What Is the Point of Schooling? Education Access and Policy in Tanzania since 1961

Learning is not always the goal of schooling. As policymakers around the world tackle the ongoing global “learning crisis,” it is important to understand the logics behind the design of education systems. Using a political settlements approach, this article documents the evolution of education policy and associated outcomes in Tanzania in tandem with shifts in elite perceptions of the goals of schooling. Tanzania has experienced four phases of political settlements, each of which encapsulated a different policy posture regarding the goal of schooling. The article also shows that electoral incentives are decidedly biased in favor of increasing access to schooling, even at the expense of quality. These findings advance our understanding of the politics of education policy and the persistent failures to invest in improving learning outcomes in many countries around the world.


Introduction
"Primary School education is indeed what we mean by education in Tanzania. Post-primary education is that which will prepare a few qualified individuals for those special kinds of service which need more training. It cannot be more than that while our National Income per head is so low" - (Nyerere, 1982, p. 30) The world is experiencing a "learning crisis" (Languille, 2014;World Bank Group, 2018).
While the last five decades have seen tremendous improvements in school enrollment rates in middle and low-income countries, the gains have been accompanied by a glaring stagnation or decline in school quality (Sifuna, 2007;Nestour, Moscovix and Sandefur, 2021). In other words, schooling has not translated into learning. This raises the question, why did politicians and policymakers preside over stagnating or declining education quality in the last 50 years?
Using a political settlements approach, this article answers this question with evidence from Tanzania. 1 In doing so it answers the question: what has been the point of schooling in Tanzania since independence? The main argument herein is that changes in the official policy objectives behind schooling -informed by shifting political settlements -explain the observed stagnation in education quality and the current learning crisis. Simply put, learning has not always the intended goal of schooling. 2 Furthermore, because of the logics of path dependence, past policy choices often constrain policymakers' ability to ensure that schooling translates into learning. Therefore, interventions aimed at addressing the global "learning crisis" are more likely to succeed if modeled to internalize different country's specific education policy histories and their underpinning political settlements. In countries that are still plagued by access gaps, electoral politics are decidedly biased in favor of increasing access, even at the expense of the quality of learning. I present evidence from multiple sources -including administrative data on school enrollment, government spending, and economic indicators; content analysis of party legislation, 1 The analysis herein only covers mainland Tanzania. 2 See Paglayan (forthcoming) for related findings in Latin America. manifestos and important policy documents; survey data from the Demographic and Health Surveys (DHS) and Afrobarometer; and interviews with key stakeholders in Tanzania's education sector. Combined, these data allow me to construct a thick description of the evolution of education policy and its impacts in Tanzania from multiple vantage points.
I show that Tanzania, as a matter of policy, rationed access to primary education before 1974 and secondary education before 2006. Nyerere's claim above was not cheap talk.
Throughout this period, only about 10 percent of primary school graduates attended secondary school. I also show that Tanzania's education policy evolved through four different periods, each with its own logic of schooling. In the first phase (1961)(1962)(1963)(1964)(1965)(1966)(1967), learning was seen as important for producing a limited skilled workforce to Africanize the state bureaucracy and economy. Thereafter political and ideological objectives drove education policy -first under the socialist Education for Self Reliance era (1967)(1968)(1969)(1970)(1971)(1972)(1973)(1974)(1975)(1976)(1977)(1978)(1979)(1980)(1981)(1982)(1983)(1984)(1985) followed by the reformist era (1985)(1986)(1987)(1988)(1989)(1990)(1991)(1992)(1993)(1994)(1995). It was not until 2005 that the government initiated reforms to improve learning with the introduction of the Competence Based Curriculum. Previous curriculum reforms (1967, 1978, and 1997) mainly focused on the content of the curriculum and not pedagogical improvements designed to improve learning outcomes. 3 Indeed, to address growing demand for qualified teachers in the 1970s, the government halved the training period for qualified Grade A teachers (O-Level graduates trained for 2 years) and then recruited primary school graduates as Grade B (with four years of training) and C teachers (one year of training).
The recruitment of Grade B teachers only stopped in 1996 (Meena, 2009).
Tanzania's education outcomes reveal the deep structural challenges facing many developing countries. For example, as of 2017 only 22.9% of women aged 15-49 had secondary or higher education. In the same vein, a mere 79,348 A-Level (Forms V-VI) students sat the 3 In 1967 the Education for Self-Reliance reforms sought to match the curriculum content with Tanzanians' lived realities and to promote socialist agrarian development. The 1979 reforms emphasized vocational training and the "Politics of Agriculture." Finally, in 1997 reforms sought to codify ideological changes from socialist single party rule to a multiparty market economy. Socialist political education was scrapped and replaced with Social Studies. In addition, new subjects like Computer Studies were introduced as well greater emphasis on science education (Nzima, 2016).
Advanced Secondary Education Examination (ACSEE) in preparation for admission into tertiary institutions. 4 This represented just 8.7% of the 2017 cohort that graduated from O-Levels (Form I-IV). In 2018, the latest year for which data is available, the transition rate from Form IV to Form V was 21.7%. 5 This is despite the fact that in 2018 79. 3% (285,885) of Form IV candidates passed the Certificate of Secondary Education Examination (CSEE).
Less than 4% of eligible children in Tanzania attend A-Levels. 6 Continuing lack of space leaves about 300,000 children out of secondary schools. The challenges to effective learning begin early. An estimated 10% of all Standard 1 pupils repeated the grade in 2017. In 2016.79% of all students enrolled in primary school had repeated a grade. 7 What explains these outcomes? The answer lies in Tanzania's varying political settlements and their influence on the government's ability to shape curriculums and ration access to schooling for instrumental political and developmentalist ends. Following independence (1961)(1962)(1963)(1964)(1965)(1966)(1967), manpower planning had a dominant influence on policy. The goals of schooling were to promote cultivate a civic identity and train a skilled workforce needed to Africanize the state bureaucracy and economy. 8 To this end, and to conserve scarce financial resources, the government explicitly targeted a ceiling of 50% primary enrollment rate (Resnick, 1967).
Thereafter (1967)(1968)(1969)(1970)(1971)(1972)(1973)(1974)(1975)(1976)(1977)(1978)(1979)(1980)(1981)(1982), the policy of Education for Self-Reliance (ESR) under socialism defined the goals of schooling. ESR emphasized the acquisition of skills needed for rural agrarian development (Nyerere, 1968;Cooksey, 1986). At the same time, the government rationed access to secondary schooling -including through restrictions of private schools -with primary school structured to be a terminal qualification. Between 1974-1978 the government imple-4 Basic education in Tanzania consists of 2 years of pre-primary, 7 years of primary level (Standard I-VII), four years of Ordinary Level (O-Level) secondary education (Form I-IV), and 2 years of Advance level (A-Level) secondary education (Form V-VI). Standard VII candidates sit the Primary School Leaving Examination (PLSE). The Certificate of Secondary Education Examination (CSEE) marks the end of O-Levels. Form VI candidates sit for the Advanced Secondary Education Examination (ACSEE) to qualify for tertiary institutions.
5 United Republic of Tanzania (2020) 6 These figures contrast sharply with those in Kenya, where in 2019 679,222 students sat the Kenya Certificate of Secondary Education (KCSE) examination to qualify for tertiary education. 7 URT, Primary Teach Deployment Strategy, 2017. 8 This was part of the government's policy of manpower planning (Hunter, 1966;UNESCO, 1968).
mented universal primary education (UPE) by waiving fees in response to popular demand (Sabates, Westbrook and Hernandez-Fernandez, 2012). However, the rationing of secondary education and focus on vocational skills for agrarian development persisted.
UPE created twin problems that defined education policy between 1982-1995. Increased primary enrollment heightened demand for secondary education. The policy also exerted enormous pressure on the fiscus. To address the first problem, the government relaxed its restriction of the growth of private secondary schools to absorb the excess demand, even as it maintained its policy of rationing public secondary education. Economic crises in the 1980s forced the government's hand with regard to the second problem. School fees were reintroduced, causing a decline in enrollment rates. Meanwhile, economic reforms eroded the socialist underpinnings of ESR thereby setting the stage for a redefinition of the goals of schooling (Samoff and Sumra, 1994).
In 1995 the government formally committed to a policy of universal primary and secondary education. Importantly, the policy viewed basic education as preparation for higher levels of schooling in the quest to produce skilled workers for a 21st century economy (URT, 1995). The reintroduction of UPE took place in 2001. Unlike in the 1970s, the government's response to increased demand for secondary education after UPE was a massive expansion in access beginning in 2006 (Habyarimana, Opalo and Schipper, 2021). The policy of universal secondary education was formally introduced in 2015. It is not a coincidence that these policy shifts came around election years -1995, 2000, 2005, and 2015. 9 While increasing access still dominates policy, Tanzania has also made efforts to improve learning outcomes. Tanzania's legacy of rationing access and under-investing in learning leaves it lagging on both fronts. Due to increased electoral competition, politicians are strongly incentivized 10 The idea that schooling may serve ends other than learning is not unique to Tanzania (Paglayan, forthcoming). Neither is official anxiety about a perceived "learning crisis." Hacsi (2002), USG (1983), andKatz (1987) document similar shifting goals of education and concern about learning outcomes in the United States of America.
11 After 13 years of the union between Tanganyika and Zanzibar to create Tanzania, TANU merged with the Afro-Shirazi Party of Zanzibar to form CCM in 1977. Despite the re-introduction of multiparty politics in the early 1990s, CCM has maintained its electoral hegemony both in Zanzibar and on the mainland. to invest in increasing access to schooling, at times at the expense of learning. This is for the simple reason that, credit claiming being the currency of electoral politics, schooling is more readily visible and attributable than learning. In addition, investments in learning are expensive, require complex reforms of the education sector, and take time to yield results.
Overall, the evidence from Tanzania suggests that the legacies of historical policy choices and contemporary electoral politics in low-income states create incentives that militate against a timely solution to the global learning crisis.
This article advances our understanding of the politics of education. A large literature examines the political incentives for public investments in education (Franck and Rainer, 2012;Stasavage and Harding, 2014;Kramon and Posner, 2016;Habyarimana, Opalo and Schipper, 2021) These works posit that electoral competition and/or responsiveness are important drivers of inputs, outputs, and outcomes in the education sector. A related literature focuses on the logics of public investments in schooling (Weber, 1976;wa Thiong'o, 1986;Farnham, 1994;Marginson, 1997;Paglayan, 2021, forthcoming). Going beyond electoral incentives, these works explore the instrumental uses of schooling by states -including identity formation, civics education, and habit formation in preparation for the labor force. I synthesize these two strands to show how elite's instrumental use of schooling interacts with the politics of mass demand for education.
More broadly, this paper highlights in the importance of historicizing education policy in low-income countries. First, the current learning crisis has deep historical roots -including legacies of under-investment in both access and quality, the impacts of low fiscal and bureaucratic capacity on investments in quality, external pressures from global policy entrepreneurs and institutions, and shifting goals of schooling from the perspective of dominant elite coalitions. Second, given the parallels between the current learning crisis and experiences during the last expansion of access in the early postcolonial period, a historical approach can facilitate a better understanding of how to balance the twin challenges of persistent access gaps and deteriorating learning outcomes in many low-income states.

The Politics of Schooling vs Learning
Education is one of the most important public goods provided by modern states (Boli, Ramirez and Meyer, 1985;Ansell and Lindvall, 2021). It is also deeply political. On the demand side, households are incentivized to be stakeholders -whether due to concerns about education's social and political functions or the perceived promise of economic returns to schooling. 12 On the supply side, ruling elites are typically vested in education as a means of molding national identities, socializing loyal citizens, or creating knowledgeable and productive workers. 13 For these reasons, several works find positive correlations between increased electoral competition and investments in mass education (Kaufman and Segura-Ubiergo, 2001;Brown and Hunter, 2004;Stasavage, 2005;D'Arcy, 2013). At the same time, historical evidence suggests that elites often shape public demand for both access and content of schooling (Alesina, Giuliano and Reich, 2021;Paglayan, forthcoming). Below I provide a conceptual framework for synthesizing these two features of the politics of education.

Unbundling Education
The bundled nature of education raises questions about how electoral competition conditions politicians' choices of investments in the sector. The education sector includes different components such as physical infrastructure, teachers, curriculum content, quality control, school management, parent or community contributions, and predictable government fiscal outlays. These components differentially contribute to schooling and learning. Having enough space in classrooms, regardless of the quality of teachers, content of the curriculum, or level of parental/community involvement, can increase access to schooling. However, translating schooling into learning requires investments in the latter components that are often not readily tangible -and which come with significant costs and added complexity.
12 These political demands may be more salient in democracies, but are also present in autocracies. See Paglayan (2021) for a discussion of autocratic incentives for education expansion.
Given the divisibility of education into multiple components, politicians can make differential investments conditional on prevailing incentive schemes. Importantly, as multi-task agents with an eye on re-election (Holmstrom and Milgrom, 1991;Mani and Mukand, 2007) , politicians are incentivized to invest in aspects of education that are easily visible and attributable from voters' perspective (Stasavage and Harding, 2014;Habyarimana, Opalo and Schipper, 2021). Supplying schooling (e.g., building physical schools) is ordinarily more visible and attributable than investments in learning (e.g., through curriculum reform, teacher training, remedial classes, and other pedagogical interventions). Furthermore, parents are more likely to attribute their children's academic successes to innate intelligence or parental investment than politicians' effort.
Levels of administrative and fiscal capacity may further complicate these electoral dynamics. In many countries the education sector attracts the largest share of government spending.
However, many low-income countries struggle to reach the recommended optimal fiscal investments of 4-6% of GDP in spending (World Bank, 2021). Given the scale and complexity of operations involved, proper management of the education sector demands administrative and fiscal capacities that are lacking in many (developing) countries. Consequently, even politicians that are intrinsically motivated to improve both access to schooling and learning may lack the administrative or fiscal means of achieving their objectives. Forced to consider the tradeoff on how to allocate their effort, the combined effects of electoral incentives and lack of administrative and fiscal capacity reinforce the political bias towards the provision of schooling -often at the expense of learning. 14

Political Settlements and Education Policy
In addition to the electoral and fiscal-administrative factors outlined above, coalitions matter for education policy formulation and implementation. Policy persistence is predicated on continued support from pivotal actors throughout the administrative apparatus -typically referred to as dominant "advocacy coalitions" (Schlager, 1995). Such coalitions may coalesce around specific interests or ideas and influence policy in ways that are independent of electoral, fiscal-administrative, or institutional variables. Stated differently, politicians and policymakers do not simply react or anticipate societal demands, but can also impose their vision on society (Skocpol, 1980;Skowronek, 1982). This is especially true in contexts where institutions such as legislatures and bureaucracies are weak and unable to constrain the behavior of pivotal policy actors in predictable ways. To explore these dynamics in the context of education policy, this section applies a political settlements approach.
The political settlements approach is a useful framework for evaluating the impact of coalitions on education policy in weak states. The framework posits that the distribution of power among administrative/policy coalitions determines the impacts of institutions and policies (Khan, 1995(Khan, , 2017. Therefore, understanding how institutions and policies shape outcomes requires knowing the coalition politics that underpin successful policy formulation and implementation. This is especially true in contexts with weak institutions. In such polities, what matters for outcomes are not formal rules or institutions but "the distribution of organizational power and mobilization capabilities" (Khan, 2017, p. 646). Overall, a political settlement is a particular distribution of organization power (e.g., manifested as a policy coalition) that is capable of reproducing itself over time. 15 A sector as complex as education -especially in weak states -readily lends itself to a political settlements analysis. Managing the sector involves balancing policy ideas and fiscaladministrative capacity against the realities of mass (electoral) politics, while facing pressure from interest groups. Under these conditions, the ability of institutions to aggregate ideas and 15 This article does not consider transaction cost institutional analysis and the political settlement approach to be in tension. Their respective explanatory powers is dependent on the strength of institutions. The political settlements approach has greater explanatory power in contexts with weak institutions -perhaps due to isomorphic mimicry (DiMaggio and Powell, 1983;Pritchett and de Weijer, 2010). This conceptualization is consistent with the general idea that institutions are only as strong as the dominant coalitions within them (Stasavage, 2003;Opalo, 2022a).
interests depends on the prevailing balance of power across pivotal coalitions within a given polity. In particular, if institutions are weak, dominant policy coalitions are able to shape both ideas about what education ought to be about and the manner in which such ideas are implemented -regardless of the formal institutional channels of policy formulation and implementation. This is in contrast to situations where there is greater predictability of policy lifecycles -e.g., the passing of legislation, budget appropriation, and policy implementation by a competent state bureaucracy subject to legislative oversight.  Education policy may not always be salient. And even when salient, the impact of political mobilization for specific policy outcomes may be blunted by coalitions dominating the political system -often with a strong bias for schooling and not necessarily learning. Finally, structural factors like the administrative and fiscal capacities of states shape the feasible set of policy outcomes -including levels of access or possible investments to improve quality of learning. These observations are an important corrective against assuming a dominant influence of mass politics and electoral incentives in driving education policy and realized outcomes. They are also an invitation to interrogate the historical evolution of education policy in developing countries without presuming a fixity of the goals of schooling. The rest of this paper does this with a case study of education policy in Tanzania since 1961.

Data and Methods
This article relies on multiple data sources to describe the evolution of education policy in Tanzania. Major policy documents provide a general understanding of the government's formal policy objectives. These include the Education Act (1969,1978), TANU Guidelines The analysis herein is primarily descriptive, with the goal being to show shifts in education policy objectives conditional on the prevailing political settlement and the resultant effects. 18 I rely on a political settlements approach to construct four periods marked by the predominance of specific ideas about the role of education in Tanzania. The claim is not that the dominant ideas were unchallenged or that there were discrete periods. The observable political, economic, and policy changes were certainly not always coterminous; and outcomes often lagged policy. TANU and technocrats within the public bureaucracy often competed for policy influence. Furthermore, despite its hegemonic status, TANU's policies were not always popular with the mass public. At times, popular pressure or economic crises forced dominant elites to implement policies that were not entirely consistent with their preferences. The next section describes the different political settlements summarized in Table 1.

Political Settlements and Schooling Objectives
Political settlements and their respective underlying ideas about the role of education shaped the evolution of Tanzania's education sector. This section documents the various features of four political settlements since 1961 with a focus on how intra-party (ideological) competition, the state administrative structure, economic change, political reforms, and global policy diffusion impacted education policy.

Party-Based Pragmatism and Manpower Planning
The first political settlement (1961)(1962)(1963)(1964)(1965)(1966)(1967) was marked by the hegemony of the ruling party, TANU, under the leadership of President Julius Nyerere. During this period manpower planning was the driving logic of education policy (Hunter, 1966;Resnick, 1967). This meant providing access and high quality education to only a limited share of the school-age population who would become part of the managerial class in both the private and public sectors. As a matter of policy, the government "[restricted] firmly the further expansion of recurrent expenditure on education until it [could] be better matched (and paid for) by economic advance" (Hunter, 1966, p. 13). Less than a third of children were able to attend primary school. Secondary enrollment was even lower, at less than 3 percent. In addition, the curriculum was Africanized and infused with civic education -including through the adoption of Swahili as the language of instruction in primary schools. Finally, the administrative elements of education were decentralized, urban and rural authorities having significant influence on policy implementation. Overall, TANU's political hegemony enabled the government to successful suppress mass expectations of a rapid expansion of schooling.
Within TANU, different factions jostled for power and influence along two main cleavages.
A national/local cleavage pitted national leaders against local elites (including bureaucrats) who dominated the productive sectors of the economy and the architecture of local government (Picard, 1980). National leaders wanted to subordinate education to nation-building and centrally planned developmentalism. Local elites saw education as means of training competent workers and administrators. The ideological cleavage pitted pragmatic modernizing nationalists against ideologue socialists (Samoff, 1994). Both wanted to Africanize the upper echelons of the economic and political realms, albeit for divergent ends. The nationalists included upwardly mobile capitalist-leaning economic elites and the small bureaucratic cohort that was inherited from the colonial era. The socialists, inspired by Nyerere, envisioned central planning as the means of achieving accelerated modernization and economic prosperity in Tanzania (Nyerere, 1968;Molony, 2014).
The factional balance of power and Tanzania's subnational administrative structure enabled a pragmatic approach to education policy. Rural authorities and urban municipalities retained significant influence on local education policy -including school construction -as was the case under colonialism. The 1962 Education Ordinance granted Local Education Authorities fiscal and administrative powers in the education sector. This policy reflected the government's desire to deflect responsibility over education access to local authorities, and to limit the financial burden on the national government. 19 This produced significant regional inequalities in education access and outcomes. Eventually, the national/local and ideological contestations were decisively settled by the late 1960s on the back of a systematic dismantling of alternative centers of power that may have posed a challenge to TANU and Nyerere's authority.
The eventual centralization of education policy reflected broader political developments in Tanzania. In 1964 the umbrella trade union organization was reconstituted as an affiliate of TANU (Chambua, 2002). Opposition parties were abolished in 1965 (Hyden andLeys, 1972), essentially obviating any possibility of the Tanzanian parliament providing checks on executive power (Tordoff, 1977;Opalo, 2019). Except for a brief military mutiny in 1964, throughout his tenure  Nyerere never faced any real threat to his authority. His total dominance was reflected in the public image of consensus politics (van Donge and Liviga, 1986). However, as noted above, the national consensus was pragmatically balanced against subnational variation in actual policy implementation.

Rationing of access was official policy. While unveiling Tanzania's First Development
Plan in 1964, Nyerere argued for a "carefully planned expansion of education," with the goal of education being to "equip Tanganyika with the skills and the knowledge which is needed if the Development of [Tanganyika] is to be achieved." A priority was to be given to adult education since, in his view, the returns to educating children would take long to materialize.
Nyerere also signaled that he understood the need to expand secondary and technical education. However, in the same speech, he emphasized the importance of efficiently utilizing Tanzania's meagre resources, observing that "Children entering university this year will still be at university in 15 years time!" 20 As such, there was need to prioritize teacher training, educating an optimal number of secondary and university students, while at the same time rationing overall access in recognition of Tanzania's fiscal-administrative constraints.

Education for Agrarian Self Reliance
The Arusha Declaration (1967) outlining socialism under Ujamaa marked the beginning of the second political settlement (1967)(1968)(1969)(1970)(1971)(1972)(1973)(1974)(1975)(1976)(1977)(1978)(1979)(1980)(1981)(1982)(1983)(1984)(1985). If the first political settlement was characterized by pragmatism in the face of intra-TANU factional competition and subnational variation, the second was dominated by Ujamaa and increasing (albeit not total) centralization of education policy. TANU came to dominate education policy at the expense of technocrats in the education ministry. Mirroring ongoing economic reforms, schools were nationalized in 1969 (Carnoy and Samoff, 1990;Temu and Due, 2000) -thereby foreclosing potential sources of deviation from official goals of schooling. A reorganization of local government centralized education policy. In 1969 Local Education Authorities were created to manage schools at the local level -a fact that contributed to continued subnational variation in school attainment. 21 Furthermore, the elimination of council elections diluted localized political incentives to increase access to schooling. Due to agglomeration effects, Ujamaa villagization facilitated the expansion of access to primary education (Maro, 1990). Universal Primary Education (UPE) was introduced in conjunction with adult literacy campaigns.
The most important features education policy in this phase were the rationing of secondary education and employment of primary education to the service of civic education, nation-building, and socialist agrarian development. 22 Many of these changes were encapsulated in the Second Five Year Development Plan (1969)(1970)(1971)(1972)(1973)(1974). The plan envisioned the achievement of UPE in 1989 and gradual expansion of secondary and tertiary education in tandem with labor demands. However, due to pressure from the TANU ranks, the Third tion policy by subordinating government operations to the party (Hartmann, 1983). 23 The guidelines urged the government to reduce subnational disparities in the education sector.
In 1972 the system of local government was abolished (including elected local councils) to check lingering subnational particularistic interests and opposition to national policies and to rationalize the implementation of Nyerere's developmentalist agenda (Hyden, 1980;Picard, 1980). 24 Compulsory villagization soon followed (von Freyhold, 1979) and with it the abolition of cooperatives in 1975 (Eckert, 2007). The net effects of these changes were the 23 Mwongozo Wa TANU, Dar es Salaam, 1971 24 The new system of decentralization (de-congestion) sought to replicate national level line ministries at the local level in order to equalize the provision of essential public goods and services. Tanzania's fiscal-administrative capacity. 25 Importantly, the Musoma Resolution caught the education bureaucracy by surprise (Omari et al., 1983) -a sign of the party's newfound supremacy over the state bureaucracy with regard to policymaking.
Until 1972 local governments were in charge of the basic education system, had elected officials, and could raise their own revenue through taxation (Semboja and Terkildsen, 1994).
They could therefore subvert the top-down policy directives or innovate during implementation. This created the distinct possibility of reproducing existing subnational disparities 25 As late as 1973 primary enrollment rate was only 50 percent (Samoff, 1987). in educational and economic outcomes (as well as in other sectors). To avoid this problem, the government abolished the local government system and replaced it with a system of regional administrations. The goal was to rationalize the allocation of scarce resources for development and to have the regions (instead of smaller sub-regional units) better coordinate the implementation of national government policies, including ESR (Samoff, 1979;Mbilinyi, 2003;Therkildsen, 2000).
Despite TANU's hegemony, residual tension existed between the party's ideological policy aims and their practical implementation through the state bureaucracy (Kiondo, 1989). In particular, Nyerere had to balance TANU's (and later CCM's) impulses for poorly planned rapid reforms against the stated policy objective of rationing access to education in line with the country's fiscal-administrative capacity. He also had to contend with the problem of subnational variation in policy implementation. Administrative reforms in 1974 (dubbed "decentralization") sought to decongest the central bureaucracy and replicate ministerial presence at the subnational level with little effect on policy implementation (Hyden, 1980). This was in part due to TANU's influence on subnational administrative organs. 26 The consolidation of single party rule between 1965 and 1980 politicized the policymaking process and decoupled planning from fiscal-administrative considerations. Ineluctably, failure followed.
Importantly for education policymaking, the combined effects of these failures and economic crises boosted the influence of technocratic bureaucrats.

Structural Adjustment of Education
Economic crises and policy failures under Ujamaa precipitated a new political settlement focused on macroeconomic reforms (1985)(1986)(1987)(1988)(1989)(1990)(1991)(1992)(1993)(1994)(1995). The defining features of this phase were political reforms and economic decline (Figure 2). Across the board budget cuts led to a severe contraction in education spending ( Figure A.3). To compound matters, UPE had led to a severe shortage of teachers, leading the government to hire scarcely trained primary school 26 See Figure D.9 summarizing party-state relations in the Appendix.
graduates as primary school teachers (Meena, 2009;Nzima, 2016). The quality of schooling deteriorated even further, forcing the government to appoint a Presidential Commission on Education in 1982 to review the sector and provide recommendations for reforms.
Other reforms were forced on the government by circumstances beyond their control. The economic crisis forced liberalization of the economy, which in turn set in motion movements towards a reorientation of the school curriculum away from socialism and towards capitalist development. The inability to fund education saw the re-introduction of school fees in secondary schools. Budget cuts eroded UPE's gains in enrollment. Increased demand for secondary education in the wake of UPE forced the government to acquiesce to the opening of ever more private secondary schools. 27 Overall, the new focus was on rationalization of spending and minimization of sectoral decline, as opposed to schooling for socialist and redistributive ends (Samoff, 1987).
Nyerere's resignation in 1985 crystallized the third political settlement. CCM's hold on on education policy and ability to suppress popular demand for access and quality waned.
Eventually, the party accepted multipartyism in 1992 ahead of the 1995 General Election.
These shifts were reflected in the moderate improvements in levels of democratization seen in Figure 2. On the economic front, Ujamaa policy missteps, droughts in the 1970s, the war with Uganda in 1979, and a global economic crisis jointly precipitated a severe recession. Consequently, the government was forced to implement structural adjustment programs which included budget cuts in the education sector -including the re-introduction of school fees. While Nyerere's successor, Ali Hassan Mwinyi, favored unwinding many Ujamaa policies, intra-CCM factional politics moderated the pace of reforms (Kjaer, 2004). In any event it was not until the late 1990s that the fiscal situation permitted the government to contemplate 27 For example, by 1974 only 11.4% of primary school graduates transitioned to secondary school -with 28.6% of them attending private schools. Just over a decade later in 1986 and following the implementation of UPE, the transition rate had dropped to 7.2% with 51% of students attending private secondary schools. The privatization of education exacerbated regional inequlities. For example, Cooksey (1986) notes that in 1986 "Kilimanjaro Region [had] only 11% of all government schools in the country, but no less than 34% of all registered private schools" (p. 185).
a new ambitious agenda of mass education.
This phase also saw a reorganization of the administrative structure of implementing education policy. The de-congestion policy of 1972 had proven unable to manage the provision of essential public services like education. As such, in 1982 the government re-introduced local government authorities with elected assemblies and appointed executives. In 1983, these units were charged with the implementation of basic education, including planning and budgeting, payment of teachers, and supervision of school management. This left the Ministry of Education in charge general education policy and higher education.

Mass Education for a Modern Economy
The present political settlement, which began with the presidency of Benjamin Mkapa in 1995, has been characterized by a departure from past ideas about school access and the role of education in society. Liberalization of the economy and the demise of Ujama forced curriculum reforms Symbolically, pro-CCM Political Studies was scrapped as a subject. While CCM remains a hegemonic party (Morse, 2014;Collord, 2021), since 1995 it has faced strong electoral incentives to increase access to schooling and signal commitment to improving learning outcomes. Importantly, the party is no longer at the core of education policymaking. The Quality became a concern, albeit one that was not readily salient given the yawning access gaps that existed. Since 2005 the government has signaled a willingness to invest in improving learning outcomes. However, such efforts have been moderated by the reality of fiscal-administrative capacity and the weight of historical under-investments. For example, at the height of the most recent expansion of access to secondary education (2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015), the government resorted to the same tactics of lowering requirements for teacher qualification -with disastrous consequences for quality (Habyarimana, Opalo and Schipper, 2021) Importantly, the CBC and BRN reforms emphasized learning, and defined it as a set of quantifiable outcomes. 30 Administratively, the management of basic education still reflects the reforms of 1983.
Administrative reforms since 1999 have deepened decentralization, giving more power over spending and policy to local governments and their assemblies. In order to improve school management, the government has embarked on direct funding through capitation grants.

Implications for Schooling and Learning
How did policy prescriptions under the different political settlements described above impact actual outcomes? I answer this question from three perspectives. First, I use survey data on reported levels of education attainment and literacy rates to construct trends in education attainment. Second, I show parallel trends from official government statistics. Both sets of data broadly align with the phases of political settlements described above. that concerns about quality of education is trending downwards, with respondents more concerned about access. Furthermore, concern about quality is not correlated with disaffection with government performance in the education sector, or the overall rating of the president, members of parliament, or members of the LGA assemblies.

Survey Data on Attainment and Literacy
In light of the four phases of political settlement described above, this section delves into actual education outcomes in Tanzania since 1961. I begin by analyzing trends in education attainment with data from seven rounds of Afrobarometer Surveys (N=13,119) and six rounds of Demographic and Health Surveys (N =55,121). 32 To describe the policy impacts of each phase of political settlement, I code the year in which each respondent graduated from primary school. During the period under study, the primary education extended to Standard Seven or Eight with pupils enrolling in Standard One at six or seven years of age. 33 For simplicity, I assume that pupils graduated primary school at the age of 14 years. The choice of primary graduation age is informed by Tanzania's historical rationing of secondary education, which presumably structured households' investments in education beyond primary school.
I then code if the respondent only graduated from primary school, secondary school, or attended any institutions of learning higher than secondary (i.e. any institution above primary school). Figure 4, with data from 7 rounds of the Afrobarometer Surveys, shows trends in the share of respondents who have graduated from primary and secondary school, or have attended any schooling beyond primary. Recall that this latter category includes those that attended secondary school. The trends indicate a clear increase in primary enrollment and completion throughout the 1950s, a brief deceleration of the expansion in the 1960s, 32 The Afrobarometer Surveys were conducted in 2001, 2005, 2008, and 2017. The DHS Surveys are from 1992, 1996, 1999, 2005, 2010 Standard 8 was abolished in 1966  and beyond. This is shown by increases in the proportion of respondents reporting having attended secondary school or higher after the year 2000. As expected, the trends also show a decline in the share of respondents who did not advance beyond primary school.
The evidence from the Afrobarometer Surveys is corroborated by data from the Demographic and Health Surveys (DHS), which only include female respondents. Figure 5 shows trends in in the share of (female) respondents who graduated from primary school (coded as those that attended up to Standard 7 in the data). First, the trends largely corroborate the Afrobarometer data. There is a increase in enrollment between the late 1960s and the late 1970s, followed by a stagnation through the mid 1990s. Second, the trends show that girls benefitted from the 2001 UPE policy than they did in the 1970s, with the share of respondents reporting having attended up to Standard 7 jumping significantly after 2000.
To evaluate the impact of schooling on learning, I rely on actual information on respondents' ability to read simple sentences -a marker of literacy -from the DSH data. The data are only available for female respondents. 34 I follow the same coding rule above, and place respondents to the year when they are supposed to have graduated primary school (age = 14 years). The assumption is that literacy upon graduation from primary school is critical for further education attainment. Figure 6 shows the trends in the share of respondents who are coded as being able to easily read a simple sentence in the DHS data. Notice that between 1961 though the 1970s the share of respondents able to easily read a simple sentence steadily increased with the gradual expansion of primary education. Then followed more than two decades of stagnation beginning with the UPE policies of the 1970s. For two decades, nearly 40 percent of pupils enrolled did not acquire simple reading skills. The gap between primary school attainment ( Figure 5) and the literacy rates in Figure 6 suggest either that even pupils who dropped out before completing primary school were able to acquire readings skills (more likely) or the respondents acquired reading skills after leaving school. 34 The similarities of trends in attainment between boys and girls in the Afrobarometer (See Figures A.2  and A.3 in the Appendix is indicative that these data are informative on boys' learning as well.

Official Statistics on Attainment
Having shown data from surveys on education attainment and literacy rates, this section shows the related trends in official education statistics. The data are from various issues of the Basic Education Statistics in Tanzania (BEST) and the Electronic Management Information System (EMIS). Figure 7 shows trends in access to primary and secondary education. Official statistics corroborate the survey data above. Primary school enrollment in Tanzania increased dramatically under the UPE policy in the late 1970s. This was followed by a stagnation in the total number of pupils in primary school for nearly two decades as a result of government disinvestment in the education sector due to economic crises. Notice that this was despite an ongoing increase in Tanzania's population. In other words, enrollment rates declined during this time period.  Notes: Graph shows the important role of private secondary schools for much of the period in which the government was effectively rationing access to secondary education. By 1990 nearly 60% of all pupils entering high school (O-Levels) attended private secondary schools.  Finally, it is worth asking if ESR achieved its objectives of boosting agrarian production and preventing uncontrolled (jobless) urbanization. Figure 8 shows trend in agricultural productivity and the growth in urbanization in Tanzania. Between 1961 and 2000, the per capita agricultural production remained steady or declined. Growth in urbanization was highest during the ESR decade in the 1970s. In other words, ESR does not appear to have worked as intended, which suggests that its longevity can be attributed to CCM's political and policy hegemony and the supporting political settlement.

Public Opinion on Education Performance
One of the implications of the conceptual framework highlighted in Section 2 is that education's influence on electoral outcomes is conditional on the political salience of the sector.
Furthermore, even when education is salient, vote choice is more likely to be driven by accessrelated and readily visible and attributable factors in the education sector such as school construction and the abolition of fees; and less so by factors like actual learning outcomes.
With this in mind, this section examines public opinion on education policy in Tanzania using data from Afrobarometer Surveys. In all the seven available surveys, the share of respondents deeming the government's performance in the education sector to be "very well or fairly well" was consistently more than 50 percent. This is despite the fact during this period the country saw a significant deterioration in the quality of learning as measured by pass rates in primary and secondary school examinations (Habyarimana, Opalo and Schipper, 2021). It also came on the back of a history of severe rationing of access to both primary and secondary education as outlined above. Given the overall positive perception of the government's performance in the education sector, how do Tanzanian voters balance the demands of access against concerns about the quality of education? The same Afrobarometer data allows me to interrogate this question. In Rounds 1 through 3, the survey asked respondents to explicitly consider the tradeoff between access and quality. Respondents were asked if they supported increasing access regardless of impact on quality, or if they were willing to pay fees in order to improve the quality of education provided by the government. Figures 11 and 12 summarize the underlying dynamics of the electoral politics of education in contemporary Tanzania. In the Round 1 Afrobarometer Survey, conducted when the government was just rolling out the new UPE policy, more than 80 percent of respondents expressed a preference for paying fees to improve school quality. Only about a fifth of respondents supported increasing access at all costs. Two years later, after some experience with the UPE policy, there appears to have been a significant shift in public opinion. In 2003, less than 60 percent were willing to pay fees to improve quality, while over 40 percent preferred increasing access even at the cost of compromising on quality. By 2005 the a majority of respondents in the survey favored increasing access, even at the expense of quality. A possible explanation of these shifts is that in the 2001 survey, respondents were still used to paying fess -a legacy of the fiscal reforms of the 1980s. However, upon realizing that UPE was about free access to primary education, many updated their preferences and started expecting the government to increase access to schooling, even if it meant compromising on quality. Notes: Figures show results from OLS estimates of approval of overall performance in the education sector and specific evaluations of the national government/president and local government authorities. A preference to pay fees for quality is not correlated with a negative evaluation fo the government's performance in the education sector or overall rating of different levels of government-despite the fact quality remains to be a major concern in Tanzania's education sector.
evaluation of the overall performance of the national government/president and LGAs. In other words, overall approval of the government's performance in the sector was consistently positive ( Figure 10) and, among those who are concerned about education quality and are willing to pay for it, such concerns were not correlated with negative ratings of government performance. Quality does not appear to have been a driver of perceptions of performance.
Overall, the survey results suggest that electoral incentives for politicians are biased in favor of access, even if it means at the cost of compromising on quality. This is particularly important because much of Tanzania remains plagued by inadequate access to schooling at all levels. As such, government investment in access will continue to be a top priority for the government of Tanzania. Whether or not the government also invests in improving learning outcomes is likely to depend not on electoral incentives, but on policymakers' normative commitment to quality education.

Conclusion
Learning is not always the goal of schooling. As policymakers tackle the global learning crisis, it is important to understand that the design of education systems is not always conducive to learning. Besides learning, states may view education as a means of cultivating loyal citizens, nation building, ideological indoctrination, or the narrow development of technical skills to meet the economy's labor demands. Furthermore, due to bureaucratic weakness and fiscal constraints, governments may not always be willing or able to provide access to schooling, or to invest in improving learning outcomes. Historical underinvestment in both access to schooling and learning means that many low-income states have to balance demands for greater access against improving learning outcomes. Under these circumstances, electoral incentives are decidedly in favor of investing in access, including at the expense of learning.
Using a political settlements approach, and with data from multiple sources, this paper documents shifts in Tanzania's education policy through four phases of political settlements.
Tanzanian policy has not always been supportive of universal access to education and/or improvements in learning outcomes. In the first 15 years of independence, the political settlement resulted in the rationing of access to both primary and secondary education. As part of its policy of manpower planning, the government aimed to train a prescribed finite number of graduates to provide the skills necessary to Africanize the managerial class in both the private and public sectors. In the early 1970s, a combination of political pressure and the need to orient the education system towards producing manpower for socialist agrarian development forced the government to adopt a universal primary education (UPE) policy.
Yet even as it did so it continued to ration access to secondary and tertiary education, preferring to have the private sector meet some of the demand at the margins. Primary schooling became a terminal tier of education, with the curriculum emphasizing functional literacy for participatory socialist citizenship and technical skills for improving agrarian productivity.
Transition rates from primary to secondary school flattened for 40 years.
Economic crises in 1980s forced the government to abandon the UPE policy and to disin- A consequence of Tanzania's historical rationing of schooling is that access to education remains far from universal. For example, despite enormous gains made since 2015, in 2020 gross enrollment in secondary school was 31.4 percent. Therefore, the government is currently grappling with the need to rapidly expand access, while also improving learning outcomes.
Under the circumstances, evidence suggests that the electoral incentives are biased towards investments in access, even if such expansions compromise quality.

A Educational Attainment
This section provides further evidence of the changes in education attainment in Tanzania since 1961 with data from the Demographic and Health Surveys (DHS) 1 and Afrobarometer. 2 The DHS data only include female respondents. However, the Afrobarometer data confirm that the trends among female respondents parallel those from male respondents. Both pieces of data support two general narratives: (1) increasing enrollment after independence and through the 1970s, with stagnation in the 1980s and 1990s; and (2) a persistent rationing of access to education beyond primary school -a policy position that only appeared to end in the early 2000s.       This section summarizes major education policy changes in Tanzania since 1961 matched to the four prevailing political settlements. Overall, Tanzania's policy evolution can be summarized as thus: • Initial pragmatism around manpower planning aimed at Africanizing both the state bureaucracy and commanding heights of the economy. This era was also characterized by effective rationing of education access due to fiscal constraints; subnational variation in education as a consequence of the administrative structure (local authorities were in charge of much of education policy implementation); and limited commitment to learning targeted at skilling the requisite manpower.
• The era of Education for Self Reliance was marked by complete subordination of education policy to the politico-ideological objectives of TANU. Cost considerations were abandoned. Education was targeted at molding loyal citizens that would support the policy objective of agrarian socialist development. To the extent that the government was concerned about (adult) literacy, it was for socialist developmentalist ends -as opposed to preparing students for education advancement. Importantly, this era was characterized by primary education as a terminal level, with access to secondary and higher education severely constrained.
• Economic crises in the 1980s ushered in the era of reforms and structural adjustment of education. During this period, the influence of CCM members on policy declined, budgetary concerns acquired urgency, fees were introduced, and the government cut back its spending on education. Consequently, much of the gains from the late 1970s UPE drive were eroded. Tanzania therefore entered the next phase with a chronically underfunded education sector that had little chance of improving learning outcomes.
• The last phase is one that is characterized by a government commitment to mass education and to improving learning outcomes.

C Administrative Structure and Education Policy
This section briefly describes the political and administrative structures through which the Tanzanian government implements education policy. The Education Ordinance (1962) gave empowered subnational units with the administrative and fiscal authority over educationwith the national government topping up localized funding. The law charged rural and urban authorities with implementing education policy. The government encouraged the formation of Local Education Authorities to manage this process. However, after the 1967 Arusha declaration, the government increased centralized administrative and policy functions. This left the subnational authorities with "more responsibility than power." 4 To remedy this problem, in 1972 the government embarked on a decentralization by de-concentration exercise aimed at making the bureaucracy more accountable to TANU and to increase participatory government at the subnational level. Elected local authorities were abolished. The decentralization process involved de-congestion, with central ministerial structures replicated at the subnational level with the aim of improving (participatory) service delivery. Subnational development committees replaced the previous system of local government. Notes: Graph shows the share of the budget allocated to education. Notice the decline in education share after the 1970s due to the bigger role of the Ministry of Regional Administration and Local Government in the provision of basic education in Tanzania.
In effect, the administrative arm of the government (regions and districts) became the implementing authorities of basic education policy in conjunction with Local Education Authorities. The national Ministry of Education was left to manage higher education and to give the sector general policy guidance -under heavy influence from TANU and its policy of Education for Self Reliance. The net effect of this administrative change was to give more power to TANU politicians and organizations and to limit the influence of technocrats in making education policy. Consequently, the ministry's budget saw a decline as a share of the total budget beginning in the late 1960s (Figure ??). Presently, the bulk of the education budget (up to 80 percent) gets channeled through the PO-RALG and Local Governments and Authorities. 5 PO-RALG remains principally concerned with increasing access to schooling in an equitable number -with emphasis on school construction and material support.  • Policy, curriculum, standards and examinations • Technical and professional inputs for the education service • Quality control and supervision • Analyze infrastructure and financial reports from Local Authorities Ministry of Regional Administration and Local Government (MRALG) • Funding of Local Authorities • Coordinate work done by Ministry of Education and Local Authorities • Determine the allocation of block grants to Local Authorities • Guidelines for planning, budgeting, management, and reporting of Local Authorities in consultation with MoE and Ministry of Finance

Regional Secretariat
• Technical support for Local Authorities on budgets and planning • Consolidated budget plan for all Local Authorities within region • Monitor and audit Local Authorities within region

Local Authorities
• Prepare and implement education plans that address local needs and priorities • Prepare annual education budgets • Manage account from own resources and from centrla government • Lead, regulate and co-operate with other actors including the private sector in the implementation of education plans The broad failure of the politicization of policymaking led to a recalibration in the early 1980s, with the government giving in to the need for technocratic policy reforms touching on the economy and multiple sectors. In 1982 the government re-introduced elected local governments, although their functions were severely constrained on account of ongoing eco-nomic crises. 7 Town and municipal councils had been re-established in 1978. It was not until the late 1990s that the government was able to reinvigorate Local Governments and Authorities (LGAs) under the Local Government Reform Program (LGRP) started in 1998. Under the program, the government sought to strengthen the fiscal, administrative, and policy functions of elected local councils. These authorities fall under the Ministry of Regional and Local Governments in the President's Office (PO-RALG/TAMISEMI), and have the primary responsibility of providing basic education in Tanzania. Table C.2 summarizes the division of labor among the Ministry of Education, PO-RALG, Local Government Authorities, and the regional administrative units.
Other government ministries and agencies involved in the education sector include the Ministry of Finance, President's Office -Public Sector Management and Good Governance (PO-PSMGG), Tanzania Institute of Education (TIE), the National Examinations Council of Tanzania (NECTA), the Tanzania Library Services (TLS), Tanzania Education Authority (TEA), the Institute of Adult Education (IAE), the Teacher Service Commission (TSC), Vocational Education Training Authority (VETA), and National Commission for Science and Technology (NCST). TheTanzania Teachers Union, Non-Governmental Organizations, Think Tanks, and Donors are also important stakeholders in the sector. 8 The activities of these different agencies are not always well-coordinated, resulting in unclear principal-agent relationships in the sector. Figure ?? highlights the difficulty of managing the sector under these circumstances with a reference to teacher management.

D Politics and Electoral Incentives
One of the core claims of this paper is that, for a long time, the government of Tanzania was able to effectively ration access to (quality) education. It was able to do so because of the regime type that existed in the country between independence and 1995. TANU/CCM's single party rule between 1965-1995 afforded the government the ability to impose policy prescription without having to face voters at the ballot. The abolition of local governments and local elections in favor of "participatory" socialism further reduced the sensitivity of education policy to politics at the local level. As shown in Figure ??, TANU/CCM was able to influence policy at both the national and subnational levels. 9 After 1964 a significant share of the civil service were members of TANU. While the government certainly faced political pressure through the period under study, it is fair to say that the party enjoyed considerable leeway in policy formulation and implementation -in the service of socialist developmentalism.
TANU/CCM's unchallenged policy dominance ended in 1995. Already in the early 1980s, several policy failures in education and other sectors had eroded the party's powers in favor fo technocratic bureaucrats. Nyerere's retirement in 1985 further dimmed the party's star. However, its organizational reach and enduring commitment to socialism under Ujamaa meant that Nyerere's successor, Ali Hassan Mwinyi, could not accelerate policy reforms. 1995 marked a watershed moment. The party, which had been losing power and influence over policy for over a decade, found itself facing multiparty competition and with it the need for greater competence. It is important to note that CCM is yet to face a serious electoral threat to its power. However, by merely holding multiparty elections -some of which are competitive -the party exposes its officials to pander to voters in ways that generate electoral pressures common in electoral democracies.
Notably, due to electoral pressure, the government then proceeded to accelerate the realization of universal basic education (primary and secondary) well ahead of schedule and with very limited resource. As late as 1995 the Minister of Education and Culture admitted that the government "was incapable of financing all school projects but would support any project initiated by parents and students." 10 Yet UPE became a reality only six years later (albeit with donor support). In the same vein, ahead of the 2005 election, CCM's manifesto pledged to increase secondary enrollment by 50 percent and A-Levels by 25 percent by 2010. The party also reiterated its commitment to the idea of at least one secondary school per ward. 11 No longer could the government ration access to basic education. Currently, government policy is to provide free basic education through lower secondary (O-Levels). The government is also committed to quality -both for instrumental reasons and with an eye on the electorate. In its 2004-2017 education report, the government notes that " [In] order to realize the vision of Tanzania becoming a Lower Middle Income Country (LMIC) by 2025, the education sector is a catalytic sector for building human resource skills and creating the mind-set of Tanzanians for economic development." 12 Perhaps the best illustration of the influence of electoral politics on education is the recent expansion of access to secondary education -ahead of official schedule. 13 After 2006, the government embarked on a massive school construction effort -building thousands of schools. It did this by leveraging CCM's extensive reach throughout Tanzanian society and the system of regional administration and local government. The increase in the number of schools was matched by student enrollment. All this happened even as learning outcomes deteriorated on account of the inability to absorb new students in an already stressed system. Electoral incentives favored providing access.