly cash income, village C has, by far, a much higher proportional spending than the other two villages. The mean cost of education in village C corresponds to 2.3 average monthly incomes while in villages A and B the yearly primary education costs correspond to 1.3 monthly incomes on average. The difference in education expenditures between the villages is significant at the 10% level. There is a difference, however not statistically significant, in educational spending between different types of households. The data suggest that nuclear families spend 40% more on education than FHHs.
There is a statistically significant difference, (at the 5% level), in average annual spending per primary school aged child between the villages; households in village B spend 19% less per child and village C spends 28% less per child than families in village A. Additionally, children who are raised in SGHs appear to be more vulnerable as these households spend 30% less on school related expenses per primary aged child compared to children in other types of households. This difference is statistically significant at the 5% level.
More than 75% of respondents in the three villages identify primary school contributions as the main problem related to primary schooling. The qualitative data illustrate how these contributions are perceived as mandatory, even if they are officially voluntary. If contributions are delayed, children will be dismissed from school or even beaten. The perception of contributions as mandatory payments is supported elsewhere in the literature (Vavrus and Moshi 2009). According to the Ministry of Education and Vocational Training (MOEVT), contributions are an agreement between parents and the school committee of a certain school. They are contextual and may vary between schools, and individual families should not be punished if they fail to pay (MOEVT 2015).

Implications at the household level
Closely related to the topic of primary education is the issue of child labor as children in many developing countries are withdrawn from schooling to undertake labor and increase household income. Statistics from the International Labour Organization (ILO) indicate that while Asia and the Pacific have the largest number of children in child labor, Sub-Saharan Africa has the largest incidence of 21% (ILO 2013). Child labor interferes with schooling as children are forced to: combine school attendance with out of school work, drop-out of school or not enroll (ILO 2015). Research indicates numerous reasons for the high incidence of child labor including the lack of secondary schooling options influencing child work already at primary school age level (ILO/ IPEC-SIMPOC 2007; Burke and Beegle 2004); high fertility rates and a traditional agricultural sector dependent on child labor (Admassie 2002). Most frequently, however, the link between child labor and high poverty rates is highlighted (ILO 2013;Admassie 2002;Webbink, Smits, andde Jong 2012 andBandara, Dehejia, andLavie-Rouse 2015).
In contrast to ILO statistics, the data collected for this study does not show that children are being withdrawn from school in order to do farming or domestic work. The number of households who had their primary school aged children withdrawn from school to do domestic work ranged from 0% to 1.5% depending on the village. The number of households that withdraw their children to do farming was slightly higher and ranged from 1.5% to 4.1%, depending on the village. In relative, as well as absolute terms, this suggests a very limited conflict between schooling and child contributions to household labor. The difference between villages is statistically significant at the 5% level for both domestic work and farming.
The households indicated a slightly higher rate when asked if any of their children have ever dropped out of primary school. The variation between villages was also larger in the case of retrospective drop-out rates. This time village B has the lowest drop-out rate at 2.9%, followed by village A at 6.0%. The rate of village C is more than double that of village A, 13.7%. This can be compared with the national drop-out rate of primary education of 1.5% for children aged 7-13 (NBS 2014). Previous research suggests children from poorer households are the ones to drop out of school (Hunt 2008) and as indicated by the household survey village C is the poorest village considering both level of income and living standard. This might therefore, at least partly, explain the high drop-out rate in village C. Other possible reasons might be the average household size and difficulties in keeping all children enrolled, walking distance to school and the seasonality of labor demands in agriculture. According to the head teacher in village A, there is no problem regarding attendance in this school. In villages B and C, though, the head teachers identify a seasonal problem of attendance during March and April when it is time for harvesting fruits. Still, this is referred to as a seasonal problem not leading to permanent drop-outs. Interesting though, neither March nor April are ranked as labor intensive by the households themselves. The two most labor intensive months are instead for village A November and December, for village B October and July and for village C September and October. The villages' geographical location and topography reflect the variances in ranking.
Throughout the individual interviews only one father said he could consider withdrawing his primary aged children from school. This was an HIV-positive man and due to his condition he was no longer able to work full time within the agricultural sector and was concerned for his future livelihood and the cost of education for his children.
Considering the high regional NER for primary education in Iringa as well as the positive parental value of education (further discussed below), the low share of children working in domestic work or in farming during school days is perhaps not that surprising. However, there is a widespread consensus in the individual interviews that outside school hours children contribute their labor to farming and domestic work.
Although school attendance does not directly conflict with the households' need for labor, education expenditures need to be handled by the households. One way, explained through interviews, to manage the contributions is the possibility to add informal income generating activities, like making and selling traditional alcohol, to the income generated by traditional farming. Similar arguments related to the need of lowincome groups to diversify their livelihoods in order to meet their basic needs are made in the literature (de Haan and Zoomers 2005;Ellis 2000b). Another possibility is to reduce the indirect cost of education, for example, children are not provided with new uniforms yearly, instead they are worn until torn. More drastically, a focus group of fathers in village C, stated that some households even depended on selling their plots of land to be able to pay for the education of their children. Other households, they argue, are relieved and happy when their children do not pass the Standard VII exam, as they are unable to provide funds for secondary education. According to one of the female focus groups in village B, even children need to engage in casual labor during weekends to earn money to cover the costs of education.
Financial support to pay for children's education also comes from outside the households. Although differences are not statistically significant between villages, 12-18% of the participating households reported that they received contributions from outside the household for their children's education.
One focus group in village A, consisting of elderly men caring for orphans, explained that previously various funders helped to pay for school contributions but as the support of these benefactors was only temporary the households currently depend on themselves. The focus group participants acknowledge that due to their age they are unable to work and they fail to pay for all contributions requested.
When contributions are delayed or defaulted on, it is generally the children that suffer the most. They are usually dismissed from school and fall behind in their school work. A focus group of mothers in village C questions how teachers handle the requests for contributions: [W]hy do the teachers ask the pupils to turn back home for a certain contribution? We parents, we are responsible for the schooling of our children, so let the teacher ask us to pay the contribution because to ask a pupil to go home for contribution while others continue with the studies will affect the pupil. The teacher should send a letter to the parents to ask for the contribution and let the child stay at school. So, when the pupil arrives at home and … find that that the parent is not at home so he or she will not feel comfortable to go back to school, then a child can decide to stay at home or go somewhere else. Other teachers beat the children, and not helping them to understand the subject, so they make them not comfortable to study. (Focus group 1, village C, 10 November 2013) A father from village B argues that his child needed to stay at home and that 'he is not comfortable to go back to school because the teacher can beat him' as the requested contributions are not paid.
In recent literature FHH and SGH have been identified as particularly vulnerable and at higher risk of poverty (Eriksen, Brown, and Kelly 2005;Zimmer and Das 2014;Mokomane 2012 andBörner et al. 2015). This is also reflected in this study as data show that both FHHs and SGHs have lower average incomes than nuclear families. The mean income 2 for all three villages is 105,274 Tsh/month while FHHs have a mean income of 71,250 Tsh/month and SGH a mean income of 51,950 Tsh/month. However, the difference is not statistically significant between groups. Considering that village C has the lowest rate of participating households representing the two categories of FHH and SGH in the household survey, its low level of income and living standard becomes even more remarkable.
Social mobility and hopes for the future Future generations' capabilities need to be maintained or increased in order for a livelihood to be socially sustainable, and a possible way of increasing future generations' capabilities is investments in education. However, considering the amounts involved and the frequencies of contributions one might think families' willingness to invest in education would be affected. The qualitative data, nevertheless, suggest there is a strong belief that education will increase both the opportunities for certain livelihoods or, rather avoid the specific livelihood of farming, or, at least, increase the productivity of farming as a livelihood. A better education is accordingly perceived to improve children's lives either through employment or improvements in agriculture. While future employment and a more modern agricultural sector, both leading to improved living standards, stand out as the main reasons for investing in education, one focus group in village C, consisting of fathers, addresses a completely different issue. Education, they argue, must be invested in to avoid problems: 'Another reason is that if we will not send our children to school they will face so many problems like diseases, pregnancies' (Focus group, 2, village C, 10 November 2013).
Employment is highlighted as the goal of education; through formal employment the child will become independent and not be a burden to the parent any longer and through formal employment the child will have a better life, live in better housing and be able to help its family financially. If not formally employed, parents expect their children to be successfully selfemployed as they through education will have attained the knowledge of reading, writing and mathematics. Education will help their children not to be poor anymore, as well as to improve the situation for their families.
If not employed, the relationship between education and improved agriculture and the commercialization sector is highlighted. This sector is believed to become more modern through agricultural machinery, mechanized weeding, inorganic fertilizer as well as the knowledge of using it. Children will be able to increase yields and thus be able to increase commercialization.
During individual interviews, parents and caregivers stress that the reason for investing in their children's education is for them to have a better and easier life; for them to become independent and to improve not only their own situation but also to improve the situation of their parents. Teachers are of a somewhat different opinion. Rural parents are not sufficiently aware of the value of education it is claimed. Parents are not as supportive as needed regarding school supplies and they, at times, prefer for their children to stay out of school and help their families with farming as a teacher in village B argues: [S]ome parents do not see the importance of education.
How do you notice this?
Because the parent asks the pupil to stay home and go on farming. (Teacher 1, village B, 6 November 2013) Despite being a heavy burden at the household level, the data suggest investments in education to be fruitful. In households in which the head of household had some kind of education, the monthly income was 60% higher compared to households in which the head of household lacked education, indicating the importance of human capital.

Summarydifferences by village and subgroup
The quantitative data suggest a difference between the three villages both in terms of income and housing standards as well as the proportion of total expenditure devoted to education, with village A as the better off village in terms of income, asset ownership and housing standard. The differences in these regards are striking in comparison to village C, which throughout appears to be the least well off village among the three. Yet, proportionally, households in village C spend the highest share on education. The qualitative data suggest fewer differences between the villages. In all three villages and among the different types of households, education is motivated as necessary to achieve positive change for both the individual and the household, to enter formal employment and thus increase incomes, to improve farming output and improve living standards. Education is highly valued in all three field sites and believed to be the most important asset parents can give their children.
Furthermore, there is a general perception of contributions being problematic and a heavy burden for households to bear. Education is obviously a relatively high expenditure in all three villages regardless of household type. Poverty is widespread in the three villages. Nevertheless, the proportion of expenditure used for education was largest in village C, and qualitative data from this village also shows households resorting to drastic measures such as selling land to be able to send their children to school. Thus, while investing in education has the potential of positive livelihood outcomes, in the case of children not completing education, these households are even more vulnerable as they have sold another valuable asset, namely access to land.
In terms of household type, FHHs and SGHs indicate less yearly spending on primary school expenses per child than nuclear families suggesting these two types of households face harder conditions than nuclear families.

Conclusion
The policy that was introduced by the Tanzanian government in 2001 stated that: No child should be denied the opportunity to participate in education because of poverty, gender, disability, or because of a lack of school uniform, fees or other parental contributions, or because of a lack of school facilities, materials or teachers. (URT, v) Sixteen years after the decision was taken to make primary education free of charge it is concluded that this goal is yet to be fulfilled. The capitation grant has throughout this time been 10 USD/pupil, however, in reality less than half of this amount reaches the schools. Instead, a heavy financial burden falls on households to keep the school system functioning and to be allowed to have their children attending school.
Education is currently one of the largest expenses for rural families. While the three villages in the study show a great difference in terms of access to urban centers, available services, housing standard and income, the data indicate less difference regarding family structure, value of education and the perception of education as an immense expenditure for the individual households. However, the proportional expenditure varies with the most remote village having the highest proportional educational spending with more than two monthly incomes yearly being spent on education. The same village has the highest drop-out rate in primary school.
Interviews and focus group discussions in all three villages suggest a strong belief in education. Without primary education it is perceived to be impossible to move out of poverty and to make any changes in society. Education is considered as necessary to become independent, to be formally employed or to work in a more modern agricultural sector. Education is highly valued and believed to be the key to change livelihoods or increase the productivity of livelihoods; that is either to move out of the agricultural sector and enter formal employment or to considerably increase agricultural productivity as well as the profit from improved agricultural commercialization. Data from Iringa Region suggest that households are forced to engage in multiple activities to secure livelihoods and to bear the living costs, of which education is one of the two top expenditure items.
While Tanzania is experiencing a positive trend in its strive towards UPE, the reform has also resulted in a substantial financial burden for rural families, especially poor households. In the poorest village, with the highest proportional educational spending, nearly 50% of households indicate that they are unable to eat lunch during the leaner seasons. Nevertheless, parents and caregivers will, most likely, keep doing their utmost to secure a livelihood that will allow them to keep their children in education, at least basic education, as education is highlighted as the one thing parents can give their children to secure their future livelihoods, while enhancing also their own future prospects.

Notes
1. Due to the AIDS-epidemic, Iringa has a high-orphan rate.
Therefore, caregivers other than parents are included in the study. Children are usually cared for by relatives, for example, grandparents and live in so called skipped generation households (SGHs). 2. Including all cases.