WPS6158 Policy Research Working Paper 6158 Can Donors and Non-State Actors Undermine Citizens’ Legitimating Beliefs? Audrey Sacks The World Bank Africa Region Poverty Reduction and Economic Management Detpartment August 2012 Policy Research Working Paper 6158 Abstract This paper addresses the conditions under which and managing these resources. The author assesses these donor and non-state actor service provision is likely to competing hypotheses using multi-level analyses of undermine or strengthen citizens' legitimating beliefs. On Afrobarometer survey data. The sample, drawn from a the one hand, citizens may be less likely to support their continuum of developing societies in Africa, allows for government with quasi-voluntary compliance when they analysis of associations between donor and non-state credit non-state actors or donors for service provision. actor service provision and the sense of obligation to On the other hand, the provision of goods and services comply with the tax authorities, the police and courts. by donors and non-state actors might strengthen citizens' The findings yield support for the hypothesis that the confidence in their government and their willingness to provision of services by donors and non-state actors is defer to governmental laws and regulations if citizens strengthening, rather than undermining, the relationship believe that the government is essential to leveraging between citizens and the state. This paper is a product of the Poverty Reduction and Economic Management Departyment, Africa Region. 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://econ.worldbank.org. The author may be contacted at asacks@worldbank.org. 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. Produced by the Research Support Team Can Donors and Non-State Actors Undermine Citizens’ Legitimating Beliefs?∗ Audrey Sacks† Keywords: Sub-Saharan Africa, Legitimacy, Aid Effectiveness JEL: H1, H2, H4,D73, Sector Board: PSM ∗ e I thank Margaret Levi, Ga¨l Raballand, Lisa Schmidt, Kate Stovel and Mike Ward for their valuable feedback. † asacks@worldbank.org. Extended Term Consultant, PREM, Africa Region 1 1 Introduction In June 2005, the Office of the U.S. Global AIDS Coordinator (OGAC) claimed that there were over 32,000 citizens of Botswana receiving anti- retroviral treatment (ARV) because of PEPFAR (U.S. President’s Emer- gency Plan for Aid Relief). In response, Segolame Ramotlhawa, director of Botswana’s treatment program, called the U.S. �gures “a gross misrep- resentation of the facts� and reported that no patient was dependent on PEPFAR for treatment. Politically, the credit claiming was important to OGAC because it needed to demonstrate progress to the U.S. Congress. For Botswana’s President, Festus Mogae, receiving credit was important because he promised universal ARV treatment and has made HIV/AIDS one of his top issues. President Mogae did not want Botswana’s citizens to think that the U.S. was the only one responsible for the country’s treatment success (Patterson, 2006; Timberg, 2005, 159-161). This exchange between Botswana’s government and OGAC officials illus- trates one of the dilemmas that foreign aid poses for donors, non-governmental organizations (NGOs), governments, and citizens. If the priorities of a gov- ernment mirror those of PEPFAR and PEPFAR provides much of the money needed to meet them, who receives the credit for success or the blame for failure? Different actors have funded the various components of the ARV treatment program: the Gates Foundation has built AIDS clinics; Merck Pharmaceutical Company has donated ARVs; the Government of Botswana pays doctors and nurses; PEPFAR money has purchased laboratory equip- ment; and Harvard has trained lab technicians (Patterson, 2006, 159-161). In this context, who do recipients believe is ultimately responsible for the HIV/AIDS program? These questions raise a broader concern about the conditions under which the provision of services by donors and non-state actors is likely to under- mine or strengthen the �scal contract between citizens and a government and citizens’ legitimating beliefs. Before addressing this concern, I will �rst examine whether the logic of the �scal contract, which is largely based on the statebuilding experiences of Europe, is relevant to developing countries with very low and idiosyncratic taxation levels. I will then assess whether there is empirical support for the �scal contract argument using cross-sectional survey data, the Afrobarometer, which was collected across 19 Sub-Saharan African countries in 2008. Second, extending the logic of the �scal contract arguement, there is 2 reason to believe that the provision of services by donors and non-state actors could either strengthen or weaken citizens’ deference to government. I will identify the conditions under which these two scenarios are likely to occur. On the one hand, citizens may be less likely to support the government with deference to its laws and regulations when they credit non-state actors or donors for service provision. The provision of services by donors and non- state actors is likely to prompt citizens to question why they should pay taxes to a government that is not providing them with anything in exchange. On the other hand, the provision of goods and services by donors and non-state actors might strengthen citizens’ legitimating beliefs and their willingness to defer to governmental laws and regulations if citizens view their government as essential to leveraging and managing these external resources. I assess these competing hypotheses using multi-level analyses of Afro- barometer survey data. The sample, drawn from a continuum of developing societies in Africa, allows us to analyze associations between donor and non- state actor service provision and the sense of obligation to comply with the tax authorities. Third, I assess the relationship between the provision of ser- vices by donors and non-state actors and citizens’ willingness to defer to two additional authorities, the police and courts. The �ndings yield support for the hypothesis that the provision of services by donors and non-state actors is strengthening, rather than undermining, the relationship between citizens and the state. 2 A Model of Legitimating Beliefs I extend the logic of the �scal contract to consider the implications for cit- izens’ perceptions of their government’s legitimacy when actors other than the state are involved in providing the basic services that are at the heart of the �scal contract relationship. My conception of government legitimacy and the reasons citizens adopt legitimating beliefs derives from the work of We- ber (1968). I use the concept of legitimacy to denote acceptance of the right of governors to rule; its effect is an increased likelihood of quasi-voluntary compliance with governmental rules and regulations. Quasi-voluntary com- pliance is compliance motivated by a willingness to comply but backed up by coercion, particularly coercion that ensures that others will be obeying the law (Levi, 1997). I treat legitimacy as a concept with two components: value-based legiti- 3 macy (a sense of obligation or willingness to obey authorities) and behavioral- based legitimacy (actual compliance with governmental regulations and laws). Theoretically, value-based legitimacy should translate into behavioral legiti- macy. Empirical work shows that value-based legitimacy is a function of four antecedent conditions: procedural justice; leadership motivations; adminis- trative competence; and, government performance.1 3 Fiscal Contract Fiscal theories of governance, which grew largely out of the state-building experiences of Europe, argue two main points. First, the shape political institutions assume reflects the government’s need for revenue (Bates and Lien, 1985; Levi, 1988; Moore, 2004; Tilly, 1990). A government has an incentive to meet citizens’ demands for political representation and public goods provision only when it is dependent on them for revenue. Second, taxpayers bene�t from government policies roughly in proportion with the share of government revenue they �nance (Bates and Lien, 1985; Boix, 2003). Several key comparative historical works demonstrate that rulers have incentives to reduce the cost of compliance by making credible commitments to citizens, according them representation in government and providing them with public goods. In exchange, citizens are more likely to willingly comply with tax demands thus reducing the costs to governments of ensuring com- pliance through coercion alone (Levi, 1988; North and Weingast, 1989; Tilly, 1990). There is also evidence from a wide range of developed and develop- ing countries that governments which rely on broadly based taxes are likely to establish a �scal contract with citizens (Bergman, 2003; Fjeldstad and Semboja, 2000, 2001; Guyer, 1992; Levi and Sacks, 2009; Lieberman, 2001, 2003; Tripp, 1997; Tyler, 1990). These works demonstrate that the more citizens perceive that government is upholding its end of the �scal contract with constituents with serious efforts to deliver infrastructure and services, the more likely it is to produce a virtuous cycle of citizen compliance with its extractions, leading to better government and greater compliance (see 1 See Levi and Sacks (2009) and Levi, Sacks and Tyler (2009) for a discussion of why these four conditions matter for legitimating beliefs and for an empirical test of the relation- ships between these four conditions and legitimating beliefs in the context of Sub-Saharan Africa 4 Figure 1).2 For example, there is also some evidence that when tax increases are linked to improvements in public goods provision, citizens are less likely to resist the tax increases. In Ghana, the government linked increases in the VAT rate explicitly to the new public spending programs, such as the Ghana Education Trust (GET) Fund in 2002 and the National Health Insurance Scheme (NHIS) in 2003 which enjoyed broad public support. The government used strategic communication to make this link in order to avoid major public protests, such as the Kume Preko protests that greeted the introduction of the VAT in 1995 and left several people dead (Osei, 2000; Prichard, July 2009). Similarly, Ghana’s government linked the introduction of a talk tax on mobile phone calls to efforts to combat youth unemployment, which helped to curb public opposition (Prichard, July 2009). 3.1 Is donor and non-state actor service provision likely to undermine the �scal contract? We are beginning to accumulate knowledge about what government can do to influence the perception of the relationship between citizens and political authorities. We know very little about what happens once non-state actors mediate that relationship. If the �scal contract rests on the provision of basic welfare bene�ts, what happens when these bene�ts are provided by actors other than the state (see Figure 2)? When is the provision of services by non- state actors likely to strengthen the �scal contract between citizens and the state? My primary hypothesis is that the provision of services will undermine legitimating beliefs when citizens credit non-state actors for high-quality ser- vice provision or blame the government for low quality service provision. Under some conditions, speci�cally, when citizens view their government as essential to leveraging and regulating external resources, non-state service provision is likely to improve citizens’ legitimating beliefs. In weak and fragile states that lack the capacity to perform the basic functions of government, other actors often step in to �ll the gap. Donors and non-state actors including NGOs, private entrepreneurs and companies, communities, and churches, are increasingly becoming major providers of hu- manitarian aid and other essential public goods including health care, refugee settlement, security while constructing hospitals, roads, schools, irrigation 2 Levi and Sacks (2009) reviews this literature. 5 networks, and other infrastructure. Estimates suggest that between a third and two-thirds of education and health services now completely bypass the state (van de Walle, 2001, 100). The rise in the number of NGOs from less than 200 in 1909 to nearly 1000 in 1956 to over 20,000 currently points to their growing importance in public goods provision (Union of International Associations (UIA), 2005).3 The World Bank, international donor institutions and governments are encouraging non-state actors to play a larger role in the provision of public goods and social welfare in developing countries and emphasizing public- private partnerships. Their rationale for expanding non-state welfare pro- vision generally emphasizes the inability of state institutions to meet basic welfare needs (Devarajan and Reinikka, 2004; Edwards and Hulme, 1996; Marcussen, 1996) or the greater ability of community-groups and NGOs to identify and tailor projects to local-level needs (Mansuri and Rao, 2004; Plat- teau, 2004). Increasingly, governments are contracting service provision out to private actors in order to create competition where government agencies are viewed as corrupt or inefficient (Smith and Lipsky, 1993). There is some evidence that foreign aid has harmful consequences for a country’s political development (Azam, 2001; Brautigam, 2000; Kaldor, 1963; Knack, 2001). Governments dependent on foreign aid also have fewer incen- tives to build or improve domestic revenue collection, are more accountable to external donors than to domestic parliaments and constituencies, and are therefore less likely to build accountable political institutions (Moss, Pet- tersson and van de Walle, 2006). One study found that local governments that rely on transfers from external donors or the national government for revenue are less responsive to their constituencies development needs than local governments that depend on local tax revenue (Gibson and Hoffman, 2005). In Zambia, for example, World Vision among other international NGOs has been providing a substantial number of services for Zambia’s citizens. World Vision estimates that about one-quarter of Zambia’s population are bene�ting from this work. In 2008 to 2009, I carried out a unique survey throughout the Southern, Eastern and Copperbelt provinces.4 In a ward in 3 The amount of discretionary funding that high-income countries have given to NGOs to promote international development assistance has risen from a negligible amount before 1980 to nearly $2 billion in 2004 (OECD, 2010). Currently, the NGO sector currently constitutes a 1.1 trillion dollar industry (Edwards, 2004, 21) 4 This �eldwork was carried out as part of my �eld work for my PhD dissertation. The 6 the Eastern province, where World Vision is particularly active in service delivery, 49.06 percent of respondents believe that public services in their community have improved over the past year. Of these respondents, 56.52 percent credit World Vision and only 26.09 percent credit the government for the improvement in service delivery. The remaining 17.39 percent of respon- dents credit other NGOs for the improvement in service delivery. Of the 47.17 percent of respondents who do not believe that services have improved over the past year, 88.0 percent blame the government and 12.0 percent blame the community. Not one respondent blamed an NGO or donor for the lack of improvement in service delivery. Further, respondents were asked if the decision was up to them, to whom would they prefer to pay their taxes. Half of the ward’s respondents said that they would prefer to pay their taxes to donors, NGOs, or other non-state actors and the other half would prefer to pay their taxes directly to the government. Unlike sectarian and rebel movements that tend to flourish where they are able to provide the goods that citizens are not receiving from their own government (Chen, 2004; Fritsch, October 2, 2001; Iannaccone and Berman, 2006), NGOs and donors do not have aspirations to wrest control from the government and are therefore unlikely to replace the state as a source of authority. Yet, they still pose a threat to states in terms of undermining cit- izens’ perceptions that the state is not ful�lling its end of the �scal contract. Several governments including Uganda, Ethiopia, and Zambia have passed bills or taken other concrete actions designed to control and regulate NGOs. Although these actions are primarily directed at limiting the ability of NGOs to oppose government policies, they are also aimed at enhancing the state’s role in service delivery in the eyes of its citizens. 4 Can the provision of services by donors and non-state actors strengthen the �scal con- tract? Many scholars argue that the effect of aid is contingent on the institutional en- vironment in place (Burnside and Dollar, 2000, 2004; Kosack, 2003; Morrison, 2009, January 2010). Where accountable government institutions are in place survey was funded by an Opportunity Fund generously provided by the World Justice Project. 7 or a leader is particularly committed to development, governments are likely to work with donors, international agencies and non-state providers to effec- tively manage and utilize aid. In these contexts, aid is likely to strengthen rather than undermine the �scal contract. The Government of Botswana, for example, has an impressive record of coordinating with donors to ensure that aid is effectively utilized to promote development, which has contributed to a record of sustained economic growth. By law, all aid activities are in- tegrated into Botswana’s planning and budgeting processes (van de Walle, 2001, 205-206). In Malawi, large-scale mission-hospital and school systems work closely with government, while retaining some management autonomy (Kadzamira et al., 2004). Other developing countries have experimented with co-production, or the provision of public services through an informal relationship between state agencies and communities, where both make sub- stantial resource contributions (Joshi and Moore, 2004). Some states may not directly partner with donors and non-state actors to provide services but rather support, co-ordinate and manage them as to improve the effectiveness of service delivery. Through performing regulatory functions, the state can assume overall responsibility for service provision without necessarily delivering services. States have come up with second- best solutions when universal regulation is impractical. For example, the government in Bangladesh has instituted a system, where the government rewards those non-state providers that achieve certain performance stan- dards. With the World Health Organization (WHO), the government issues vouchers to poor pregnant women, which they can only use to pay accredited providers of antenatal care and delivery (Chowdury et al., 2004; Palmer et al., 2006). Recipient governments might also elicit credit for donor and NGO aid when politicians are able to effectively influence constituents’ assignments of credit and blame (Crant and Bateman, 1993; Weaver, 1986). Governments throughout the world regularly claim credit in the media for non-state actor service provision. The vast majority of citizens are likely to have poor information about the state’s formal relationship with donors and non-state actors and its reg- ulation of aid and non-state service provision. Citizens are likely to evaluate the state’s role in development activities based on their every-day observation of and interaction with street-level bureaucrats including agricultural exten- sion agents, teachers, health care officials, social workers (Kelly and Swindell, 2003; Lipsky, 1980; Tendler, 1997). When citizens observe bureaucrats col- laborating with NGO staff in the �eld, they are likely to forge a positive 8 impression of the state’s role in development and give credit to the state for helping to leverage external resources. Citizens are also likely to give the state credit where mechanisms to voice complaints about non-state actors exist and where bureaucrats are able to effectively respond to complaints. Under these conditions, non-state service provision is likely to strengthen the �scal contract. 5 Data and Methods I explore the relationship between external service provision and deference to government using Afrobarometer survey data from 19 Sub-Saharan African countries (see Table 1). Africa is an especially good place to examine these issues because of the large amount of variation both within and across African countries in the extent to which non-state actors, donors and other states are active in service provision and the extent to which governments are relatively effective and fair. Government responsiveness, corruption and reliance on non-public resources vary considerably among localities with consequences for citizen understanding of and relationship to government (Gibson and Hoffman, 2005). This project relies on the fourth round of Afrobarometer data that surveys Africans’ views towards democracy, economics, and civil society with random, strati�ed, nationally representative samples. In 2008, trained enumerators conducted face-to-face interviews in local languages with 26,513 respondents across 19 countries.5 The sample is designed as a representative cross-section of all citizens of voting age in a given country. The dataset used for this paper has a multilevel structure; individuals are nested within primary sampling units (PSUs), which are nested within countries. The PSUs are the smallest, well-de�ned geographic units for which reliable population data are available and they tend to be socially homoge- nous, thereby producing highly clustered data. In most countries, these will be Census Enumeration Areas (Afrobarometer, 2005, 37-38). Although re- spondents were not sampled based on their ethnic affiliation, there is likely to be a high level of clustering in the dataset around ethnicity. In other work, I discuss the advantages of multilevel modeling (Levi and Sacks, 2009). Treating the dependent variable as a binary outcome and taking into account the multilevel nature of our data, I estimate random intercepts for 5 I excluded Zimbabwe from the analysis because of missing data on key variables. 9 PSUs, countries, and ethnic groups with the system of equations described in Equation 1: Level 1 (Individual) : Log[(Pij(kl) )/(1 − Pij(kl) )] = β0j(kl) + Xj(kl) + aij(kl) Level 2 (PSU) : β0j(kl) = δ0j(kl) + b(kl) Level 3 (Country and Ethnic Group) : δ00(kl) = λ0(k) + γ0(l) + c0 + d0 In this model, i, j, k, and l index the individual, PSU, country, and ethnic group levels. The Greek symbols refer to parameters estimated at particular levels of analysis: β, δ, λ, and γ refer to �xed effects at the individual, PSU, country, and ethnic group levels, respectively. Xj(kl) refers to individual socio-demographic factors and individual-level perceptions of bureaucracy, procedural justice and service delivery. C0j(kl) includes the objective indica- tors of service delivery that are measured at the enumeration-level. C0(kl) includes country-level controls, such as GDP per capita and foreign aid per capita. The random effects of intercepts at the individual, PSU, country, and ethnic group level are described by aij(kl) , b0j(kl) , c0(k) , and d0(l) . The logistic multilevel model expresses the log-odds (i.e. the logit Pij(kl) as a sum of a linear function of the explanatory variables and random-group and random effect deviations. 6 Measures 6.1 Dependent Variable My dependent variable is whether respondents agree with the statement: “The tax department always has the right to make people pay taxes.� Re- spondents’ answers were originally coded as “strongly disagree�, “disagree�, “neither disagree nor agree�, “agree�, “strongly agree�, “don’t know� and “refused to answer�. For two reasons, I dichotomized this variable from a �ve-point scale by collapsing “strongly disagree� and “disagree� into one category and the other three responses into another category. First, I am only interested in the comparison of people who accept government’s right to make people pay taxes versus those who express some ambiguity about this right and those who reject this right. Second, I estimated a multilevel ordered probit and the results from this model are very similar to the results 10 obtained from the multilevel logistic model. Thus, I am con�dent that I did not lose substantial statistical information by dichotomizing this variable. 6.2 Independent Variables 6.2.1 Socio-Demographic Variables I control for standard socio-demographic variables that can affect citizens’ acceptance of government’s right to make people pay taxes. A question prob- ing respondents on their household income was not included in the fourth round of Afrobarometer surveys. Asking respondents to quantify their in- come can be problematic in the context of developing economies, where in- dividuals are often embedded in barter or commodity exchange, rather than, market economies. There are, however, reasonably good proxies including whether respondents own a television, radio, car, and mobile phone, and use the internet. Age, education, employment, and urban or rural residence are demographic factors that also affect household resources. 6.2.2 Experience with Paying Taxes or Fees It is difficult to assess just how ubiquitous taxes are in ordinary Africans’ lives. There has not been any systematic effort to take stock of the types and amount of taxes citizens pay across Africa. Similar to pre-modern European states, African states’ revenue raising capacity is generally low. On average, the tax-to-GDP ratio in Sub-Saharan Africa is around 21 percent, compared with the OECD average of about 32 percent. In Tanzania and Uganda, the total tax share drops to about 10 percent. Historical data suggests that the tax share of many European countries did not reach 15 percent of GDP until World War II when incomes were substantially higher than they are in many African countries (Fjeldstad and Rakner, 2003, 3). The types and amount of taxes citizens pay varies both within and be- tween countries. We do know there are taxes on agricultural crops, but the rates and processes of collection vary within countries (Kasara, 2007). User fees from electricity, water, sanitation, and other services comprise the major- ity of local revenue in South Africa (Hoffman, 2007). In Tanzania, Fjeldstad and Semboja (2001) count ten major categories of taxes, eighteen major categories of licenses, forty groups of charges and fees, and seventeen items listed as other revenue sources. In some countries including Kenya, Malawi, 11 Tanzania, Uganda, and Zambia, national level semi-autonomous revenue au- thorities are in charge of collecting a limited number of taxes. The Ugandan Revenue Authority (URA), for example, collects all major indirect taxes and some direct taxes (income tax, but not property or social security tax). Even with the presence of national revenue authorities, there is often little or no coordination in setting and collecting taxes between the various levels of government in Africa (Fjeldstad and Rakner, 2003, vi). The Afrobarometer asks respondents whether they have paid �ve differ- ent types of taxes and fees within the year preceding the survey: fees for a government service, such as education or health care; licence fees to lo- cal government, such as a bicycle, cart, or market stall; property rates or taxes; public utility fees; and, income taxes. The majority of respondents, 68 percent, have experience with paying a fee for a government service. 47 percent of respondents have had experience with paying a public utility fee. One-quarter of respondents have experience with paying a license fee to a local government. 24 percent of respondents have experience with paying a property rate or tax. Only 22 percent of respondents have had experience with paying an income tax. Table 2 shows a large amount of variation between countries in the per- centage of respondents who report having paid a tax or fee to government. For example, 83.7 percent of respondents from Senegal have experience with paying a fee for a government service compared to 38.4 percent of respondents from Mali. Forty-two percent of respondents from Ghana have experience paying a license fee to their local government compared to 3.5 percent of re- spondents from Mozambique. Although the percentage of respondents who have experience paying an income tax is low, there is a substantial difference between Liberia and Mali. Forty percent of respondents from Liberia have experience paying an income tax compared to 10.3 percent of respondents from Mali. I control for respondents’ experience with paying an indirect tax, an income tax, and a direct tax, property rates or taxes.6 6.2.3 Government Effectiveness Quasi-voluntary compliance is most likely among those who believe that a government is trying to meet its end of the �scal contract by delivering 6 I also included controls for experiences with paying fees for licenses, government ser- vices and utilities. Because these variables were not signi�cant at the p< 0.05 level, I left them out of the regression tables. 12 essential infrastructure and services or quickly developing the competence to do so. I measure three dimensions of government’s performance in ful�lling its �scal obligations to citizens. First, I include a measure of respondents’ food security, speci�cally, whether a respondent or a household member ever went without sufficient food in the year preceding the survey. Sacks and Levi (2010) �nd that food security is strongly related to government performance in providing and maintaining infrastructure, reliable bureaucracies, and law and order, controlling for in- come, rainfall levels, the competitiveness of elections, and the presence of civil liberties and political rights. Second, I include two measures of citizens’ perceptions of government service delivery. I include an indicator of citizens’ perceptions of how well the government is maintaining roads and bridges. I also include an indicator of citizens’ perceptions of how well the government is providing water and sanitation services. Third, I include two objective indicators of government service delivery. Survey enumerators identi�ed whether there was a sewage system and school in each of the primary sampling units.7 6.2.4 Perceptions of Administrative Competence and Honesty I include two measures of citizens’ perceptions of the honesty and trans- parency of the bureaucracy. First, I include a measure of whether citizens believe that a large portion of tax administrators is corrupt. Second, I in- clude a variable indicating whether citizens approve of how well their local government is handling the collection of license fees on bicycles, carts and barrows.8 Third, both the size of a country and the size of the government may affect a government’s ability to detect and punish evaders. I include the 7 I also include a country-level indicator of government performance, the World Bank Governance indicator of government effectiveness, in the model. This indicator measures perceptions of the quality of public services, the quality of the civil service and the degree of its independence from political pressures, the quality of policy formulation and implemen- tation, and the credibility of the government’s commitment to such policies (Kaufmann, Kraay and Mastruzzi, 2006, 4). This variable is not signi�cant at the p<.05 level. 8 I included two additional measures in the model neither of which were signi�cant at the p<0.05 level. One is a measure of citizens’ approval of how well their local government council provides citizens with the information about the councils budget (i.e. revenues and expenditures). The other, the World Bank governance indicator, control of corruption, measures the extent to which public power is exercised for private gain, as well as capture of the state by elites and private interest (Kaufmann, Kraay and Mastruzzi, 2006, 4). 13 natural log of a country’s population size and government consumption as a share of GDP; the latter is the most widespread measure of government size (Adsera and Boix, 2002; Alesina and Wacziarg, 1998; Rodrik, 1998). Tax compliance is also related to the government’s ability to effectively detect and punish tax avoiders, tax evaders, and tax arrears. Although an imperfect measure of states’ deterrent capacity, Afrobarometer includes two survey questions on perceptions of government enforcement and monitoring capacities. One question probes respondents on how often ordinary people who break the law go unpunished. The other probes respondents on how often officials who commit crimes go unpunished. This latter question is also a measure of perceived government fairness - the extent to which a government implements the law evenly across all social groups. 6.2.5 Procedural Justice I include two indicators of procedural justice. The �rst probes respondents on how often people are treated unequally under the law. The next taps citizens’ perceptions of the government’s treatment of their ethnic group. Speci�cally, respondents were asked how often their ethnic group is treated unfairly by their government.9 6.2.6 Donor and Non-State Actor Provision of Services I include a measure of who citizens believe is responsible for collecting income taxes. I expect citizens who perceive tax collection to be the responsibility of non-state actors to be less likely to be willing to pay taxes to the state’s tax department than citizens who perceive tax collection to be the responsibility of the state. Perceptions of the effectiveness of donor and non-state actor provision of services are assessed using the following items. Respondents were probed on how much they believe the following non-state actors and donors do to help their country: the United Nations; international donors and NGOs; international businesses and investors; China; and the United States. 9 I also include Freedom House’s political liberties and civil rights ratings for the 19 countries in the sample. These two variables should capture the relative equality of influence in making policy. They indicate whether citizens are able to express their voice without fear of repression and whether elections are free and fair. Neither of these variables are signi�cant at the p<0.05 level. 14 Respondents were probed on whether they believe the following donors and non-state actors exert too much, too little or just the right amount of influence over their government: domestic civic organizations and NGOs; and, international donors and NGOs. These two variables indirectly capture citizens’ normative views of non-state service provision, speci�cally, whether donors and non-state actors are helpful or harmful for their country. I would expect citizens who believe that donors and non-state actors have too little influence over their government to be less likely to willingly defer to their government. These respondents probably believe that their government needs the assistance of donors and non-state actors to provide essential goods and services. A perception that donors and non-state actors exert too much influence over their government may not necessarily translate into a greater willingness to defer to the government. Some may view the influence donors and non-state actors exert over their government positively and believe that the more influence donors and non-state actors exert over their government, the better off their country. Others may view the influence donors and non-state actors have over their government negatively. If citizens view the influence donors and non-state actors exert over their government negatively, it is not clear whether they are more or less likely to be willing to pay taxes. Citizens have multiple strategies for confronting the state. On the one hand, some may want to bolster the strength of the state to collect revenue from, and provide public goods to, citizens. An increase in a state’s extractive capacity should correspond to less dependence on donors and NGOs and an increase in state autonomy. On the other hand, others may want to try and replace the regime with a stronger government that is committed to standing up to donors and non-state actors and re-claiming the state’s sovereignty. Finally, I also include several country-level measures of donor and non- state actor activities. First, I include a measure of the number of intergovern- mental organizations (IGOs) and NGOs of which a country or territory is a member, whether directly or through the presence of members in that coun- try for 2007 (Union of International Associations (UIA), 2005). I excluded dead, inactive or uncon�rmed organizations. Second, I include a measure of the amount of foreign aid a country receives from bilateral and multilateral donors. Speci�cally, I include the natural log of net Official Development Assistance (ODA) received per capita (current US $) for 2007. Third, I in- clude a measure of food aid. The Food and Agricultural Organization (FAO) compiles food aid shipments, which represents a transfer of food commodi- 15 ties from donor to recipient countries, on a total-grant basis or on highly concessional terms (FAO). Fourth, I include a measure of country’s depen- dence on health-related aid, speci�cally, external health-related resources as a percentage of the total expenditure on health for 2006 and 2007. Unfortunately, there are no quantitative indicators of a country’s man- agement and regulation of foreign aid and non-state service provision. Nei- ther are there quantitative indicators of citizens’ perceptions of the extent to which bureaucrats collaborate and cooperate with donors and non-state actors in development activities. I include a country-level indicator, regu- latory quality, which captures perceptions of the ability of the government to formulate and implement sound policies and regulations that permit and promote private sector development. Although it should indirectly capture the quality of a government’s regulation of foreign aid and non-state services, it is a poor substitute for a measure of the quality of a country’s regulation of foreign aid and non-state service provision. 7 Results I estimate the model described in Equation 1 and examine the extent to which legitimating beliefs are related to government performance and administra- tive competence, as well as procedural justice. I then test whether including indicators of the provision of services by non-state actors and donors affect the relationships between government effectiveness, procedural justice and legitimating beliefs. I conclude by examining whether perceptions of donors and non-state actors have consequences for citizens’ willingness to defer to the courts and to the police. Because of the difficulties of directly interpreting multilevel logistic pa- rameters, I focus my discussion on predicted probabilities and �rst differ- ences, or changes in the expected values, of reporting a willingness to defer to the tax department. The point estimates of the �rst differences of re- porting deference to the tax department are shown in Figure 3. Regression results and replication code are available upon request. 16 7.1 Are government performance, administrative com- petence and procedural justice related to legiti- mating beliefs? In a wide range of developing societies in Africa, I �nd considerable evidence of a link between government performance, administrative competence and honesty and procedural justice and citizens’ willingness to defer to govern- mental tax authority. As illustrated by Figure 3, citizens’ beliefs about how fairly their government is treating members of their ethnic group has the largest relative effect on deference to the tax department followed by percep- tions of corruption within the tax administration. The relationship between perceptions of the state’s enforcement capacity among ordinary citizens and the willingness to defer to the tax department is negative. I argue that perceptions of government effectiveness are correlated with the acceptance of a government’s right to make people pay taxes. There are three components to government effectiveness: government performance and administrative competence and honesty. The objective indicators of government performance — the presence of a sewage system and a school in respondents’ neighborhoods — are not signi�cant at the p<0.05.10 There are three possible explanations for the lack of association between the objective indicators of service delivery and the willingness to defer to the tax depart- ment. First, there is reason to believe that better service delivery may affect citizens’ willingness to defer to the tax department only through its effect on improved outcomes that matter for citizens’ livelihoods. Unless improved services and infrastructure have a positive impact on citizens’ welfare, indi- viduals are unlikely to credit the government for these outputs (Sacks and Levi, 2010). The Afrobarometer’s objective measures of service delivery only denote the presence or absence of infrastructure and services. The data do not indicate the condition of the services and infrastructure. Citizens may perceive and reward relative improvements or sanction de- teriorations in services, rather than the absolute level of service quality they receive. If services deteriorate or improve, taxpayers may alter their beliefs about governments’ performance and should attempt to adjust their terms 10 I also tested whether there is a relationship between the presence of a concrete road, health clinic, post office and electricity grid in the enumeration areas and respondents’ willingness to pay taxes. None of these objective indicators except for the presence of an electricity grid were signi�cant at the p<0.05 level. The presence of an electricity grid is negatively associated with the willingness to defer to the tax department. 17 of trade with government. Because the dataset used for this study is cross- sectional rather than longitudinal, I am unable to examine whether citizens adjust their beliefs and behavior to relative changes in service delivery. I was only able to test whether there are associations between the absolute service quality across neighborhoods, countries and ethnic groups and deference to the tax department, police and courts. Further, citizens may not be attributing goods and services to the gov- ernment. Rather, citizens may be attributing goods and services, such as roads, electricity grids, sewage systems, health care and education to vari- ous non-state actors including the following: the private sector; NGOs and community-based groups; churches, mosques and other religious institutions; traditional leaders; and, bilateral and multilateral donors. Survey questions on the Afrobarometer only indicate the presence or absence of services and infrastructure, and the quality of these services, but these questions do not probe respondents on who they believe are providing these services. Each of the indicators of perceptions of government performance is sig- ni�cant at the p<0.05 level. Food security is positively associated with a willingness to defer to the tax department. The difference in the predicted probability of deferring to the tax department for respondents who have never experienced food insecurity and for responden fts who have experienced food security at least once within the past year is 4 percentage points. A dis- inclination to pay taxes on the part of the hungry may reflect an inability to pay. It may also reflect a belief that government is not ful�lling its end of the �scal contract. Satisfaction with government’s maintenance of water and sewage corresponds to a percentage point increase of 3.4 in the predicted probability that one will defer to the tax department. Similarly, satisfaction with government efforts to maintain roads and bridges corresponds to a per- centage point increase of 3.0 in the predicted probability that one will defer to the tax department I �nd support for the hypothesis that citizens’ assessment of adminis- trative competence is positively related to their acceptance of government’s right to make people pay taxes. There are two dimensions to administrative competence: the honesty of state agents; and the extent to which the bureau- cracy can monitor and enforce tax regulations. I do not �nd any evidence of a relationship between citizens’ perceptions of the transparency of their local council in terms of providing information about its revenue and expenditures and their willingness to defer to the tax department. A perception that tax collectors are honest rather than corrupt corresponds to a percentage point 18 increase of 8.1 in the predicted probability that one will defer to the tax department. Analyses point to a positive relationship between a belief that the bureau- cracy is competent and able to deliver promised services and a willingness to defer to the tax department. The indicator of citizens’ evaluations of how well their local government is handling the collection of license fees is sig- ni�cant at the p<0.05 level. Approval of local government’s handling of the collection of license fees corresponds to an extra 6.4 percentage points in the predicted probability of deferring to the tax department. The relationship between con�dence in the state’s enforcement capacity and the willingness to defer to the tax department is mixed. A belief that the government is likely to enforce tax regulations among government officials does not appear to be correlated with citizens’ willingness to defer to the tax department. Also, �ndings point to a statistically signi�cant but negative relationship between a belief that the government is likely to enforce tax reg- ulations among citizens and willingness to defer to the tax department. A belief that the government is likely to enforce tax regulations among citizens corresponds to a percentage point decrease of 5.3 in the predicted probabil- ity of deferring to the tax department. One explanation for this somewhat counter-intuitive �nding is the following. Although citizens might believe that the state is likely to enforce the law among citizens, they may also view this enforcement as unfair, overly harsh or arbitrary. If citizens believe that the government is selectively enforcing the law among members of a partic- ular political party, ethnic group, or region, then they may be less likely to be willing to defer to the tax department. As I expected, citizens’ judgments of the fairness in government proce- dures and decisions is associated with their willingness to defer to the tax department. A perception of a government’s treatment of one’s own ethnic group as fair translates into a predicted probability of 8.1 percentage points that one will be willing to defer to the tax department. A belief that the government treats the populace fairly, either all or most of the time, corre- sponds to a percentage point increase of 4.1 in the predicted probability that one will be willing to defer to the tax department. Overall, I �nd a large relative effect of holding an all-around positive assessment of government on the predicted probability that individuals will grant their government the right to make people pay taxes. The combined effect of a government that is both fair and effective corresponds to an extra 31.7 percentage points in the predicted probability that an individual will 19 accept government’s right to make people pay taxes. These results suggest that the more effective and fair the government, the more likely that its population will accept the government’s right to make people pay taxes, my indicator for legitimating beliefs. With the available African data, I cannot identify how respondents’ beliefs about taxation translate into actual compliance behavior. 7.1.1 Control Variables Although I control for several socio-demographic variables, very few are sig- ni�cant at the p<0.05 level. Gender, age, ownership of a radio and television, use of the internet, and employment status are not signi�cant at the p<0.05 level. Respondents who live in an urban rather than a rural area are more likely to be willing to defer to the tax department (p<0.05), as are respon- dents who use a mobile phone every day (p<0.05). Respondents who have experience with paying property taxes or fees (p<0.001) are also more likely to be willing to defer to the tax department. Yet, there is no association between experience with paying income taxes and the willingness to defer to the tax department. Although including random intercepts for ethnicity im- proves the model’s �t, the intra-class correlation for ethnicity is quite small. Ethnic affiliation only explains 1 percent of the variation in deference to the tax department. Another step in the quantitative analysis is to test whether including country-level indicators improves the model’s �t beyond individual socio- economic characteristics and perceptions of government effectiveness and fairness. I estimate coefficients for each of the following indicators for 2007: GDP per capita; population size; government consumption as a share of GDP; Freedom Houses measures of civil liberties and political rights; and World Bank Governance Indicators, government effectiveness and corrup- tion. Not one of these indicators is signi�cant at the p<0.05 level when I include the variables in the equation together as a group or when I include them separately. The coefficient estimates for the individual-level explana- tory variables barely change when I include these country-level indicators. The most likely explanation for why I am not �nding any robust country- level effects is that there is more variation in attitudes within countries than between countries. The intra-class correlation coefficient for the country level is only 0.03. Thus, only 3 percent of the variation in deference to the tax department can be attributed to country-level factors. By contrast, 20 the intraclass correlation coefficient for the PSU level is 0.14. I attribute this �nding to within-country variation in perceptions of government effectiveness and procedural justice. To the extent that governments favor certain groups or regions of a country in their provision of public goods, while targeting other groups with repression or discrimination, I would expect to �nd a large amount of within country variation in the willingness to defer to government authority. Even countries with high government effectiveness scores may have certain sub-populations who perceive the government to be unfair or incompetent. 7.2 Does the provision of services by non-state actors and donors affect the �scal contract between citi- zens and the state? The likelihood ratio test of model �t and the AIC suggest that including the measures of donor and non-actor provision of services in model 2 is an improvement over model 1. I also present separation plots (Greenhill, Ward and Sacks, 2011), as a visual method for assessing the predictive power of the regression results. The plots allow us to evaluate model �t based upon the models’ ability to assign high-probabilities to actual occurrences of the event of interest (e.g. willingness to defer to the tax department) and low- probabilities to non-occurrences of the event of interest (e.g. unwillingness to defer to the tax department). The plots separate the events and non-events into two separate plots. In these plots, the dark and light panels correspond to the actual instances of the events and non-events, respectively. The colors correspond to ranges of probabilities; the darkest shade of red corresponds to a probability of 0.8 and higher and the lightest shade of red corresponds to a probability of 0.1 and lower. A perfect model would produce a plot with complete separation between the events and non-events. All the events would be in the upper band of the plot, which would be represented in the darkest shades of red. All the non-events would be in the lower band of the plot, which would be represented in the lightest shades of red. The separation plots suggest that model 2 is a slightly better performing model than model 1 (see Figure 4). A comparison of Figure 3 and Figure 5 demonstrates that the effects of government effectiveness and procedural justice do not change when I include measures of donor and non-state actor service provision. I include the perception-based indicators of donor and non-state actor 21 service provision in separate models due to high correlations among the var- ious indicators.11 Figure 7 illustrates the effect of donor and non-state actor service provision on citizens’ willingness to defer to the tax department. With one exception, analyses point to a positive relationship between donor and non-state actor service provision and the willingness to defer to the tax de- partment. Respondents who believe that traditional actors are responsible for collecting taxes are less likely to be willing to accept the authority of the tax department than respondents who believe that the government is responsible for collecting taxes. I expected respondents who believe that non-state actors, rather than the state, are responsible for collecting taxes to be less likely to be willing to defer to the tax department. The results suggest that this is the case. A belief that traditional leaders instead of the central or local government are responsible for collecting taxes corresponds to a percentage point increase of 5.8 that one will be unwilling to defer to the tax department. I argue that non-state actor and donor service provision is likely to un- dermine the �scal contract between a state and citizens by providing citizens with essential goods and services that they believe the state has an obligation to provide. The analysis of the Afrobarometer data does not yield support for this hypothesis. Instead, I �nd a positive relationship between donor and non-state service provision and deference to the tax department. In other words, citizens who believe that donors and non-state actors are helping their country by delivering essential goods and services are more likely to be will- ing to defer to the tax department. This positive relationship is consistent across all of my measures of donor and non-state actor service provision.12 Of the indicators of donor and non-state actor service provision, the mea- sure of service provision by the UN has the largest relative effect on the acceptance of the tax department’s right to make people pay taxes. A belief that the UN is doing a lot, rather than a little, to help one’s country cor- responds to a percentage point increase of 8.4 in the predicted probability of deferring to the tax department. A belief that international donors and NGOs are doing a lot to help one’s country corresponds to a percentage point increase of 7.2 in the predicted probability of deferring to the tax department. 11 These regression tables are available upon request. 12 Using different survey data collected in Zambia, I �nd a negative association between donor and non-state actor service provision and reported compliance. With the exception of church provision of welfare, I do not �nd any association between donor and non-state actors service provision and legitimating beliefs. 22 Further, a belief that the US or China is doing a lot to help one’s country corresponds to a percentage point increase of 6.8 and 7.3, respectively, in the predicted probability of deferring to the tax department. These �ndings support my hypothesis of the possible positive spill-over effects of donors and non-state actor service provision on citizens’ percep- tions of the state. If the government is able to signal to citizens that it is playing a key role in effectively leveraging and regulating external resources, then citizens are likely to accord the government credit for non-state ser- vice provision. I test this alternative hypothesis by including a measure of a country’s regulatory quality in the model and examine whether this variable mediates the relationship between citizens’ perception of the helpfulness of donors and their willingness to defer to the government. The indicator of regulator quality is not signi�cant at the p<0.05 and including the variable does not affect any of the coefficient estimates. Do perceptions of donor and non-state actor helpfulness and influence vary according to respondents’ wealth? An alternative hypothesis is that very poor citizens, who have little experience with paying taxes, might not care from whom they receive services and goods as long at they help to improve their welfare. Poor citizens might credit the government for any resources they receive regardless of whether the actor delivering the service is a state, non-state or donor agency. By contrast, wealthier citizens, who likely have a higher tax burden, may be less willing to give credit to a government that is not directly providing them with goods and services to which they believe they are entitled. I test this alternative hypothesis by including two interaction variables in the model. The �rst is an interaction term between experience with paying property fees or taxes in the last year and perceptions of the helpfulness of donors and non-state actors. The second is an interaction term between experience with food insecurity in the past year and perceptions of the helpfulness of donors and non-state actors for the country. Neither of these variables are signi�cant at the p<0.05. I also argue that citizens who believe that donors and non-state actors exert too little influence over the government are less likely to be willing to defer to their government than citizens who believe that donors and non-state actors exert too much influence over their government. My �ndings support this argument. A belief that international donors and NGOs have too little influence, rather than too much influence, corresponds to a percentage point increase of 4.4 in the predicted probability that one will be unwilling to de- 23 fer to the tax department.13 A belief that international donors and NGOs have just the right amount of influence, rather than too little influence, cor- responds to a percentage point increase of 2.8 in the predicted probability that one will defer to the tax department. Analyses point to a similar ef- fect of domestic civic groups and NGOs on citizens’ willingness to defer to the tax department. The predicted probability that one will defer to the tax department is 2.9 and 4.1 percentage points higher for respondents who believe that international businesses and domestic civic groups and NGOs, respectively, have too much influence, rather than too little influence, over their government. This �nding supports either one of two arguments. Citizens may view the influence donors and non-state actors exert over their government positively. Alternatively, it may be the case that citizens who believe that donors and non-state actors exert “too much influence� want to strengthen the state. By quasi-voluntarily complying with government rules and regulations, citizens are increasing the amount of revenue flowing to the state, thereby increasing the strength of the state. Next, I examine whether including country-level indicators of the provi- sion of services by donors and non-state actors, including a measure of the number of NGOs and IGOs present in each of the 19 countries, as well as for- eign aid flows, food aid, and health aid, improves model 2’s �t. None of the country-level measures are signi�cant at the p<0.05 level when I include the variables in the full model. The variable indicating the number of NGOs and IGOs per country is positive but only signi�cant at the p<.01 level. Includ- ing these country-level variables do not affect the predicted probabilities and �rst differences of individual-level measures of donors and non-state actors. To summarize, �ndings from multilevel logistic regressions suggest that the relationships between government effectiveness, procedural justice and deference to the tax department do not change once I account for citizens’ perceptions of donors and non-state actors. There is little evidence that donors and non-state actors are undermining rather than strengthening citi- zens’ willingness to defer to the tax department. 13 Alternatively, a belief that international donors and NGOs have too much influence, rather than too little influence, corresponds to a percentage point increase of 4.4 in the predicted probability that one will defer to the tax department. 24 7.3 Do perceptions of non-state actors and donors have consequences for deference to the police and to the courts? Findings suggest that perceptions of non-state actors and donors have a positive effect on citizens’ willingness to defer to the tax department. Do perceptions of non-state actors and donors also have an effect on citizens’ willingness to defer to the police and to the courts? The Afrobarometer asked respondents whether they agree or disagree with the following: “The police always have the right to make people obey the law� and “The courts have the right to make decisions that people always have to abide by�. I estimate two additional models to test whether perceptions of donor and non-state actor service provision have consequences for citizens’ willingness to defer to the courts and to the police. I include the same control variables as model 1. I also include the same measures of procedural justice, govern- ment performance, and enforcement. I replace the indicator of respondents’ perceptions of corruption among tax officials with an indicator of perceptions of corruption among the police and an indicator of perceptions of corruption among judges and magistrates. I also replace the measure of beliefs about who is responsible for collecting taxes with a measure of beliefs about who is responsible for maintaining law and order in their community. Regression results suggest that perceptions of government performance, administrative competence and honesty, and procedural justice are positively associated with the willingness to defer to the police and to the courts.14 As illustrated in Figure 6 and Figure 8, the effect sizes of procedural justice, government performance and administrative competence and honesty on the willingness to defer to the courts and to the police are smaller than the effect sizes of these measures on the willingness to defer to the tax department. The results point to a positive association between some measures of donor and non-state service provision and deference to police and to courts. Citizens who believe that donors and non-state actors exert too much influence, rather than too little influence, are more likely to be willing to defer to the courts and the police. Yet, I do not �nd any relationship between perceptions of the helpfulness of donors and non-state actors and the willingness to defer to the courts and to the police. 14 Regression results for each of the different model speci�cations are available upon request. 25 Similar to my �ndings from model 1 and model 2, I do not �nd any asso- ciation between the objective indictors of government performance and the willingness to defer to the police and to the courts. Except for the measure of food security, each of the indicators of government performance is signi�cant at the p<0.05 level. A positive evaluation of government performance in maintaining water and sewage corresponds to an extra 1.7 and 4.2 percent- age points in the predicted probability of granting courts the right to make binding decisions and the police the right to impose the law, respectively. Similarly, a positive evaluation of government performance in terms of its maintenance of roads corresponds to an extra 4 and 3.4 percentage points in the predicted probability of granting courts the right to make binding decisions and the police the right to impose the law, respectively. Also, consistent with my �ndings from model 1 and model 2, �ndings point to an association between perceptions of administrative competence and a willingness to defer to the police and to the courts. Analyses suggest that the relationship between perceptions of government enforcement of the law among citizens who violate it and a willingness to defer to the courts and to the police is negative. Speci�cally, a belief that the government is likely to enforce the law among ordinary citizens corresponds to a percent- age point decrease of 6.1 and 4.5 in the predicted probability that a citizen will be willing to defer to the courts and to the police, respectively. The predicted probability of deferring to the courts and to the police is 3.2 and 3.9 percentage points higher for respondents who view the state bureaucracy as competent in its collection of fees and licenses. Perceptions of adminis- trative honesty are strongly correlated with citizens’ willingness to defer to courts and to police. A belief that all judges and magistrates and police are honest is associated with a percentage point increase of 14.2 and 10.5 in the predicted probability of deferring to the courts and to the police, respectively. Similar to my �ndings from model 1 and model 2, citizens’ judgments of the fairness in government procedures and decisions appear to correlate with their beliefs regarding deference to the courts and to the police. A belief that the government treats citizens fairly, either all or most of the time, corresponds to an extra 2.9 percentage points in the predicted probability of deferring to the courts. Perceptions of whether the government treats citizens equally under the law is not correlated with deference to the police. A belief that the government treats members of their own ethnic group fairly either all or most of the time corresponds to an extra 3.9 and 4.8 percentage points in the predicted probability of citizens deferring to the courts and to 26 the police, respectively. Findings point to a negative relationship between beliefs that non-state actors and donors are responsible for providing law and order and the will- ingness to defer to the courts and to the police. Speci�cally, the predicted probability of deferring to the courts and to the police is 3.3 and 3.6 per- centage points higher for respondents who believe that the community is responsible for providing law and order. Respondents who believe that tra- ditional authorities, rather than the state, are responsible for providing law and order are less likely to be willing to defer to the police; the �rst difference is 2.8 percentage points. I do not �nd a relationship between perceptions of the helpfulness of donors and non-state actors and the willingness to defer to the courts and to the police. Yet, results point to a positive relationship between perceptions of the influence that donors and non-state actors exert over one’s government and a willingness to defer to the courts and to the police. A belief that inter- national donors and NGOs exert too much influence, rather than too little influence, over their governments corresponds to a percentage point increase of 2.5 and 4.2 in the predicted probability of deferring to the courts and to the police, respectively. Similarly, the predicted probability of deferring to the courts is 2.2 percentage points higher for respondents who believe that domestic civic organizations and NGOs exert too much, rather than too lit- tle, influence over their government. The predicted probability of deferring to the police is 4.8 percentage points higher for respondents who believe that domestic civic organizations and NGOs exert too much, rather than too lit- tle, influence over their government. The �rst difference of deferring to the police is 4.3 percentage points higher for respondents who believe that inter- national businesses exert too much, rather than too little, influence over the government. To summarize, �ndings suggest that perceptions of procedural justice, government performance, administrative competence, and honesty have con- sequences for citizens’ willingness to defer to the police and to courts. Yet, the effect sizes of perceptions of procedural fairness and government com- petence are not as large in the regression on the willingness to defer to the courts and to the police as in the regression on the willingness to defer to the tax department. The correlations between donor and non-state actor service provision and the willingness to defer to the courts and to the police are also weaker than the correlations between donor and non-state actor service provision and the willingness to defer to the tax department. There does not 27 appear to be a relationship between perceptions of the helpfulness of donors and non-state actors and the willingness to defer to the police and to the courts. Individuals who believe that donors and non-state actors exert too much, rather than, too little influence over one’s government is associated with the willingness to defer to the court and to the police. Findings also suggest that citizens who believe non-state actors are responsible for provid- ing law and order are less likely to be willing to defer to the police and to the courts than respondents who believe the state is responsible for providing law and order. 7.4 Conclusion This paper demonstrates that the logic of the �scal contract is relevant to a wide variety of contemporary African states. Findings from a cross-national analysis of survey data from Africa link citizens’ legitimating beliefs — in- dicated by a willingness to defer to the tax department, the police and the courts — to a government’s ful�llment of a �scal contract. Citizens who are satis�ed with their government’s provision of services and goods are more likely to be willing to defer to the tax department, courts and police than citizens who disapprove of government service provision. There is also evi- dence that the more competent, honest and fair a government is in serving its citizens, the more likely it is to produce a virtuous cycle of citizen compliance with its extractions, leading to better government and greater compliance. There is little evidence that the presence of infrastructure in neighborhoods is itself linked to citizens’ legitimating beliefs. In Sub-Saharan Africa, NGOs, donors, churches, traditional leaders, and other non-state actors provide a signi�cant portion of essential goods and ser- vices. The provision of services by non-state actors and donors is not unique to Africa and occurs throughout the world in both OECD and non-OECD countries. The growth of non-state actor service provision has sparked a debate in political science and sociology and in the aid community over the conditions under which non-state actors are likely to affect citizens/state re- lationships (Bratton and van de Walle, 1997; Brown, 1998; Cammett and Lynch, 2008; Moore, 2007; Smith and Lipsky, 1993; White, 1999). I offer some preliminary answers as to whether donors and non-state actors are un- dermining or strengthening citizen/state relationships. Findings suggest that across a wide range of African countries, including fragile states like Liberia and stronger states like Botswana and South Africa, donors and non-state 28 actors are strengthening, rather than undermining, citizens’ legitimating be- liefs, as measured by their willingness to defer to the tax department, the police and the courts. Citizens who believe that donors and non-state actors, including domestic and international NGOs and international businesses, are doing a lot to help their country, rather than a little, are more likely to be willing to defer to the tax department. People who perceive that donors and non-state actors exert too little, rather than too much, influence over their government, are less likely to be willing to defer to the tax department, po- lice and courts. The opposite is true for those who perceive that donors and non-state actors exert too much influence, rather that too little influence, over their government. The absence of behavioral and longitudinal data on compliance limits the conclusions that can be drawn from this study. We know very little about how citizens’ reported willingness to comply with taxes, the police and the courts corresponds to their actual compliance behavior. We know very little about how changes in service delivery over time affect citizens’ legitimating beliefs. We are also lacking a measure of the quality of governments’ regulation of foreign aid and non-state actor resources. Yet, this paper offers one of the few empirical tests of whether citizens’ perceptions of non-state actors and donors are associated with their beliefs about their government’s legitimacy. 29 8 Figures Government Performance Legitimating Compliance Beliefs Figure 1: There is a large literature that examines the relationship between the �scal contract, or how government extractions contributes to the emergence of responsive, accountable and representative governments. Non-State Actor Provision of Services Government Performance Legitimating Beliefs Compliance Figure 2: The focus of this paper is on the impact of non-state actor provision of services on the “�scal contract� between citizens and the government. 30 First difference of deferring to the Tax Department −0.1 −0.05 0 0.05 0.1 0.15 0.2 Gov't treatment of ethnic group: fair q No corruption within tax administration q Approve gov't efforts to collect fees q Gov't treatment of citizens: fair q Food security q Approve gov't efforts to maintain water q & sewage Approve gov't efforts to maintain roads q State enforcement of law among citizens q −0.1 −0.05 0 0.05 0.1 0.15 0.2 Probability Figure 3: This �gure graphically illustrates the point estimates of the �rst dif- ferences of deferring to the tax department. The horizontal bars in the �gure indicate 95 percent con�dence intervals. The results suggest that procedural fair- ness, speci�cally, citizens’ perceptions of the government’s treatment of their ethnic group, has the largest relative effect on deference to the tax department followed by administrative competence and government performance in delivering services. 31 Accept authority of tax department (n=15991) Probabilities: over 0.8 0.7 − 0.8 0.6 − 0.7 0.5 − 0.6 0.4 − 0.5 Do not accept authority of tax department (n=8700) 0.3 − 0.4 0.2 − 0.3 0.1 − 0.2 under 0.1 (a) Model 1: This model includes perceptions-based indicators of gov- ernment service delivery, bureaucratic competence and honest and pro- cedural fairness. Accept authority of tax department (n=15991) Probabilities: over 0.8 0.7 − 0.8 0.6 − 0.7 0.5 − 0.6 0.4 − 0.5 Do not accept authority of tax department (n=8700) 0.3 − 0.4 0.2 − 0.3 0.1 − 0.2 under 0.1 (b) Model 2: This model also includes perception-based indicators of donor and non-state actor service provision. Figure 4: The separation plot suggests that both models perform well in assigning high probabilities to actual cases of accepting the authority of the tax department and low probability to actual cases of rejecting the authority of the tax department. The darker red bands, which represent the high probabilities of the event occurring, are wider on the upper deck of the plot (which consists of the actual events) than in the lower deck (which consists of the non-events). The grey bands, which correspond to low probabilities of the event are wider in the lower deck than they are in the upper deck. 32 First difference of deferring to the Tax Department with measures of donors and non−state actors −0.1 −0.05 0 0.05 0.1 0.15 0.2 No corruption within tax administration q Gov't treatment of ethnic group: fair q Approve gov't efforts to collect fees q Gov't treatment of citizens: fair q Food security q Approve gov't efforts to maintain water q & sewage Approve gov't efforts to maintain roads q State enforcement of law among citizens q −0.1 −0.05 0 0.05 0.1 0.15 0.2 Probability Figure 5: This �gure graphically illustrates the point estimates of the �rst differ- ences of deferring to the tax department once we take into account perceptions of donors and non-state actors. Measures of non-state actor and donor provision of services do not signi�cantly affect the relationships between procedural justice, administrative competence government performance and legitimating beliefs. 33 The effect of donor & non−state actor service provision on deferring to Courts −0.1 −0.05 0 0.05 0.1 0.15 0.2 No corruption among judges q Approve gov't efforts to maintain roads q Gov't treatment of ethnic group: fair q Approve gov't efforts to collect fees q Gov't treatment of citizens: fair q Int'l donors/NGOs: too much influence q Domestic orgs./NGOs: too much influence q Approve gov't efforts to maintain water q & sewage Community responsible for law & order q State enforcement of law among citizens q −0.1 −0.05 0 0.05 0.1 0.15 0.2 Probability Figure 6: This �gure illustrates the �rst differences of the willingness to accept the authority of the courts. 34 The effect of donor & non−state actor service provision on deferring to the Tax Department −0.1 −0.05 0 0.05 0.1 0.15 0.2 UN: helps a lot q China: helps a lot q Int'l donors/NGOs: help a lot q US: helps a lot q Int'l donors/NGOs: q too much influence Domestic orgs./NGOs: q too much influence Int'l businesses: q too much influence Int'l donors/NGOs: q right amount of influence Traditional authorities q responsible for taxes −0.1 −0.05 0 0.05 0.1 0.15 0.2 Probability Figure 7: This �gure illustrates the effect of donor and non-state actor service provision on deferring to the tax department. I model the effect of each of these indicators of donor and non-state actor service provision on deference to the tax department in separate models. Service provision by the UN followed by China has the largest relative effects. 35 The effect of donor & non−state actor service provision on deferring to police −0.1 −0.05 0 0.05 0.1 0.15 0.2 No corruption among police q Gov't treatment of ethnic group: fair q Domestic orgs./NGOs: q too much influence Int'l businesses: q too much influence Int'l donors/NGOs: q too much influence Approve gov't efforts to maintain water q & sewage Approve gov't efforts to collect fees q Approve gov't efforts to maintain roads q Traditional leaders responsible for q law & order Community responsible for law & order q State enforcement of law among citizens q −0.1 −0.05 0 0.05 0.1 0.15 0.2 Probability Figure 8: This �gure illustrates the �rst differences of the willingness to accept the authority of the police. 36 9 Tables Table 1: Afrobarometer Round 4 (2008) Country Number of Respondents Benin 1200 Botswana 1200 Burkina Faso 1200 Cape Verde 1264 Ghana 1200 Kenya 1104 Lesotho 1200 Liberia 1200 Madagascar 1350 Malawi 1200 Mali 1232 Mozambique 1200 Namibia 1200 Nigeria 2324 Senegal 1200 South Africa 2400 Tanzania 1208 Uganda 2431 Zambia 1200 Total N 26513 Table 2: Experiences with paying taxes and fees across Africa (Afrobarometer 2008) Income Tax Property Tax Public Utility Fee Fee for Service License Fee Ghana 38.83 34.00 65.50 58.58 24.08 South Africa 38.63 41.21 63.71 59.29 14.88 Mali 33.36 24.68 23.70 63.23 39.77 Cape Verde 29.21 34.31 81.13 83.99 17.09 Namibia 26.08 28.50 62.92 80.92 23.67 Nigeria 25.60 23.62 55.46 60.71 26.98 Mozambique 24.44 37.90 44.20 72.56 46.83 Kenya 23.07 24.23 42.33 80.43 43.01 Botswana 17.50 13.17 61.92 74.42 9.92 Zambia 17.43 21.85 39.70 77.23 24.52 Liberia 16.92 18.00 14.42 56.25 18.08 Uganda 16.34 19.29 49.01 76.14 35.05 Tanzania 16.31 26.66 36.01 79.88 23.10 Benin 15.00 16.00 47.42 76.67 22.58 Senegal 14.11 12.59 61.37 80.05 16.30 Madagascar 13.19 31.93 36.30 75.19 38.22 Malawi 11.10 7.77 21.20 38.68 18.86 Burkina Faso 10.61 13.36 32.92 81.92 34.36 Lesotho 8.42 3.83 28.92 44.42 3.00 37 Table 3: First differences of donor and non-state actor service provision on will- ingness to accept the tax department’s authority First Difference Lower 95 % C.I. Upper 95 % C.I. UN: helps a lot vs. not at all 0.08 0.04 0.12 China: helps a lot vs. not at all 0.07 0.04 0.11 International donors/NGOs: help a lot vs. not at all 0.07 0.03 0.11 US: helps a lot vs. not at all 0.07 0.04 0.10 International donors/NGOs: too much influence vs. too little influence 0.05 0.03 0.07 Domestic organizations/NGOs: too much influence vs. too little influence 0.04 0.02 0.06 International businesses: too much influence vs. too little influence 0.03 0.01 0.05 International donors/NGOs: right amount influence vs. too little influence 0.03 0.01 0.05 Traditional authorities responsible for taxes vs. state responsible -0.06 -0.10 -0.02 Table 4: First differences of donor and non-state actor service provision on will- ingness to accept authority of the courts First Difference Lower 95 % C.I. Upper 95 % C.I. No corruption among judges vs. a lot of corruption 0.14 0.08 0.19 Approve government efforts to maintain roads vs. disapprove 0.04 0.02 0.06 Government treatment of ethnic group: fair vs. unfair 0.04 0.01 0.07 Approve government efforts to collect fees vs. disapprove 0.03 0.01 0.05 Gov’t treatment of citizens: fair vs. unfair 0.03 0.00 0.05 International donors/NGOs: too much influence vs. too little 0.03 0.01 0.05 Domestic organizations/NGOs: too much influence vs. too little 0.02 0.00 0.04 Approve government efforts to maintain water and sewage vs. disapprove 0.02 0.00 0.03 Community responsible for law and order vs. state responsible -0.03 -0.06 0.00 State enforcement of law among citizens likely vs. unlikely -0.06 -0.08 -0.04 Table 5: First differences of donor and non-state actor service provision on will- ingness to accept authority of the police First Difference Lower 95 % C.I. Upper 95 % C.I. No corruption among police vs. a lot of corruption 0.11 0.05 0.16 Gov’t treatment of ethnic group: fair vs. unfair 0.05 0.02 0.08 Domestic organizations/NGOs: too much influence vs. too little 0.05 0.02 0.08 International businesses: too much influence vs. too little 0.04 0.02 0.07 International donors/NGOs: too much influence vs. too little 0.04 0.02 0.07 Approve Gov’t efforts to maintain water and sewage vs. disapprove 0.04 0.02 0.06 Approve Gov’t efforts to collect fees vs. disapprove 0.04 0.02 0.06 Approve Gov’t efforts to maintain roads vs. disapprove 0.03 0.02 0.05 Traditional leaders responsible for law and order vs. disapprove -0.03 -0.06 0.00 Community responsible for law and order vs. state -0.04 -0.07 -0.01 State enforcement of law among citizens likely vs. unlikely -0.05 -0.07 -0.02 38 References Adsera, Alicia and Charles Boix. 2002. “Trade, Democracy, and the Size of the Public Sector: The Political Underpinnings of Openness.� Interna- tional Organization 56(2):229–262. Afrobarometer. 2005. “Round 3.�. URL: http//wwww.afrobarometer.org Alesina, Alberto and Romain Wacziarg. 1998. “Openness, country size and government.� Journal of Public Economics 69:305–321. Azam, Jean-Paul. 2001. “The Redistributive State and Con�cts in Africa.� Journal of Peace Research 38(4):429–444. Bates, Robert H. and Da-Hsiang Donald Lien. 1985. “A Note on Taxa- tion, Development and Representative Government.� Politics and Society 14(1):53–70. Bergman, Marcelo. 2003. “Tax Reforms and Tax Compliance: The Diver- gent Paths of Chile and Argentina.� Journal of Latin American Studies 35(3):593–624. Boix, Carles. 2003. Democracy and Redistribution. Cambridge, U.K.: Cam- bridge University Press. Bratton, Michael and Nicholas van de Walle. 1997. Democratic Experiments in Africa: Regime Transitions in Comparative Perspective. Cambridge, U.K. ; New York: Cambridge University Press. Brautigam, Deborah. 2000. Aid Dependence and Governance. Stockholm: Almqvist and Wiksell International. Brown, L. David. 1998. Creating Social Capital: Nongovernment Develop- ment Organizations and Intersectoral Problem Solving. In Private Action and Public Goods, ed. Walter W. Powell and Elizabeth S. Clemens. New Haven, CT: Yale University Press. Burnside, Craig and David Dollar. 2000. “Aid, policies, and growth.� Amer- ican Economic Review 90(4):847–868. 39 Burnside, Craig and David Dollar. 2004. “Aid, policies, and growth: Reply.� American Economic Review 94(3):781–785. Cammett, Melani and Julia Lynch. 2008. “Providers and Politics: The Effect of Private Healthcare on Citizen Trust in Government in Western and Eastern Europe.� Prepared for presentation at the Annual Meeting of the American Political Science Association. Chen, Daniel L. 2004. Club goods and group identity: Evidence from Is- lamic resurgence during the Indonesian �nancial crisis. Department of Economics, MIT. Chowdury, Alamgir Farouk, Simon Delay, Naushad Faiz, Iftekher Haider, Brian Reed, Pauline Rose and Priti Dave Sen. 2004. Bangladesh: Study of Non-State Providers of Basic Services. Birmingham, UK: University of Birmingham, International Development Department. Crant, J. Michael and Thomas S. Bateman. 1993. “Assignment of Credit and Blame for Performance Outcomes.� The Academy of Management Journal 36(1):7–27. Devarajan, Shantayanan and Ritva Reinikka. 2004. Making Services Work for Poor People. New York: Oxford University for the World Bank. Edwards, Michael. 2004. Civil Society. Malden, MA: Polity Press. Edwards, Michael and David Hulme. 1996. “Too close for comfort? The im- pact of official aid on nongovernmental organizations.� World Development 24(6):961–973. Fjeldstad, Odd-Helge and Joseph Semboja. 2000. “Dilemmas of Fiscal Decen- tralisation: A Study of Local Government Taxation in Tanzania.� Forum for Development Studies 27(1):7–41. Fjeldstad, Odd-Helge and Joseph Semboja. 2001. “Why People Pay Taxes: The Case of the Development Levy in Tanzania.� World Development 29(12):2059–2074. Fjeldstad, Odd-Helge and Lise Rakner. 2003. “Taxation and tax reforms in developing countries: Illustrations from sub-Saharan Africa.� Chr. Michelsen Institute (CMI), Bergen, Norway. 40 URL: http://www.cmi.no/publications/�le/?1551=taxation-and-tax- reforms-in-developing-countries Fritsch, Peter. October 2, 2001. “Lesson Plan: Religious Schools in Pakistan Fill Void-and Spawn Warriors.� Wall Street Journal A1, A24. Gibson, Clark and Barak Hoffman. 2005. “Fiscal Governance and Public Services: Evidence from Tanzania and Zambia.� Department of Political Science, University of California, San Diego. Greenhill, Brian, Michael Ward and Audrey Sacks. 2011. “The Separation Plot: A New Visual Method for Evaluating the Fit of Binary Models.� American Journal of Political Science 55(4):991–1002. Guyer, Jane. 1992. “Representation Without Taxation: an Essay on Democ- racy in Rural Nigeria, 1952-1990.� African Studies Review 35(1):41–79. Hoffman, Barak. 2007. “Assessing the Quality of Local Government in South Africa.� Center on Democracy, Development and Rule of Law, Stanford University. Iannaccone, Laurence R. and Eli Berman. 2006. “Religious extremism: The good, the bad, and the deadly.� Public Choice 128(1-2):109–129. Joshi, Anuradha and Mick Moore. 2004. “Institutionalized Co-production: Unorthodox Public Service Delivery in Challenging Environment.� Journal of Development Studies 40(4):31–49. Kadzamira, Esme, Dominique Moran, Jo Mulligan, Nebert Ndirenda, Khwima Nthara, Brian Reed and Pauline Rose. 2004. Malawi: Study of Non-State Providers of Basic Services. Birmingham, UK: University of Birmingham, International Development Department. Kaldor, Nicholas. 1963. “Will Underdeveloped Countries Learn to Tax?� Foreign Affairs 41(2):410–419. Kasara, Kimuli. 2007. “Tax Me If You Can: Ethnic Geography, Democracy, and the Taxation of Agriculture in Africa.� American Political Science Review 101(1):159–172. 41 Kaufmann, Daniel, Aart Kraay and Massimo Mastruzzi. 2006. “Governance Matters VI: Aggregate and Individual Governance Indicators 1996-2006.�. URL: https://openknowledge.worldbank.org/handle/10986/7473 Kelly, Janet M. and David Swindell. 2003. “Service Quality Variation Across Urban Space: First Steps Toward a Model of Citizen Satisfaction.� Journal of Urban Affairs 25(3-4):271–288. Knack, Stephen. 2001. “Aid Dependence and the Quality of Governance: Cross-Country Empirical Tests.� Southern Economic Journal 68(2):310– 329. Kosack, Stephen. 2003. “Effective aid: How democracy allows development aid to improve the quality of life.� World Development 31(1):1–22. Levi, Margaret. 1988. Of Rule and Revenue. Berkeley: University of Cali- fornia Press. Levi, Margaret. 1997. Consent, Dissent and Patriotism. New York: Cam- bridge University Press. Levi, Margaret and Audrey Sacks. 2009. “Legitimating Beliefs: Concepts and Indicators.� Regulation and Governance 3(4):311–333. Levi, Margaret, Audrey Sacks and Tom R. Tyler. 2009. “Conceptualizing Le- gitimacy, Measuring Legitimating Beliefs.� American Behavioral Scientist 53(3):354–375. Lieberman, Evan. 2001. “National political community and the politics of income taxation in Brazil and South Africa in the twentieth century.� Politics & Society 29(4):515–555. Lieberman, Evan. 2003. Race and Regionalism in the Politics of Taxation in Brazil and South Africa. Cambridge: Cambridge University Press. Lipsky, Michael. 1980. Street-Level Bureaucracy: Dilemmas of the Individual in Public Services. New York: Russell Sage Foundation. Mansuri, Ghazala and Vijayendra Rao. 2004. “Community - Based and - Driven Development: A Critical Review.� The World Bank Research Observer 19(1):1–39. 42 Marcussen, Henrik Secher. 1996. “NGOs, the State and Civil Society.� Review of African Political Economy 23(69):405–423. Moore, Mick. 2004. “Revenues, State Formation, and the Quality of Gov- ernance in Developing Countries.� International Political Science Review 25(3):297–319. Moore, Mick. 2007. “How does Taxation Affect the Quality of Governance?� Tax Notes International 47(1):79–98. Morrison, Kevin. 2009. “Oil, non-tax revenue, and the redistributional foun- dations of regime stability.� International Organization 63(Winter):107– 138. Morrison, Kevin M. January 2010. “What Can We Learn About the Resource Curse from the Aid Effectiveness Literature?� Department of Government, Cornell University . Moss, Todd, Gunilla Pettersson and Nicholas van de Walle. 2006. An Aid- Institutions Paradox? A Review Essay on Aid Dependency and State Building in Sub-Saharan Africa. Washington D.C.: Center for Global De- velopment (CGD). North, Douglass C. and Barry R. Weingast. 1989. “Constitutions and Com- mitment: The Evolution of Institutions Governing Public Choice in Sev- enteenth Century England.� Journal of Economic History 49(4):803–832. OECD. 2010. “International Development Statistics Online.�. URL: www.oecd.org/dac/stats/idsonlins Osei, Philip. 2000. “Political Liberalisation and the Implementation of Value Added Tax in Ghana.� Journal of Modern African Studies 38(2):255–278. Palmer, Natasha, Lesley Strong, Abdul Wali and Egbert Sondorp. 2006. “Contracting Out Health Services in Fragile States.� British Medical Jour- nal 332:718–721. Patterson, Amy S. 2006. The Politics of AIDS in Africa. Boulder, Co: Lynne Rienner. Platteau, Jean-Philippe. 2004. “Monitoring Elite Capture in Community- Driven Development.� Development and Change 35(2):223–246. 43 Prichard, Wilson. July 2009. “The Politics of Taxation and Implications for Accountability in Ghana 1981-2008.� IDS Working Paper 330. Rodrik, Dani. 1998. “Why do more open economies have bigger govern- ments?� Journal of Political Economy 106(5):997–1032. Sacks, Audrey and Margaret Levi. 2010. “Effective Government and its Consequences for Social Welfare in sub-Saharan African Countries.� Social Forces 88(5):2325–2351. Smith, Steven Rathgeb and Michael Lipsky. 1993. Nonpro�ts for Hire: The Welfare State in the Age of Contracting. Cambridge, Mass; London, Eng- land: Harvard University Press. Tendler, Judith. 1997. Good Government in the Tropics. Baltimore and London: The Johns Hopkins University Press. Tilly, Charles. 1990. Capital, Coercion, and European States, A.D. 990-1992. Oxford, U.K.: Basil Blackwell. Timberg, Craig. 2005. “Botswana’s Gains Against AIDS Put U.S. Claims to Test.� The Washington Post . Tripp, Aili Mar. 1997. Changing the Rules: The Politics of Liberalisation and the Urban Informal Economy in Tanzania. Berkeley: California University Press. Tyler, Tom R. 1990. Why People Obey the Law. Princeton and Oxford: Princeton University Press. Union of International Associations (UIA). 2005. “International organiza- tions and NGOs.�. URL: http://www.uia.org/statistics/pub.php van de Walle, Nicolas. 2001. African Economies and the Politics of Perma- nent Crisis, 1979-1999. Cambridge, UK: Cambridge University Pres. Weaver, R. Kent. 1986. “The Politics of Blame Avoidance.� Journal of Public Policy 6(4):371–398. Weber, Max. 1968. Economy and Society. Berkeley: University of California Press. 44 White, Sarah C. 1999. “NGOs, Civil Society, and the State in Bangladesh: The Politics of Representing the Poor.� Development and Change 307:307– 332. 45