Social Contracts in Sub-Saharan Africa Concepts and Measurements

The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. that a social contract framing can help to diagnose and explain complex development challenges such as entrenched inequalities, binding constraints, poor service delivery, weak institutions, and why decades of policy and institutional reforms promoted by external development actors had uneven effects on countries’ development paths. This paper proposes three potential contributions of social contract diagnostics for development institutions: i) to offer an integrative perspective connecting the different elements of a country’s development spheres (political, social, and economic), ii) to help anticipate indirect effects from international development aid on the social contract which could lead to unforeseen consequences, and iii) most importantly, a social contract lens places the citizens front and center of the development effort and not only has beneficiaries but also as major stakeholders and actors with intrinsic agency.


Policy Research Working Paper 9788
In 2019, an Independent Evaluation Group review on the growing use of social contracts terminology by the World Bank concluded that social contract diagnostics are useful analytical innovations with relevant operational implications, particularly in situations of transition and social unrest. But it also found that the World Bank had no formal, conceptual framework or shared understanding of social contracts, leading to uneven quality of use. This paper proposes a framework and quantitative measures to describe social contracts. First, the paper presents a literature review on social contract theory and its applications in development. Second, it proposes a conceptual framework based on three core aspects of social contracts: (i) the citizen-state bargain, (ii) social outcomes that form the contents of the social contract, and (iii) resilience of the social contract in terms of how citizens' expectations are being met. Third, an empirical measurement strategy is described to quantify these aspects through six dimensions and 14 subdimensions using available indicators from multiple sources. An empirical analysis then successfully tests some of the framework's predictions and finds indicative evidence for an operationally interesting result: that state capacity without civil capacity is often not sufficient to generate thicker and more inclusive social contracts, and that these better outcomes lead to less misalignment with expectations and to less social unrest. Fourth, the quantitative measures are used to present three comparative maps for the general characterization of social contracts at the cross-country level.
This paper is a product of the Governance Global Practice. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The author may be contacted at mcloutier1@worldbank.org.

Introduction
Social contract theory has been central to the Western understanding of statehood and citizenship for many centuries. The applicability of the theory outside that context and within the development community is not straightforward. Following the wave of countries acceding to independence in the 1960s, a large body of Africanist scholarship addressed the challenges of state-building and governance in Africa, sometimes in the context of widespread political turbulence, civil conflict, military rule and state failure (Watts 2018). Recently and as part of a departure from the Washington Consensus, most international development institutions, including the World Bank, have started integrating more sociopolitical considerations into their engagement with client countries. One increasingly popular approach is to analyze the context of a country through a social contract lens. One reason why this approach is gaining traction is because it connects with three current debates in international development: the policy implementation gap, 2 the diagnostic of binding constraints to development, and resilience to fragility from conflict and social unrest.
It is undeniable that social contract terminology has been appearing more frequently in the World Bank's lexicon used in regional strategies and reports. Due to this trend, the Independent Evaluation Group (IEG) 3 published in 2019 a review of the utilization and usefulness of a social contract lens for the World Bank's engagement. The report identified 21 Systematic Country Diagnostics (SCDs) that "use a social contract framing to diagnose and explain complex development challenges such as entrenched inequalities, poor service delivery, weak institutions, and why decades of policy and institutional reforms promoted by external development actors could not fundamentally alter countries' development paths" (Independent Evaluation Group, 2019). The IEG evaluation concludes that social contract diagnostics are useful analytical innovations with relevant operational implications, especially in countries experiencing social unrest where social contracts are in transition, but that the World Bank has no formal conceptual framework or shared understanding of social contracts, leading to uneven uses.
The current paper was commissioned as a background paper for a regional report on social contracts in Sub-Saharan Africa. The regional report was requested by the World Bank's Chief Economist for Africa following two similar regional reports based on the social contract lens produced for the Middle East and North Africa (World Bank, 2015) and for the European and Central Asia (World Bank, 2019) regions. The main purpose of the paper is to propose a conceptual framework and an empirical strategy to understand and measure important aspects of social contracts. The secondary objectives are then to use this framework to classify social contracts according to proposed typologies, to see how different measures of the social contract correlate to the World Bank targets of stability, equity and prosperity, and finally to propose preliminary diagnostic tools for social contracts.
Laying down a conceptual framework that offers a clear definition of social contracts and the associated terminology is a necessary preliminary step to quantifying them. The framework is based on a definition from an Organization for Economic Co-operation and Development (OECD) report from 2008 which defines social contracts as: "dynamic agreements between state and society on their mutual roles and responsibilities". This definition has the advantage of being simple while still introducing three key aspects that are central to the framework. First, the definition is based on the idea of an agreement and therefore 3 on a form of implicit and explicit bargaining mechanism between the citizens and the state. Recently, a similar approach was used by the World Development Report (WDR) 2017 (World Bank 2017a) on Governance and the Law to account for outcomes from the political arena. This paper therefore builds on these ideas of settlements and bargains. Second, the definition mentions the roles played by the actors in determining the social outcomes that result from that agreement. The policies, programs, and laws represent the contents of the social contract and impact how resources and rents are allocated within society. Third, the definition also introduces the idea that social contracts assign responsibilities which leads to actors forming expectations on what they are entitled to receive. By mentioning dynamicity, the definition also suggests that social contracts are not static equilibria but that they exhibit pathdependency, feedback loops and self-reinforcing cycles. This means that all its elements are profoundly endogenous and there are no clear causality directions between any of them. Building on this definition and principles, the paper presents a coherent model and measurable dimensions to characterize the bargain between citizens and the state, the social outcomes, and the alignment between expectations and perceptions which impacts the evolution of the social contract.
In the framework, these three aspects of social contracts: i) citizen-state bargain, ii) social outcomes, iii) resilience and dynamicity, are incorporated through six measurable dimensions. Two dimensions directly influence the citizen-state bargain: State capacity and Civil capacity. 4 Three dimensions characterize the output of that bargain in terms of social outcomes: Thickness, Inclusiveness and Responsiveness. Finally, the Alignment dimension captures how the three social outcomes are aligned with the citizens' expectations. The paper describes an empirical methodology to quantify these dimensions, themselves composed of 14 sub-dimensions, using readily available indicators taken from multiple data sources. The main data sources used are the Varieties of Democracy Index (V-DEM), the Worldwide Governance Indicators and perceptions surveys such as the Afrobarometer.
The indicators are used to perform empirical tests of the assumptions of the framework and to present indicative evidence of a major implication of the model. First, the analysis shows that as predicted by the framework, both higher civil capacity and higher state capacity are strongly correlated with higher social outcomes. This result supports the core assumptions that the state-citizen bargain influences the social contract. Furthermore, the statistical model also shows strong correlation between better social outcomes and a higher alignment. A less intuitive prediction of the model is that perceptions on these outcomes should matter more for Alignment than the expert-based measures. This prediction is empirically supported by showing that the perceived level of outcomes from citizen surveys better predicts Alignment compared with more expert-based measures. This suggests that expectations and perceptions play a major role in determining citizens' satisfaction with the social contract. Taken together this evidence suggest that the framework is based on sound assumptions and has a robust level of explicative power.
The main empirical finding of the analysis is to present indicative evidence that state capacity without civil capacity is not sufficient to generate high levels for the social outcome indicators. This finding has important operational implications that are particularly relevant for international development 4 institutions who often focus most of their engagement on building state capacity through trainings and technical assistance and often invest less on building citizen capacity to hold states accountable.
The framework and its empirical measurements are then used to propose different typologies to classify social contracts in Sub-Saharan Africa and to introduce some preliminary diagnostic tools for social contracts. The first of these typologies, related to the resilience of the social contract, maps countries along the two dimensions of alignment and responsiveness. 5 This paper proposes that this typology could potentially help anticipate or explain social unrest based on whether resolving misalignments in the social contract is possible through peaceful and open dialogue. The assumption is that a social contract with misaligned outcomes and expectations could find a peaceful resolution to the issue if there is enough responsiveness from the state but that without it, social unrest and a breakdown of the social contract could be likely, due to the lack of alternative. A second typology presents countries along the dimensions of state capacity and civil capacity and is related to identifying potential binding constraints to development within social contracts. The third typology presents countries along the dimensions of thickness and inclusiveness and is related to the social outcomes of the social contract.
The IEG (2019) report mentions that a social contract framing can help to diagnose and explain complex development challenges such as entrenched inequalities, binding constraints, poor service delivery, weak institutions, and why decades of policy and institutional reforms promoted by external development actors had uneven effects on countries' development paths. This paper proposes three potential contributions of social contract diagnostics for development institutions: (i) to offer an integrative perspective connecting together the different elements of a country's development spheres (political, social, and economic), (ii) to help anticipate indirect effects from international development aid on the social contract which could lead to unforeseen consequences, 6 (iii) most importantly, a social contract lens places the citizens front and center of the development effort and not only has beneficiaries but also as major stakeholders in the country's fate as well as actors with agency.
The rest of the paper is organized as follows. The first section presents a literature review on social contract theory, its use in Sub-Saharan Africa and in development, and on past efforts to measure or quantify social contracts. The second section defines concepts related to social contracts, and describes the proposed framework connecting them together. The third section describes the empirical methodology used to measure the framework, the data sources, and caveats surrounding the measurements. The fourth section presents the findings from the empirical analysis using the indicators, including the tests of a few predictions of the framework and indicative evidence of the fact that given similar state capacity, countries with higher civil capacity are associated with thicker, more inclusive and more responsive social contracts. The fifth section describes the three typologies based on the framework and their interpretation. Finally, section 6 concludes by discussing ongoing applications of the framework and future lines of research. 5 Alignment measures the alignment between expectations and perceptions, while responsiveness measures how open to dialogue and diverging ideas the state is. The term alignment is used instead of more ambiguous and loaded terms such as legitimacy and trust. 6 An excellent ethnography illustrating such a case is "The Anti-Politics Machine" by James Ferguson which describes the unforeseen social impacts of the Thaba-Tseka Development Project in Lesotho from 1975Lesotho from -1984. The project misunderstood the social function and value of cattle which led to the project having a negative impact on the people it was trying to help.

A brief history of social contract theory
Social contract theory can trace its origins as far back as Plato's Republic which described a social arrangement for what the philosopher saw as a perfect society. The field has also been a staple of Euro-American political theory since the seventeenth century with philosophers such as Hobbes, Locke and Rousseau discussing the role and purpose of the state, and of the ruler. Watts (2018) offers an exhaustive literature review on the topic 7 and suggests that the literature on social contracts can be described along two main dimensions, between the classical and the modern, and between the contractualist and the contractarian approach. This section will also cover recent social contract theory application in many areas such as development and peacebuilding.
In the classical tradition of social contract theory, covering mostly philosophers and social scientists from before the 20 th century, social contract thinkers were mostly interested in the legitimacy of the state and on limits on the ruler's power. A good definition that fits that category is suggested by Lessnof (1990): "A social contract theory is a theory in which a contract is used to justify and/or set limits to political authority, or in other words in which political obligation is analyzed as a contractual obligation". For the classical theorists such as Hobbes, Locke and Rousseau, the social contract is an agreement whereby individuals who, absent a state, have natural rights to the means of their preservation, consent to lay down those rights and subject themselves to the coercive power of the state subject to everyone making a similar commitment. According to Watts's 2018 review, social contract theory thus aims to show -either normatively or as a heuristic -that members of some society have reason to endorse and comply with the fundamental social rules, laws, institutions, and/or principles of that society while also identifying limits to the ruler's power.
In the 20 th century, a rediscovery of social contract theory occurred, but from a more practical and applied perspective. For modern social contract theorists, the existence of the state is often taken as a given and the focus is placed on identifying the social institutions and policies that reflect justice as a foundational virtue of a society or that try to optimize some function of the state. Among some modern versions of contract theory like that of John Rawls (1971) for example, the focus is not on political obligation and the legitimacy of political authority but on the justice of social systems and social structures in what he calls "sufficiently developed societies". Social justice requires that the basic structure of society conforms to "principles of justice" which ensure "the proper distribution of benefits and burdens of social cooperation" and the proper assignment of basic rights and duties… as a thought experiment". Another development of modern social contract theory is the expansion or broadening of what civil society means such as in T.H. Marshall's (1963) essay on citizenship which saw this phenomenon in evolutionary terms. According to Marshall, civil society passed through a sequence of three stages. The first stage, civil citizenship, established the rights necessary for individual freedom: rights to property and personal liberty, and especially the right to justice. The second stage, political citizenship encompassed the right to participate in the exercise of political power through a variety of democratic institutions and mechanisms (for example holding office or by voting). The third and final stage, social citizenship, not only encompassed rights to a modicum of economic security, but also entailed a more far-reaching right, 6 namely "a share in the full social heritage, and to live the life of a civilized being according to the standards prevailing in the society." In more recent academic literature related to social contracts, the central role of civil society and the distributive function of the state are consolidated. In Acemoglu, Johnson, Robinson (2005) the focus is on the degrees of predation and inclusion/exclusion of political and economic institutions. The different combinations of political and economic institutions lead to more or less stable outcomes and on different growth paths. Another example of this more institutional angle on social contracts, North et al. (2009) see the primary challenge as violence and the existence of powerful groups who "actively use or threaten to use violence to gather wealth and resources." According to North and his co-authors, for economic development to occur, violence must be restrained through the establishment of a monopoly on violence. They argue that developing societies limit violence through the manipulation of economic interests by the political system, in order to create rents so that powerful groups and individuals find it in their interest to refrain from using violence, so-called limited access orders (LAOs), which are social arrangements that discourage the use of violence by organizations.
North and his co-authors are also closely related to the concepts of roving and stationary bandits introduced by Olson (2000) which explains the origin of the state through extractive versus productive social contracts. In Olson's description of the origin of social contracts, the initial anarchy, involves uncoordinated competitive theft by "roving bandits" which destroys the incentive to invest and to produce, leaving little for either the population or the bandits. Both can be better off if one of the bandits settles as a dictator, a "stationary bandit" who monopolizes and rationalizes theft in the form of taxes. A secure autocrat has an encompassing interest in his domain that leads him to provide a peaceful order and other public goods that increase productivity. At its most basic, a social contract is thus traditionally conceptualized as the exchange of taxation from the citizens for security provided by the state. This minimalist view of the social contract is clearly illustrated by Mancur Olson's description of the roving vs stationary bandits. Social contracts based on an exchange of taxation for protection are sometimes referred to as fiscal contracts such as in Persson (2009, 2011) who considered the power to tax (fiscal capacity) and the power to enforce property rights and regulate a market economy (legal capacity). However, modern social contracts often involve more elements than this simplistic exchange, such as the provision of public goods and the protection of freedoms and capabilities outlined in Amartya Sen's (1993) work and in recent work by Besley (2020) which takes an approach based on civic culture and on intrinsic reciprocity, where the forces that shape reciprocal behavior are internalized in preferences.
In addition to classical vs modern, a second typology of social contract theory distinguishes between the contractualists and the contractarians. These distinct perspectives of the notion of social contract, described in Weale (2013), rest on the conditions under which negotiation over the social contract takes place: namely mutual advantage versus impartiality. The contractarian approach is based on liberal principles and a self-interest view. It assumes negotiation to involve the promotion of individual selfinterest (which might include family, kith and kin) over a baseline (a "state of nature") of non-cooperation, all predicated on a principle of mutual advantage. One example of such an approach is therefore the claim that justice rests upon a principle of reciprocity among members of society. The motive for agreement has to be the need to secure personal interests rather than the broader public interest which is why the forms of social co-operation over the baseline of original non-agreement has to be advantageous for all. Contractarianism stems from the Hobbesian line of thought holding persons as self-interested and that a rational assessment of the best strategy for attaining the maximization of self-interest will lead them to act morally and to consent to state authority. Some of the modern contributors to that approach are Friedrich Von Hayek, Robert Nozick and James Buchanan.
In the contractualist or social rights-based approach, negotiation is conceived of as an exchange of reasons that are to be considered impartial from the point of view of the participants in the contract. The goal is locating rules for social organization that can be justified to all participants in the compact. Rather than a state of nature (non-agreement), the contractualist argues that justifiable rules can be reasonably accepted by all members of society. Viewpoints are represented and arguments are proposed for joint acceptance rather than claims being made over a social surplus. Motives of contracting parties reflect the desire to justify their actions to one another rather than rules to protect interests or advance their good. If contractarians are said to be driven by the rational pursuit of personal interest, contractualists see agents as concerned with reasonable agreement on principles of justice that are capable of governing society. Contractualism is based on Kantian principles which hold that rationality requires the respect of persons which in turn requires that moral principles be such that they can be justified to each person (i.e. individual motivation is a commitment to public reason justifying moral standards). Over the last half century, contractualism in its various forms has seen a new lease of life, in part triggered by the work of John Rawls and Brian Barry in normative political theory and in part through the rise of the "new contractualism" (contractualist doctrines of governance) associated with neoliberalism and the refiguring of state welfare around conditional benefits, mandatory obligations in return for state benefits.
According to Watts (2018), the first direct and serious engagements with social contracts in development theory emerged in the 1990s and early 2000s from four very different directions. A first direction emerged from the Middle East in the work of John Waterbury (1997) charting, as a result of the collapse of oil prices and structural adjustment programs of the 1980s, the delayed shift from what he called "social contracts to extraction contracts". As rents and finances diminished, Middle East and North Africa (MENA) states turned to capital and citizens to in effect refigure the material basis of the social contract. A second direction of engagement emerged around social welfare politics in Latin America and East Asia (see Haggard and Birdsall 2002 and Huber and Niedzwiecki 2015 for a review) examining the changing forms of social protection in "incipient welfare states". The third engagement led to the socialist eastern bloc as scholars attempted to explain why during the 1980s the Soviet Union and the Eastern Bloc states experienced social stability despite declines in the use of repression. An answer suggested was in the 'implicit social contract' that citizens should remain quiescent to the degree that the party-state provided secure jobs and services, subsidized housing and (controlled) consumer goods (Cook, 2007). A fourth and final thread emerged from Africa and pertained to food security. Here Alex de Waal's (1996) research addressed what he called anti-famine contracts which was in large measure a response to Amartya Sen's important work on starvation and hunger. De Waal emphasized the complex forms of inclusion and exclusion associated with these contracts and the fact that donors ("the humanitarian international") sometimes circumvented the anti-famine contracts and state authority.
Another arena in which social contract language has appeared is post-conflict reconstruction and peacebuilding in which OECD, NORAD and UNDP have been leaders (for a review see UNDP 2014 or NOREF 2016). As a new UNDP (McCandless, 2018) report on fragile and conflicted states notes: "A resilient national social contract is a dynamic agreement between state and society, including different groups in society, on how to live together, how power is exercised and how resources are distributed. It allows for the peaceful mediation of conflicting interests and different expectations and understandings of rights and responsibilities (including with nested and/or overlapping social contracts that may transcend the state) over time, and in response to contextual factors (including shocks, stressors and threats), through varied mechanisms, institutions and processes". As suggested by this strand of the literature, the rise of non-state armed groups is almost by definition a marker of a fractured or contested social contract which in some cases leads to various forms of formally negotiated peace (involving UN peacekeeping forces, finance and administrative personnel).
From the international institution literature also comes a commonly used definition of social contracts.
"The social contract emerges from the interaction between a) expectations that a given society has of a given state; b) state capacity to provide services, including security, and to secure revenue from its population and territory to provide these services (in part a function of economic resources); and c) élite will to direct state resources and capacity to fulfil social expectations. It is crucially mediated by d) the existence of political processes through which the bargain between state and society is struck, reinforced and institutionalized. Finally, e) legitimacy plays a complex additional role in shaping expectations and facilitating political process. Legitimacy is also produced and replenished -or, conversely, eroded -by the interaction among the other four factors... Taken together, the interaction among these factors forms a dynamic agreement between state and society on their mutual roles and responsibilities -a social contract." (OECD 2008, Concepts and Dilemmas of State-building in Fragile Situations: From Fragility to Resilience, p.17, emphases added.) The definition proposed by the OECD (2008) captures many important and actionable aspects of social contract and therefore will be used as the starting point for the conceptual framework developed n this paper.

Social contract theory in Sub-Saharan Africa
Watts's (2018) review mentions that over the last four or five decades a complex and sophisticated body of scholarship from Africa has addressed the challenges of state-building and governance, sometimes against the backdrop of widespread political turbulence, civil conflict, military rule and in some instances, something approaching state collapse. A key starting point of that literature (and of the present paper) is a recognition of the heterogeneity of African political settlements and uneven development trajectories. This rejects the notion of a monochrome continent marked by systemic state deficits, policy failure, and poor governance in favor of a far more nuanced understanding of the unevenness of national state capacities (asymmetrical state capabilities which vary across sector, scale of government and over time), the variable relations between political settlements and accumulation, and the varied forms of civil society engagement, mobilizations (and de-mobilization) associated with gradual and uneven regime shifts toward democratic governance. In this context, many of the assumptions that underlie the modern approach to social contract theory, such as an emphasis on the dichotomy of state and citizens, or even taking the existence of a state for granted, do not translate seamlessly to the study of Sub-Saharan Africa. On one side, the historical context of colonial legacies (pre-colonial empires, settler states, peasant or resources-fueled economies, forms of direct and indirect rule), and the paths to Independence (liberation movements, negotiated settlement) shaped the African state. On the other side, the theory was not adapted for some important aspects of the Sub-Saharan context such as emerging, fragile or even failed states, and the juxtaposition of customary leaders serving as relays between state and citizens.
A major branch of the literature related to social contracts coming out of Africa is centered around the concept of neopatrimonialism. Patrimonialism is a form of organization which pervades rational-legal political and administrative systems characterized by "officials hold [ing] in bureaucratic organizations with powers formally defined but exercise those powers…as a form of private property" (Clapham 1985). When this phenomenon is substituted for state and institution building, the result is linked to numerous public sector failures: limited tax revenue, the privatization of public office, market distortions and short terminvestment horizons. Another branch of research could be defined as historical and world systemic. Mahmood Mamdani's important book Citizen and Subject (1996) covers different sorts of nationalist movements and the ways in which culture, broadly construed, shaped what he called "decentralized despotism" in the postcolonial period, spoke to a number of key concerns: the continued role of so-called customary law and customary institutions and chiefly powers, the tensions between 'indigenes' and 'settlers' (new comers) in multi-ethnic states, the incomplete character of citizenship, and the entrenched tensions and separation politically between town and country. These historical approaches powerfully established the path-dependencies of forms of colonialism, the co-existence of multiple and overlapping forms of authority in Africa (chieftainship, local government, religious authority), and the enduring significance of culture (ethnicity and religion) in understanding political change.
Another body of work related to social contract in Africa comes from Paul Nugent's work (2010). The large scope of his work aims to address "what is actually happening" within the vast diversity of political networks of African states and how "institutions are molded to deal with everyday realities". Nugent (2010) adopts a very broad definition of social contracts understood as the "sets of understandings that underpin the relations between rulers and ruled". For Nugent, social contracts are the product and expression of social and political bargaining, which are more or less contentious, more or less inclusive, more or less robust, and more or less democratic and open. All social contracts involve an ordering of power, and differing orderings of powers have direct consequences for forms of authority, public capabilities and developmental outcomes. He distinguishes between three kinds of contract covering the long arc from the colonial period to the present -coercive, productive and permissive. Coercive is a contract predicated on the capacity of rulers to "render intolerable the lives of their subjects" through de facto forms of protection racket in which personalized (rather than routinized) predation is traded against popular protection ("being spared") from predatory acts. Productive contracts entail sovereign authority negotiating over social interests -a classical almost Hobbesian contract -which typically involves negotiation over "payment of taxes as well as rights of access to…. land". And finally, permissive contracts -widespread and almost endemic -which is a "halfway house" in which governing authority claims sovereign rights but does not fully exercise them in return for a "measure of de facto compliance".

Utilization of social contract terminology by the World Bank
In the 1980s and early 1990s, the Washington Consensus approach to development involved many reforms (liberalizations, privatizations and austerity) aimed at reducing or limiting the size and scope of the state, particularly in the economy. These reforms were later criticized for their shortcomings which spurred a re-thinking of the approach to state building and development. This led to the increasing realization that many intangible factors have an impact on the success of development programs, and the rise of "inclusive liberalism" (Craig and Porter 2006) and the new mantras of good governance, accountability and state capacities. The World Bank's analytical work presents many examples of this shift in development philosophy. In the World Development Report 2011 Conflict, Security and Development post-conflict remediation addresses the need to construct "inclusive-enough coalitions" and to engage with informal institutions and patronage networks, in the attempt to acquire "broader societal legitimacy" and "proactive communication … to build public understanding and support". Central to the report's model is the grounding of post-conflict transitions in fairness and the "political inclusion of all citizens". The World Development Report 2017 directly addresses the importance of the law and the power bargains that surround policies and programs. The report argues for the need for the World Bank to integrate an analysis of these intangible factors when designing projects. These reports, while not framed in terms of social contract, contain all the components of social contract theory (legitimation, agreement and negotiation, pact arrangements and so on).
Recently, the World Bank has also started engaging directly and in an increasing manner with social contract theory. In the exposition of its twin goal strategy, the World Bank calls for "social contracts within each country demanding that the poor be a priority in the policy environment that supports the growth process" (Independent Evaluation Group, 2019). Recent examples show the growing use of the social contract concept in the World Bank's engagement such as the 2015 MENA strategy and the 2018 Europe and Central Asia (ECA) Region Flagship report Towards New Social Contracts. The 2015 MENA strategy represents one of the World Bank's most ambitious attempt to translate social contract diagnostics into country operations. According to the Independent Evaluation Group (IEG), anchoring the MENA strategy around social contract renewal created new policy reform opportunities and improved the World Bank portfolio's coherence and targeting of lagging regions. In the 2018 ECA Region flagship report Towards New Social Contracts it is explicitly stated that "the report put an economic interpretation on the concept [of social contract]" and defines it as the "individuals' agreement for the broad outline of economic policies if the outcomes of these policies coincide with their preferences". Redefining social contracts as an economic equilibrium helped justify the World Bank's foray into social contract renewal and its recommendations for what social contracts should look like. However, stripping social contracts of their socio-political dimensions could be seen as problematic because of its reductivism (IEG 2019).
Following this recent use of the terminology of social contract, an IEG review on its usefulness was requested and was published in 2019. It concludes that: "Social contract diagnostics are useful analytical innovations with relevant operational implications, especially in countries where social contracts are in transition." The report also finds that at the country level, 21 Systematic Country Diagnostics (SCDs) use a social contract framing to diagnose and explain complex development challenges. However, the World Bank currently has no formal conceptual framework nor a shared understanding of social contracts, leading to a lot of variation in the value of its use and to uneven uses. For example, in many cases reports and diagnostics find that social contracts are "broken", and they recommend their "renewal" or "rebuilding" but without offering a satisfying definition of what it means for the social contract to be broken or on how it can be rebuilt. Within the 21 SCDs examined by the IEG team, there is uneven rigor in how "social contract" is used, leading to ambiguity over how social contract diagnostics can guide actions. Despite adopting a social contracts approach in their diagnostic, 45 percent of the SCDs sampled did not provide a definition of the concept. Another 23 percent employed a state-focused definition that exclusively emphasized the role of the public sector to deliver public goods and services, provide social protection (social safety nets and subsidies) and create an environment conducive for growth. In contrast, 9 percent had a citizen-focused definition that emphasized greater citizen engagement as central to increasing long-term growth, as well as economic and social inclusion for peace and social stability. Finally, less than a quarter provided a more nuanced definition that combined -to some extent -the roles (and rights) of, and relationships between, the state, citizens and other stakeholder groups.

Previous effort to quantify the social contract
In all the literature already mentioned, the efforts to measure quantitatively social contracts have been rare due to their multi-dimensionality and impalpable nature. One previous effort to do something similar to what this paper attempts to do is Hoogeveen (2018) which argues theoretically that the quality of a country's social contract and its statistical capacity are closely and positively correlated, and that this capacity behaves in a way that would be expected for a social contract indicator. The paper suggests that using statistical capacity as a proxy for social contracts offers a way to introduce them as variables in empirical studies. The author finds that a higher Social Contract Index (SCI) is associated with improved service provision (social services, economic services as well as civil rights and press freedom) but finds no association with income, poverty or inequality. Many other papers have measured different aspects related to social contracts even if they did not try to find indicators for social contracts as a whole. Some of the most common aspects measured are related to state capacity, fiscal contracts (such as in the recent Innovations in Tax Compliance paper from Prichard et al., 2019), and service delivery. For example, Milante and Woolcock (2019) argue that fiscal policy is a central component of the 'long route of accountability' binding citizens and the state. However, they admit that in fragile states the political dynamics shaping the extent to which this 'route' does in fact deliver incrementally better key services (such as security and health) to citizens -and for which citizens, in turn, give the state due credit -are highly fraught. Therefore, they suggest that such measures need to be closely accompanied by solid theory, experience and context specific knowledge.

Section 2: Conceptual Framework
The main purpose of the present paper is to propose a social contract framework that is amenable to empirical measurement. An obstacle to achieving this objective is that the social contract itself is not observable or explicitly described anywhere. The usual approaches to circumvent this difficulty involve finding a proxy for the quality of the overall compact as done in Hoogeveen (2018) or to focus on measuring specific dimensions of the social contract such as inequality, prosperity or stability. The approach taken in this paper tries to combine both these methods. As in the latter approach, the framework identifies specific dimensions of the contract to be measured and classifies them as being either inputs of the social contract bargain or as being outputs such as the social outcomes. Then, the framework offers a theoretical lens to interpret these individual dimensions in a way that informs on the overall compact such as in the former method. It is this integrative and systematic view that enables the claim of measuring tangibly the social contract. However, one does not need to agree with the theory behind the framework to still find the measurements of the individual dimensions useful by themselves.
Following Hoogeveen (2018), the framework is based on the OECD definition of social contracts. More specifically, the framework makes use of the definition of a social contract as "a dynamic agreement between state and society on their mutual roles and responsibilities". Despite its apparent simplicity, this definition has the advantage of introducing in a few many core aspects of social contracts covered in the framework. First, by emphasizing that social contracts are agreements, although often implicit ones, the definition also introduces the idea that some form of bargaining is involved and therefore that the parties to the contract have bargaining positions and bargaining powers. Second, by mentioning that this agreement defines the roles played by the parties, the definition suggest that the social contract impacts their involvement and opportunities in the economic, social and political life of the country. Third, by introducing the notion of responsibility, the definition indicates that social contracts also inform both parties on what they can expect from each other and what level of outcome would be perceived to be satisfactory. Finally, by recognizing that the agreement is dynamic, the definition emphasizes that these elements are endogenous and path-dependent, but also that a framework trying to understand social contracts needs to consider how the elements evolve over time. This aspect also suggests the existence of self-reinforcing cycles and feedback loops between all the dimensions.
Approaching the social contract from a citizen-state bargain perspective is useful for many reasons. First, it directly connects with many of the ideas that were suggested in the World Development Report 2017 that can be adapted and extended. In the World Development Report 2017, the distribution of power, meaning who holds power and the power asymmetries among different actors and agents, is a key element of the ways in which the policy arena (both the rules of the game and the outcomes of the game) actually functions and what quality of public goods and services are delivered. Second, it suggests a straightforward way of explaining some of the observed variation in outcomes across different contexts. Third, it gives the framework some predictive power since changes in the bargaining powers should naturally lead to changes in the bargain. These assumptions that underlie the framework can be tested empirically to provide support for the robustness of the model.
In the framework, the first dimension measured is the bargaining power of the citizens. It is referred to as civil capacity and represents the capacity of citizens to resolve the collective action problem and to present a united front to make demands to the state and to hold it accountable to its commitments. The civil capacity dimension is based on three sub-dimensions: i) citizen mobilization: how engaged citizens are on public forums and how involved in politics they are, ii) citizen organization: how well organized civil society is, for example in terms of number of CSO and their membership, and iii) citizen cooperation: absence of fractionalization or group grievances among citizens. Despite civil capacity being a relatively new term, there is a large literature covering related concepts and their role in the social order of societies. For example, social capital, which was introduced by Alexis de Tocqueville in the 19 th century and popularized by Putnam in his two important books Making Democracy Works (Putnam et al., 1993) and Bowling Alone (Putnam, 2000). In the latter, Putnam takes the presence of non profit organizations as the main indicator of social capital to argue that it has been decreasing in the United States for some time and how this phenomenon weakens democracy.
Whereas social capital emphasizes the effective functioning of social groups through interpersonal relationships, a shared sense of identity, a shared understanding, shared norms, shared values, trust, cooperation, and reciprocity, the concept of civil capacity focuses more on how effectively this social capital can be leveraged in the citizens' relation with the state. Such an interpretation has been made previously by Guiso et al (2016). The authors studied whether a historical shock could have persistent effects on civic capital, which they measured as the number of nonprofit organizations per capita (following Putnam), to explain current differences in per capita income between Italian cities. Easterly et al (2006) also tested a similar hypothesis in their paper which present evidence that measures of ''social cohesion,'' such as income inequality and ethnic fractionalization, endogenously determine institutional quality, which in turn causally determines growth.
The second dimension measured in the framework is state capacity which refers to the deployment of the state's reach over the national territory. The indicator specifically measures the state's capacity in three domains: i) deployment of state authority: capacity to enforce the monopoly of violence and laws over the whole territory, ii) resources mobilization and taxation, and iii) the efficiency of its bureaucracy to spend these resources efficiently. These three sub-indicators mirror the three dimensions of state capacity outlined in Hanson and Sigman (2013): extractive capacity, coercive capacity, and administrative capacity. State capacity is an important target for the World Bank which has numerous projects and programs that aim to build it through technical assistance, trainings and management system development. On the academic side, the administrative or bureaucratic capacity is a growing topic in the development literature, including at the World Bank's Bureaucracy Lab. On this topic, an important paper is from Rasul and Rogger (2016), which shows using data from the Nigerian Civil Service that management practices matter for the bureaucrats and their capacity to deliver services. Another paper on administrative capacity is by Khemani (2019). It emphasizes that the crux of state capacity is the culture of bureaucracies-the incentives, beliefs and expectations, or norms, shared among state personnel about how others are behaving. Khemani's paper then describes a principal-agent model to illustrate this observation and offers insights for what reform leaders can do to strengthen state capacity for public goods.
The second group of dimensions measured by the framework describe social outcomes that result from the roles agreed upon through the citizen-state bargain. There is a multitude of potential social outcomes that could be measured and for manageability the framework only focuses on three categories referred to as: thickness, inclusiveness and responsiveness. The Thickness dimension measures the involvement of the state in providing public services, public goods and redistribution of wealth (social welfare, safety nets, etc.). It is composed of two sub-indicators: i) public services: measured as quality of public education and public health, and ii) social safety nets: measured as the amount of social transfers to the poor. This indicator can also be interpreted as being on a thin to thick spectrum where a thin social contract would be one of economic laissez-faire and minimal involvement of the state in the economy while a thick social contract would be characterized by a more involved state. This aspect is related to the size of the government in the economy for example as measured by public expenditures vs GDP but nuanced by where these expenditures are located, for example in health or education vs in the military or the corporate sectors.
The second outcome dimension is inclusiveness which measures whether the social contract is geared toward benefitting the broad population or a select few. It is composed of two sub-indicators: i) equal opportunity: measured as impartiality of the courts and the respect of economic rights, and ii) rule of law: measured as the absence of abuse of power or corruption. This outcome can also be expressed as a spectrum, going from an extractive social contract to an inclusive one. An inclusive social contract would involve fairness of opportunity and an important level of socio-economic mobility for the general population. The ways a social contract could be extractive is through corruption, clientelism and barriers to entry (in business or in politics) or through social exclusion of certain minorities. An extractive social contract would exhibit the kind of rent extraction and extractive institutions described in the work of Acemoglu and Robinson, including the one previously mentioned, Acemoglu, Johnson and Robinson (2005).
The third outcome dimension is responsiveness which measures how open the state is to diverging opinions and its use of repression to silence them. It is composed of two sub-indicators: i) human rights: measured as the absence of political violence and torture as methods of political repression, and ii) freedom of expression: measured as the respect of the citizens' freedom of expression and the absence of media censure. This outcome can be interpreted as being on a spectrum from a repressive social contract to a responsive one. A responsive social contract features freedom of the press and of expression, while a repressive social contract features censure, political killings and torture. Responsiveness also introduces the different mechanisms, interfaces, and intermediaries that are involved in the bargaining and implementation of the final contract. An important literature related to this dimension is the choice of political system and the roles of election. For example, Acemoglu et al.
(2019) discuss whether democracy is linked to different outcomes such as growth. These considerations are one of the possible forms that the citizen-state bargain can take and therefore are deeply connected to the social contract. However, a deep dive into all the specific bargaining mechanisms is beyond the scope of this paper and could be included in future work or an extension of the framework. When describing these three social outcomes, the framework takes into account the different information contained in two common types of data: self-report or perceptions surveys (such as the Afrobarometer) and expert-based assessments (such as V-DEM or the Worldwide Governance Indicators) . While the two approaches attempt to measure the same thing, the quality of education services for example, they both contain certain unavoidable measurement biases. It is generally recognized that one of these biases that affects perception surveys is related to the frames of reference or the subjective expectations of the surveyed population. In other words, two different populations can sometimes perceive an equivalent outcome differently. For example, an expert's assessment of the primary education sector could give the exact same score to two different countries, but the population of country X could perceive that level of education services to be satisfactory while the population of country Y perceives it as being disappointing. This is because expectations for public services differ from one population to another. These expectations are based on multiple factors, including the citizens' perception of their state's capacity, their civil capacity, the country's history, culture, etc. Based on this observation, the framework presented in this paper proposes that comparing expert assessments with perception surveys can provide some information on citizens expectations and for this reason whenever possible the two are presented side by side.
Expectations are especially important in the context of the third aspect of social contracts which is concerned with the sustainability and resilience of the social contract. The final dimension measured by the framework for the resilience aspect is alignment which measures citizens satisfaction with the social outcomes of the social contract. This aspect is measured through two indirect indicators: i) civil compliance: measured by the absence of civil protests, civil disobedience, and other acts of citizen discontentment such as demonstrations and resistance, ii) popular support: taken from surveys on citizen's support and confidence in their government and in their politicians which is also sometimes related to notions of trust and legitimacy. Trust and legitimacy are terms that are often used in the literature and they often take different meaning depending on the author. For example, in Fukuyama (1995), the author notes that trust can have two dimensions: trust in the state and trust in the government of the day. Trust in the state refers to the extent to which citizens believe that 'the state' (in this case the executive branch of the state) has the expertise, technical knowledge, capacity, and impartiality to make good judgements. That it is has an adequate number of people with the right training and skills to carry out the tasks they are assigned, and to do them in a timely and professional manner. While trust in the government of the day is more about trust in the politicians that form the government. The alignment dimension can also be expressed as a spectrum from a misaligned social contract to an aligned one.
Citizens' expectations of what an aligned social contract is changes from country to country, depending on history, culture, preferences, and countless other factors. One of the propositions of this paper, which is substantiated by the empirical analysis presented in the next section, is that citizens' expectations are also linked to their perceptions of the state's capacity and their own civil capacity. The logic behind this claim is that citizens are probably aware of the level of service delivery they can expect given the state's capacity or of the protection of their rights they can expect given their civil capacity to hold the state accountable. When applied to the alignment dimension this means that one could imagine citizens in a country with little civil capacity or state capacity expecting little from the state which could, for example, make a thin, extractive and repressive social contract be perceived as well aligned.
This final aspect included in the framework and introduced by the OECD definition also relates to the dynamic nature of social contracts. Foremost, this means that all the elements in the framework are profoundly endogenous and interdependent. The state and civil capacities at time T influence the outcomes at time T, but they are themselves a product of the outcomes of time T-1. An example to illustrate this endogeneity could be to imagine an increase in the thickness of a social contract through an expansion of the delivery of public education. This could then lead to a more educated population composed of citizen that can then better organize and collaborate, increasing their civil capacity over time and with it increasing their capacity to demand even more education services from the state in a selfreinforcing cycle of incremental changes.
Alternatively, history is full of examples of revolutions and critical juncture moments when social contracts seem to be renegotiated drastically within the span of a few years or when a complete breakdown leads to many years of social unrest and civil war. In the framework, this is captured by presenting two possible paths for the evolution through time of the social contract. First, when the level of alignment is above a certain threshold, meaning that citizens' perceptions of the social outcomes are relatively in line with their expectation, the evolution of the agreement is incremental. This would be the case of the education example given above. Second, when the alignment level has been decreasing for a certain time or when a sudden change in one of the previous aspects generates a large shock in alignment, the social contract can potentially break down, meaning that it cannot keep going on the same path. In this situation, renegotiation is possible but only if there is an open dialogue. If that is not the case, conflict and unrest could be unavoidable. This second case, presented in figure 1 below, has an end to the cycle of bargaining and outcomes. This is only to illustrate that breakdowns of the social contract are often associated with prolonged periods of civil strife and political blockage. They often end with a new social order and discontinuity in the evolution of the factors of the bargain following a peace process, a change of regime or external intervention.
The dynamic aspect of social contracts also indicates that they exhibit high path dependency. This implies that the period T state-citizen bargain is far from being the only factor defining the social contract at time T. It is also influenced by the historical context of the country, the preferences of its citizens and many other factors that cannot be fully captured by a quantifiable framework. Despite this caveat, the empirical analysis presented in the next sections shows indicative evidence that the citizen-state bargain as presented in this paper remains an important factor of the social contract and possess a significant predictive and explicative power. This predictive capability of the framework can be useful to analyze exogenous changes to the capacities of a country coming from either the intervention of outside actors such as the World Bank or the introduction of new technologies which can dramatically impact capacities. For example, such an interpretation has been proposed to link how higher social media penetration had impacted the organization of social movement during the Arab Spring (Howard et al. 2011).
To summarize, the framework proposes to characterize social contracts along three interconnected aspects: 1) the citizen-state bargain, 2) the social outcomes, and 3) resilience and dynamicity. These three aspects can be described and measured using the 6 dimensions and their 14 sub-dimensions presented in the previous discussion. All these elements and their interactions are illustrated in figure 1 below. These 6 dimensions could form the basis of typology to discuss social contracts as, for example, extractive or inclusive, thin or thick, repressive or responsive, and mis-aligned or aligned.

Section 3: Empirical Methodology and Caveats
This section describes the empirical measurement strategy of the framework introduced above and discusses its limitations. The process of constructing the measures for the 6 dimensions consists of identifying proxy indicators for the 14 sub-dimensions that composed them. These proxies come from readily available indicators from multiple sources. The main databases used as sources include the Varieties of Democracy Index (V-DEM), the Worldwide Governance Indicators and the Afrobarometer surveys. The sources of the indicators for each of the sub-dimensions are presented in table 1 below. The final score for the empirical measure of a dimension is the equal weighted average of the score for each of its sub-dimensions, with all the indicators used as proxies scaled to a [0,1] interval. The framework can be presented in its long format by showing the measures for the 14 sub-dimensions or in a shorter format by showing only the 6 main dimensions. In the empirical analysis presented below, the data used is centered around the year 2015 but with a range going from 2013 to 2017 due to limited availability of certain sources. The number of countries in the sample consists of a maximum of 178 countries for the global cross-section, with some countries missing for certain variables, and around 22 countries for the Sub-Saharan Africa focus due to the use of the Afrobarometer as the main data source for that purpose. Table 2 below presents some summary statistics for the measures of the 14 sub-dimensions while the summary statistics of the underlying indicators are presented in Annex 1. As previously indicated, the data sources can be split into two types: expert-based and perception surveys. The expert-based sources 8 have the advantage of being available for a wider range of countries and longer time horizons which allow for a larger sample for cross-country comparisons and time-series analysis. The self-reported sources 9 are based on surveys and measure the perceptions of citizens. They are available for fewer years, usually coming in the form of successive waves and for fewer countries. Both types of data sources come with their own biases and measurement errors. Another limitation of the empirical strategy proposed in this paper comes from the approach of using secondary indicators as proxies which have limited accuracy to reflect the targeted sub-dimension. The variables included in the framework were selected based both an availability and accuracy criteria for the dimensions described in the previous section. The accuracy is imperfect, meaning that sometimes their scope is broader than the dimension they are meant to capture and other times it is narrower and does not fully capture the dimension. These limitations were addressed as much as possible by considering a large selection of data sources and many combinations of the indicators. New data sources and indicators are continuously being introduced and the framework could gradually become more precise as they are integrated.
The empirical strategy also comes with more general limitations. The first one which has already been mentioned is that the citizen-state bargain is only one of the factors influencing social contracts at any point in time. Many other factors are certainly at play such as a country's history, culture, citizens preferences and path-dependence. This limits the potential explicative power of the framework. A final caveat is that many additional aspects of social contracts could not be incorporated into the framework due to their qualitative nature or because they were too specific to a given context or country. An example 19 of an aspect that could not be incorporated in the framework at the time being is the sub-national variation within a country of the different dimensions. Including this aspect would require a level of granularity in the data that would go beyond the scope of this conceptual paper. A second example of a missing element concerns the intermediaries and bargaining mechanisms that frame the state-citizen relation. The qualitative and country specific nature of this aspect means it requires a country per country analysis which makes it difficult to quantify. A full social contract assessment for a country would involve a mixed-method methodology which would combine the quantitative framework presented in this paper with a qualitative assessment done by local scholars and experts. Intermediaries, bargaining mechanisms and sub-national variations could also be integrated through extensions of the framework and further research. Despite these caveats, the empirical analysis of the next section offers indicative evidence that the assumptions behind the framework and some of its predictions are verified.

Section 4: Empirical Analysis
This section presents a few empirical tests of the validity of the framework and introduces indicative evidence of an important result. The validity tests check whether the main assumptions underlying the framework and some of the predictions implied by it are corroborated in the data. The first assumption to be tested is the central idea of the framework that the citizen-state bargain partially explains social outcomes of the social contract. The second assumption tested is that on average countries with thicker, more inclusive and more responsive outcomes should feature higher levels in the alignment dimension. The third assumption tested is that if alignment depends on expectations and that perception surveys incorporate some of these expectations, then perceived measures of the social outcomes should be more strongly correlated with alignment then expert-based measures. Finally, this section presents indicative evidence for the important proposition that state capacity is necessary but not sufficient for social contracts to exhibit thickness, inclusiveness and responsiveness. This empirical result has direct operational implications for the World Bank and its partners.
To perform these tests and to support the result, the empirical analysis is based on a cross sectional analysis of country-level indicators centered around the year 2015 following the methodology described in the previous section. Unless specified, the indicators are the ones based on expert opinions due to their wider coverage of countries. The sample size varies for each regression depending on the availability of data with the highest sample size being 165 countries and the smallest being 122 countries. When regressions are limited to Sub-Saharan African countries, the sample is the 22 countries covered by the 2015-2016 Afrobarometer wave. This empirical strategy, combined with the previously mentioned profound endogeneity of all the dimensions as well as the data caveats, does not allow for any causality claim. However, the indicative evidence gathered is promising and further research using time series and instrumental variables is currently ongoing. Table 3 below presents evidence that higher state and citizen capacities are correlated with higher level of social outcomes, supporting the assumption that the citizen-state bargain is a significant factor determining social contracts and that the framework has some explicative power. The first two columns use Thickness as the dependent variable and both capacities as an independent variable. Columns three and four show regressions with Responsiveness as a dependent variable and the last two columns show regressions as the dependent variable. All six regressions control for a country's GDP per capita to show some robustness as well. In each of the 6 columns, the relevant capacity is statistically significant with a p-value below 0.01 and is also economically significant with a relatively high coefficient. As all variables are scaled between 0 and 1 making it easy to interpret the coefficient. For example, according to the table, an increase of one standard deviation (equal to 0.158) for the civil capacity indicator would be associated on average with a 0.138 higher score for Inclusiveness, which is slightly more than half a standard deviation (equal to 0.218) for this indicator.  Table 4 below presents evidence that on average higher levels of social outcomes are associated with better alignment of the social contract. Column 1 of the table presents the regression of Alignment on Thickness, column 2 presents the regression of Alignment on Inclusiveness, and column 3 presents the regression of Alignment on responsiveness. The table below shows that all three outcomes are statistically and economically significant, even when controlling for GDP per capita. Responsiveness and inclusiveness seem to be the most strongly correlated with a higher R-square and t-stat while Thickness is the least strongly correlated.  Table 5 below presents evidence that perceptions survey indicators are more strongly correlated with the alignment indicator then expert based survey. This claim is based on the assumption that citizens' expectations and perceptions are the determinant factors of a social contract's alignment. Because these regressions use the Afrobarometer database, the sample is limited to a group of African countries covered by the 2015-2016 Afrobarometer wave. It is important to keep in mind that a smaller sample also limits the statistical power. In the table below, columns 1 to 3 regress alignment on the three outcomes from expert-based indicators while columns 4 to 6 regress alignment on the three outcomes using perception survey indicators. As predicted, the perception-based columns are more correlated with their equivalent expert-based columns although not all of them are statistically correlated anymore. This is probably due to the smaller sample size. In both the expert-based and perception-based columns the inclusiveness indicator seems to be the most strongly correlated with alignment suggesting that perceptions on fairness of the social contract are especially important to citizens. The previous evidence supports the idea that the framework is based on valid assumptions. The regressions also suggest indicative answers to interesting and useful questions regarding the relative importance of the two capacities for each of the social outcomes. Looking back at table 3, it seems that state capacity is more strongly correlated with thickness and inclusiveness, while civil capacity is more strongly correlated with responsiveness. This result is consistent with a common-sense reasoning given that the first two dimensions require more involvement from the state to deliver the services and oversee their distribution while the components of responsiveness, associated to open-access regimes, are more associated with a strong civil society. Another key message of the bargaining mechanisms is that both capacities would be necessary for each outcome. Without state capacity, no level of civil capacity would make the state capable of delivering education services. Conversely, without civil capacity all of a state's capacity could be allocated on things other than public services, enforcing fair laws or protecting rights, or it could simply be extracted as economic rent by the leaders. The first of these direction (that state capacity is necessary for quality outcomes) should come intuitively to most in the international development community which has been building developing countries' state capacity for decades through technical assistance, trainings and reforms. The second direction (that civil capacity is necessary to achieve high levels of social outcomes) is less familiar and requires some empirical support.
A corollary claim rephrasing the two previous observations is that: "state capacity is necessary but not always sufficient to achieve high levels of social outcomes." To support this proposition, figure 2, figure 3, and figure 4 plot the indicators for the social outcomes plot thickness, inclusiveness and responsiveness over state capacity. Countries with high levels of civil capacity are identified with the black dots in the figure. A country is considered to have a high level of civil capacity if its indicator ranks in the top third of scores, i.e. is above 0.617 (the 66 th percentile). The figures also include lines that illustrate the best linear fit for each group of countries. The regression results for all three of these relations are in table 7 of annex 2. The regressions of table 7 use a dummy variable to show the interaction between the two capacities. The regressions show that the differences in slope or in levels presented in figures 2, 3, and 4 are statistically significant.
There is a clear positive correlation between state capacity and the outcome plotted. This positive association is showed by the positively slopped fitted lines shown in all three figures. There is a natural concentration of low civil capacity (red dots) countries on the left sides of the three figures which indicates some correlation between civil capacity and state capacity. Luckily, the correlation between the two capacities is weak enough to allow for sufficient variation to distinguish in each graph two distinct relations represented by the different colored lines. This distinction is statistically significant in all three cases as shown in table 7. In all three figures, countries that have high state capacity but low civil capacities (the red dots on the right side of the plots) generally have worst outcomes than countries with equivalent state capacity but higher civil capacity. This observation supports the proposition that state capacity is necessary but not always sufficient for higher levels of social outcomes. Figure 2 below plots the Thickness indicator against state capacity for the two types of countries (high civil capacity vs low). It shows that when state capacity is low, both high civil capacity countries and low civil capacity countries have low levels of thickness. However, as state capacity increases thickness increases faster in countries with high civil capacity than it does for low civil capacity countries. This translates in a steeper sloped dark line than red line. This difference in the slope is statistically significant at p<0.05 even when controlling for GDP per capita and using robust standard errors as shown in table 7.

Figure 2 Inclusiveness and the interaction between the capacities
Source: Author's calculations Figure 3 and figure 4 below plot responsiveness and inclusiveness against state capacity. They both show that that across all levels of state capacity, low civil capacity leads to lower levels of the outcome on average. In both cases, this is illustrated by the black dots being situated at the top of the graphs and by the black line being above the red line. Also, in both cases the difference between the two lines is statistically significant at p<0.05 for inclusiveness and p<0.01 for responsiveness even when controlling for GDP per capita and using robust standard errors as shown in table 7. This higher statistical significance for the responsiveness case also reflects the previously discussed stronger relation between responsiveness and civil capacity. These figures, combined with the regressions in table 7 of annex 2, provide indicative evidence that state capacity benefits from being combined with high civil capacity. Civil capacity seems to act as a complement, rather than a substitute, or a catalyst to state capacity to promote virtuous cycles of development. More evidence is required before this result can be robustly established but hopefully this paper can motivate such research. If substantiated, this finding could have an important impact on World Bank operations and policies given that at the moment most of the effort is spent on building state capacity while civil capacity is not seen as much as a priority.

Section 5: Cross-National Comparisons
The IEG report concluded that social contract terminology could be a useful analytical innovation with relevant operational implications particularly in situations of social unrest, but that currently, the World Bank has no formal conceptual framework or shared understanding of social contracts, leading to invalid uses. The previous sections proposed a conceptual framework to begin building that understanding and an empirical methodology to measure its dimensions. The current section describes a way to arrange the measurements in a manner that facilitates their interpretation. Developing a standardized terminology and typology for social contracts would enable an easier integration in the many World Bank documents and reports that could benefit from them such as the Systematic Country Diagnostics, the Country Partnership Frameworks, the Risk and Resilience Assessments and the Risk and Peace Building Assessments among others. Regarding these last two examples, the social contract typologies introduced in this section could readily complement the Fragility, Conflict and Violence typology which classifies countries on their unrest levels but based on a limited selection of indicators. 10 The measures of a country's social contract dimension are most useful when compared relatively to the ones of other countries from the region or some other reference group, or when compared to scores from the same country over multiple time periods. Additionally, some of the dimensions of the framework form 26 natural pairings due to their interrelations. Based on these two observations, the following paragraphs describe three social contract comparative maps, or compasses, graphed for easier interpretation: i) Resilience, ii) Citizen-state bargain, and iii) Social outcomes. Figures 5 through 7 below present the compasses for Sub-Saharan African countries. 11 In the figures, the red lines represent the median scores of the country group for each dimension represented. The red dots identify the seven countries that appear in the Sub-Saharan Social Contract regional report as case studies. As in section 4, the data used is centered around 2015.
The first compass is related to Resilience and social unrest and is presented in Figure 5 below which maps countries along the two dimensions of responsiveness and alignment. Theoretically, this typology could contribute to anticipating or explaining social unrest and breakdowns of the social contract based on the hypothesis that a misalignment between citizens' perceptions on social outcomes and their expectations could be resolved through peaceful renegotiation only if there is space for open dialogue. Without enough responsiveness from the state, the social contract could struggle to adjust to changing expectations, perceptions, preferences or bargaining powers and a growing misalignment could eventually lead to a breakdown, conflict or social unrest. In the graph, data points in the southwest quadrant (bottom left) are characterized by low levels on both dimensions and could be associated with low resiliency social contracts. Countries in the northeast quadrant (upper right) have higher than median scores for alignment and responsiveness which could be associated with resilience. Countries in the northwest (upper left) quadrant have a responsiveness above the median but below the median alignment. Finally, countries in the southeast (bottom right) quadrant have below median responsiveness levels but above median alignment levels. The hypotheses underlying the framework suggest that countries in the northwest quadrant would have a tendency to move eastward as responsiveness leads to better aligned policies if possible, given the capacities. Likewise, countries in the southeast quadrant would have a tendency to move westward as the lack of responsiveness would hinder the capacity of the social contract to adapt to changing conditions. Interestingly, both the northeast and southwest would feature stationary equilibria, ceteris paribus, due to self-reinforcing cycles. Rapid changes, exogeneous to the framework, can also occur that have the potential of pushing countries on a new trajectory. Recent, examples of such shift include Ethiopia, The Gambia, and Zimbabwe which could offer interesting case studies. At the moment, these predictions are solely theoretical and further empirical research is needed to confirm them. Robust tests of these predictions are beyond the scope of this paper but are a natural next step of the research.
The second compass, presented in figure 6 below, maps countries along the dimensions of civil capacity and state capacity. The purpose of this typology is to present some indication on whether the level of one of the capacities is constraining the social contract. Such a binding-constraint can come from a state having limited available resources to provide public goods and services, or from citizens struggling to mobilize or coordinate enough to sway policy. The graph indicates a strong correlation between the two capacities, something that was also previously observed in the empirical analysis of section 4. A possible explanation for this phenomenon is the kind of feedback loop that was previously discussed, and which relates to the strong endogeneity of the dimensions. An example of a feedback loop between education and civil capacity was given in section 2. It can also be extended to illustrate feedback loops between civil capacity and state capacity, by pointing out that improved access to education would not only increase civil capacity but would also increase state capacity as the state could hire better trained bureaucrats. Moreover, a quality education system requires a certain level of state capacity to recruit and pay the teachers as well as some level of civil capacity so that the state devotes the resources to it. This illustrates feedback loops for capacities with themselves or between them.

Figure 6 Capacities Across Sub-Saharan Africa
In figure 6, the correlation between the two capacities is apparent from the high number of countries situated in either the northeast quadrant (high capacity types) or the southwest one (low capacity types). Interestingly, a non-trivial set of countries also fall into the two other quadrants. Countries in the northwest quadrant exhibit high levels of civil capacity for their levels of state capacity, while countries in the southeast quadrant have low levels of civil capacity given their state capacity. Theoretical predictions on the direction of the evolution of the dimensions, similar to the ones made for the previous typology, are suggested by the framework. Given the endogeneity and feedback loops mentioned, it is possible that both the southwest and northeast quadrant feature a natural tendency toward stationarity through selfreinforcing cycles and everything else constant. Meanwhile, an argument can be made to suggest that countries on the northwest and southeast corners would have a tendency to gradually move toward one of the two stationary quadrants following a step process in either direction. However, more research is needed to disentangle whether the lower dimension would drag the other one down or, vice versa, whether the higher dimension would pull the lower one up. To disentangle these effects, future research could add a time-series dimension to the analysis and look at natural or quasi-experiments.
The third compass, presented in figure 7 below, maps countries along the two social outcome dimensions of thickness and inclusiveness. In the graph, thicker and more inclusive social contracts are situated in the northeast corner while thinner and more extractive social contracts are in the southwest corner. The positive correlation seen on the graph is consistent with the framework's assumption that the citizenstate bargain explains to some extent both these outcomes. Beyond highlighting countries with higher scores on the on the social contract outcomes index and those featuring lower scores on those outcomes, this typology is also informative on outliers that do not fit the predictions of the model. Countries in the northwest corner have high scores for the inclusiveness index given their level on the thickness index, while countries in the southeast corner have relatively high levels of thickness given their level of inclusiveness. In some cases, these outliers could present opportunities and entry points for potential policy dialogue. For example, a country might have gone through a period of high citizen mobilization and political will around a social project to improve health services which boosted the thickness score higher relatively to the inclusiveness score. It is possible that this mobilization of state and civil capacities also laid down favorable conditions, for an equivalent reform program to reduce corruption or increase the fairness of the justice system, which would boost inclusiveness of the social contract. The fit of the framework is also limited due to the fact that the citizen-state bargain is only one of the factors contributing to a country's selection of policies which is one of the framework's limitations previously discussed.
In summary, this section proposed three compasses to interpret the empirical measures of the dimensions included in the framework. According to the first compass, a resilient social contract exhibits relatively high level of alignment between expectations and perceptions, as well as a high level of responsiveness which allows for adjustments to the compact when conditions change. According to the second compass, the constraints on the outcomes of the citizen-state bargain could be either from limited civil capacity, state capacity or both. Finally, according to the third compass, the performance of a social contract can be skewed toward inclusiveness or thickness, relatively low for both or relatively high for both. Perceptions on these outcomes are not directly included in this section due to the narrower samples associated with perception surveys.

Concluding Remarks
This paper proposed a framework to help conceptualize and measure three important aspects of social contracts: the citizen-state bargain, the social outcomes, and the resilience of the contract. This work is particularly relevant given the growing use of social contract terminology the World Bank's lexicon and following a 2019 IEG review which concluded that social contract diagnostics are useful analytical innovations with relevant operational implications, especially in countries experiencing social unrest where social contracts are in transition, but that the World Bank currently has no formal conceptual framework or shared understanding of social contracts, leading to uneven uses. The terminology and typologies proposed in this paper offer a first effort at building this shared understanding and standardizing the utilization of social contract theory by the World Bank.
The paper described an empirical strategy to quantify these three aspects through 6 dimensions and 14 sub-dimensions. The strategy makes use of readily available indicators from a wide range of data sources including from both expert-based and perceptions-based surveys. A preliminary empirical analysis of the data provided indicative evidence that many of the assumptions and predictions of the framework were founded, including that both types of capacity are strongly positively correlated with social outcomes, that better social outcomes are correlated with more aligned social contracts, and that, due to their link with expectations, perception measures better predict alignment than expert-based measures. The empirical analysis also offered supportive evidence for an important finding, that given similar state capacity, countries with higher civil capacity are associated with thicker, more inclusive and more responsive social contracts. This finding has implications for World Bank operations and programs and could inform binding constraint analysis done in country diagnostics.
The empirical analysis presented in the paper is still preliminary and only offers indicative evidence to support many of the claims and predictions of the framework. Further work is needed to bring definitive proof on causality claims and to dig deeper into the implications. Future research could involve a time series analysis to look at the evolution of social contracts through time or assess causality links through an instrumental variable setup or quasi-experiments. Moreover, the current framework is incomplete as many aspects including bargaining mechanisms, sub-national variations and intermediaries are missing from it. However, the framework offers a foundation upon which extensions could be built to incorporate these elements and many more. Ongoing work in Haiti is specifically interested in exploring questions of decentralization and local governance through the social contract frame which is motivating the development of an extended framework to take this aspect into account. The framework could continuously improve through more of these case studies and applications.
Based on the framework and its empirical measurements, the paper described different compasses to classify social contracts and demonstrated their application on a sample of Sub-Saharan African countries. One of these compasses relates to resilience and maps countries along the two dimensions of alignment and state responsiveness. The paper suggested that this compass could help anticipate or explain social unrest based on whether or not a peaceful resolution to misalignments of the social contract is possible through dialogue. The assumption made was that a social contract with misaligned perceptions of the social outcomes with the citizens expectations could find a peaceful resolution to the issue depending on the responsiveness level from the state but that not enough responsiveness would lead to social unrest and potentially a breakdown of the social contract. The second compass maps countries along the dimensions of state capacity and civil capacity and was used to discuss issues of binding constraints that limit social contracts from being more inclusive, responsive and thicker. The third compass presented countries along the dimensions of thickness and inclusiveness and is related to the observed social outcomes that form the content of the social contract.
The compasses contribute to develop a standardized terminology and typology for social contracts that would facilitate their integration into the many World Bank documents and reports that could benefit from them, such as the Systematic Country Diagnostics, the Country Partnership Frameworks, the Risk and Resilience Assessments and the Risk and Peace Building Assessments, among others. Regarding these last two examples linked to social unrest, the social contract typologies introduced complement the Fragility, Conflict and Violence typology which classifies countries on their unrest levels but is based on a limited selection of indicators. By itself, the empirical framework only offers a partial understanding of social contracts due to the limitations of the data and of the model itself. To have a more complete understanding, it is necessary to combine the empirical approach with a qualitative and contextual one. A robust social country diagnostic toolkit would be centered around a mixed-method approach that would emphasize the local knowledge of scholars and interlocutors of the specific country being investigated.

31
The IEG (2019) report mentions that a social contract framing can help to diagnose and explain complex development challenges such as entrenched inequalities, binding constraints, poor service delivery, weak institutions, and why decades of policy and institutional reforms promoted by external development actors had uneven effects on countries' development paths. This paper proposes three potential contributions of social contract diagnostics for development institutions: i) to offer an integrative perspective connecting the different elements of a country's development spheres (political, social, and economic), ii) to help anticipate indirect effects from international development aid on the social contract which could lead to unforeseen consequences, and iii) most importantly, a social contract lens places the citizens front and center of the development effort and not only has beneficiaries but also as major stakeholders and actors with intrinsic agency.