Policy Research Working Paper 11139 Territorial Inequalities A Note on State Discontinuity Lidia Ceriani Luis F. Lopez-Calva Samuel D. Restrepo-Oyola Poverty and Equity Global Department A verified reproducibility package for this paper is June 2025 available at http://reproducibility.worldbank.org, click here for direct access. Policy Research Working Paper 11139 Abstract When the state does not reach with the same effectiveness determine the density of state presence. By examining the everybody and everywhere, it creates pockets of exclusions discontinuity of the state in 32 developing countries using and hampers the opportunities of individuals belonging extensive data sets from the International Integrated Public to the excluded groups to participate fully economically, Use Microdata Series International and the Armed Con- socially, and politically. This paper develops an index to flict Location and Event Data Project, the paper explores measure state discontinuity, defined as the uneven presence the underlying drivers of this discontinuity and identifies and responsiveness of the state across the territory. Differ- potential policy intervention points. ent layers of capacity, service provision, and responsiveness This paper is a product of the Poverty and Equity Global Department. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The authors may be contacted at lflopezcalva@worldbank.org; srestrepooyola@worldbank.org; and lidia.ceriani@unibo.it. A verified reproducibility package for this paper is available at http://reproducibility.worldbank.org, click here for direct access. RESEA CY LI R CH PO TRANSPARENT ANALYSIS S W R R E O KI P NG PA 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 Territorial Inequalities: A Note on State Discontinuity Lidia Ceriani,* Luis F. Lopez-Calva,† and Samuel D. Restrepo-Oyola‡ Keywords: Social Contract, State Effectiveness, State Density, Inequality Indices. JEL Codes: D31, D63, I30. * Corresponding author. University of Bologna, Department of Political and Social Sciences. Email: lidia.ceriani@unibo.it † The World Bank, Poverty and Equity Global Practice. Email: l�lopezcalva@worldbank.org ‡ The World Bank, Poverty and Equity Global Practice. Email: srestrepooyola@wordbank.org 1 Introduction: The Spatial Dimension of a State “Any large social process or event will inevitably be far more complex than the schemata we can devise, prospectively or retrospectively, to map it.” (Scott, 1998, p. 309) The concept of a state has two fundamental elements: population and territory. The way a group of people is organized within a delimited territory involves a set of rules and policies whose enforcement requires a “government,” as a set of organizational forms that deliver basic services to its members through delegated authority. The capacity of a state is typically de�ined as the effectiveness of its organizational forms to deliver on collectively agreed objectives (World Bank, 2017). As Scott (1998) proposed, the �irst fundamental purpose — an indispensable function— of a state is to make the population “legible” within a speci�ic territory. In his in�luential book “Seeing Like a State”, the author describes his �indings after looking at the evolution of societies in Southeast Asia, concluding that “The more I examined these efforts at sedentarization, the more I came to see them as a state’s attempt to make a society legible, to arrange the population in ways that simpli�ied the classic state functions of taxation, conscription, and prevention of rebellion … I began to see legibility as a central problem of statescraft” (Scott, 1998). Legibility, in turn, can only have an instrumental value as a way to effectively register, exercise control and provide the services societies expect from a government. The state, as an abstract concept involving population and territory, requires a government —a set of organizational forms—whose shape, dynamics, and effectiveness depends on the way power is exercised among its members, and this is what we call governance (World Bank, 2017). Concretely, states play a signi�icant role in modern economies, providing goods and services that can be both collective, in terms of the internalization of their bene�its, like security and law enforcement, and individual, such as healthcare and education (Stiglitz, Sen, & Fitoussi, 2009). Actors in society typically consider the state responsible for providing a range of essential services, which range from infrastructure (e.g., roads, bridges, public transportation systems, water, and sewage systems); education; healthcare; public safety and security; social welfare; environmental protection (e.g., measures to address pollution, conservation efforts, and response to climate change); justice and regulation (e.g., courts, legal aid services, consumer protection, workplace safety); and cultural and recreational facilities; public utilities (e.g., electricity, gas, and telecommunications services). Such services are funded fundamentally by taxation. 2 The extent to which states ful�ill the provision of such services, and even the prioritization of different types of services, varies signi�icantly from country to country, given governance dynamics and its evolution over time. For example, most European OECD countries saw the creation and rapid evolution of the welfare states only in the aftermath of World War II. However, the variability in the extent to which states ful�ill the provision of services also varies signi�icantly within countries. This is where the territorial dimension becomes particularly important. Using O’Donnell’s (2010) de�initions, the state is not necessarily the focal point of a collective identity that transcends social con�licts and divisions. Instead, it can exhibit a discontinuous presence and responsiveness to the needs and interests of citizens across different territories and social groups. When states effectively provide essential services such as education, healthcare, and public safety, it enhances social cohesion, fostering a sense of shared identity and belonging among citizens, and trust and legitimacy in the state itself. This, in turn, reinforces the social compact, which is the implicit agreement between citizens and the state regarding the rights and responsibilities of each party. This creates a self-reinforcing positive governance dynamic: members of society are more likely to comply with laws and pay taxes (enhancing the state’s functions of cooperation and coordination), fostering better societal outcomes. In instances where the state fails to extend its reach with equal strength and ef�icacy to all individuals and regions, it fosters pockets of exclusion within society. This exclusionary dynamic hampers the ability of individuals belonging to these marginalized groups to fully engage and participate in economic, social, and political spheres, weakening outcome- legitimacy. It aggravates grievances and it increases the likelihood of political unrest or insurgency. When the state fails to ensure equitable access to its services and opportunities, it perpetuates inequality and undermines cooperation and social cohesion. There are no well-established indices of state density –presence over a speci�ic territory, which is a pre-condition for effectiveness. The concept of state discontinuity (O’Donell, 2010) has not been implemented in a way that can lead to comparisons across territories. In this paper, we propose to assess the discontinuity of the state on the territory as the inequality of the state’s density in the different geographical areas, where for density we mean the overall potential effectiveness of state intervention in speci�ic policy domains. 2 The State and Public Services Vital public services such as education, healthcare, safe drinking water, and security have been commonly considered in the literature as strong sources of legitimization for states. The main driver is (outcome) legitimacy, namely, the fact that people experience the presence of the state through the provision of services they demand and regard as valuable 3 (Mcloughlin, 2024), and because states historically have played a quintessential role in providing them. Public service delivery can take various forms, depending on the context, goals, and available resources. It may involve direct provision by state agencies, where the state directly manages and delivers services to citizens (e.g., an education system based on a system of public schools owned by the state, where teachers are state employees). Alternatively, states may opt for outsourcing service provision to private or non-pro�it organizations. In some cases, private provision with regulation is employed, where private entities deliver services under strict state oversight and regulation to ensure quality and accountability (e.g., an education system based on private provision with regulations to stipulate the curriculum, textbooks, teachers’ pay and conditions, and general norms and standards). Another model is the public-private partnership (PPP), which combines public sector goals with private sector ef�iciency, allowing for shared investment and risk in the development and management of public infrastructure and services (Barrera-Osorio et al., 2020, provide the analysis of a PPP program in education in Pakistan). The operationalization of service delivery may involve a range of actors, including central state agencies, which often set national policy and allocate funding; local governments, which are responsible for implementing services at the community level; and other specialized actors, such as quasi-public institutions, non-governmental organizations, or private contractors, which may carry out speci�ic tasks or manage certain services under the oversight of the state. Each model offers distinct advantages and challenges, and the choice of approach often depends on factors such as cost ef�iciency, service quality, and political considerations (see Poutvaara & Jordahl, 2020, for a recent literature review on the desirability of outsourcing the provision of public services). However, regardless of the mode of delivery, we posit that the state remains ultimately responsible for ensuring the effective provision of services. This responsibility encompasses guaranteeing that services meet established standards of quality, accessibility, and equity, and that public resources are used ef�iciently. While the state may delegate certain functions or roles to external entities, it retains the oversight and accountability necessary to protect the public interest and ensure that services are delivered in a manner that aligns with societal needs and policy objectives. Thus, when a state reliably and consistently supplies public services, it legitimizes itself (World Bank, 2017). Furthermore, inef�iciencies in providing these services translate into negative consequences for the social welfare of citizens (Cuadrado-Ballesteros, Garcı́a- Sá nchez, & Prado-Lorenzo, 2012) that could foster social malaise and be expressed in the form of demonstrations, riots, and violence in general. Beath, Christia, and Enikolopov 4 (2012), and Khanna and Zimmermann (2017) show that government spending on public services is associated with a signi�icant decrease in insurgent violence —analyzing the cases of Afghanistan and India, respectively— while Justino (2015) shows that government expenditure on social services such as health, education, infrastructure, and welfare decreases both the outbreak and escalation of riots across India. Therefore, public services are pivotal for state legitimacy and people’s well-being. In this paper, we suggest a methodology to build a comparable indicator of state density and apply it with a limited set of available information. The methodology makes use of indicators related to education, health, water and electricity, and security provision as fundamental interventions of the state that allow us to evaluate its lack of homogeneous consistent presence and responsiveness to the interests of all citizens on a given territory. The methodology, however, can be expanded to a broader set of indicators of state functions. Education is a key concern of citizens and has had a long-standing interest in the literature. Mcloughlin (2024) uses the illustrative case of free education in Sri Lanka to explore how the “ideational” properties of public services are constructed and manifested in the political process of state legitimation. The author argues that the relationship between service delivery and states’ legitimacy is not an instrumental equation and warns us that we should pay closer attention to when, how, and why public services carry the underlying normative ideas against which authority is judged. Bereketeab (2020), using Eritrea as an empirical case study, examines the role of education in nation-building in postcolonial Africa. The author argues that leaders during this period, aware of the transformative power of education in the region, exerted signi�icant efforts to convert education into a key element to build the nation, and used it to cultivate a sense of national identity and nationhood. The provision of public health also plays a crucial role in building state legitimacy and is one of the main concerns for citizens. Bondarenko, Krasil'nikova, and Shishkin (2009) carried out a survey of public opinion about equal access to medical services in the Russian Federation and found an unconditional demand expressing that the state should provide medical services to citizens free of charge and believe that providing medical services for a fee is an intolerable violation of people’s rights and unjust. Mobarak, Rajkumar, and Cropper (2011) report that voter preferences and income correlate with the distribution of publicly provided health services across counties in Brazil. Ratigan (2022) studies health care delivery, perceptions of health policy, and state legitimacy in China. The author shows that public health delivery matters to the state’s legitimacy, but this link is conditioned on people’s expectations of the state’s provision of healthcare. Access to piped water and electricity also represent primary concerns for people in low- income and lower-middle-income economies. For instance, when asked about the most pressing problems in their municipalities, Latin American people placed “basic services delivery de�iciencies (such as water, electricity, etc.)” in their top three concerns, only 5 surpassed by those about public security and economic adversities (Latinobaró metro, 2018). These two basic services are often overlooked in high-income economies because their homogeneity across the territory is taken for granted. Nevertheless, the story unfolds differently in low-income economies, where the access to piped water and electricity is heterogeneous, and the quality of these services varies dramatically between the rural and urban sectors. Following these results, security represents broadly common interests for citizens. In Mexico, for instance, the lack of security is considered the main concern for the population: more than 42 percent of respondents in a survey mentioned security-related issues as their main concern (Flores-Macı́as, 2018). Historically, security concerns were at the core of public consent for state capacity building and legitimacy (Besley & Persson, 2009; Dittmar & Meisenzahl, 2020), and today there is a consensus regarding the centrality of public safety as one of the state’s main functions (Flores-Macı́as, 2018; Tilly, 2009). Moreover, when the state loses direct control of vast areas of the regions it is supposed to control, it allows the possibility that armed actors and individuals af�iliated with armed groups engage in criminal activities. Sá nchez de la Sierra (2020) shows how these groups in the Democratic Republic of Congo privately provide essential functions of a state and often gain more legitimacy than the central government. In the former Federally Administered Tribal Areas of Pakistan, state legitimacy is strongly perceived by civilians who have received public services from the state and weakly by those who have been exposed to rebel services (Kubota et al., 2024). Henceforth, security lies at the core of public interest and is considered here because of its importance in the hierarchy of public goods. 3 Framework: An Index of State Density and Discontinuity Let us assume that there are i = 1,2,...,n domains of state intervention, and j = 1,2,...,m geographical subdivisions of the state. In the following, we will �irst de�ine what we mean by effectiveness, density, and discontinuity of the state. 3.1 Effectiveness The effectiveness of the state is de�ined at the local level as the degree of the state’s intervention in a speci�ic policy domain. De�inition 3.1 (Effectiveness). ∈ ℝ represents the state effectiveness in the region j with respect to intervention i, and it is a positive function of the level of reach, ef�icacy, or ef�iciency of the state in that region for that speci�ic intervention. 6 Note that the speci�ic de�inition of depends on the characteristics of the policy domain taken into account, and on data availability. For some areas of intervention, it can be expressed as the percentage of the population successfully reached by a service, for instance, the share of the population living in an area connected to electricity, water, or the sewerage system; the share of immunized children; or the share of households reached by radio or mobile signal. For other domains of state intervention, it can be expressed as the frequency of a service (e.g., how often in the week solid waste is collected), or as the number of service providers per inhabitant (e.g., number of doctors per inhabitant). Other de�initions of state effectiveness may include the number of hours in a day a given service is accessible (e.g., electricity); a measure of the capillarity of the network of streets and railways in the region; and so forth. Ef�icacy allows us to include dimensions that on the other hand can be measured by an evaluation of the quality of the service received (e.g., share of literate individuals, but also infant mortality rates, to cite two). Ef�iciency enables us to incorporate dimensions that can be evaluated based on the extent to which unnecessary resources are minimized in the production of a given output (e.g., some measure of waiting times in accessing health services). The de�inition of state effectiveness is intentionally broad to encompass any area of state intervention relevant to the cases under study. Far from being exhaustive, the list of examples above shows that state effectiveness can be measured in a variety of forms, cardinality, and scales. Nevertheless, since in the following, we will need to aggregate the different domains of state intervention, it is of utmost importance to reduce to the same metric and support the different indicators. For this reason, ̃ . we introduce the concept of normalized effectiveness, ̃ is the normalized state effectiveness in region j De�inition 3.2 (Normalized Effectiveness). with respect to intervention i: ei j − ̃ = (1) − where and are, respectively, the lower and upper bounds of state effectiveness with respect ̃ = 0; on the other hand, if the state reaches to intervention i. If the state fails at being effective, ̃ = 1. the highest possible standard, Note that the de�inition of the normalization boundaries and may be set following an absolute or a relative de�inition. Following the absolute approach, we may believe that there are universal best and worst in the de�inition of state effectiveness in some domain. On the other hand, by embracing the relative approach, we may consider the observed maximum and minimum level of state effectiveness at a particular point in time as benchmarks. Also, 7 we may select different criteria for the de�inition of the lower and upper bounds as a way to normalize the indicator, as is the case for the Human Development Index (UNDP, 2017). For example, for the indicator associated with the health dimension, life expectancy, the lower bound is set to 20 years, which is based on the observation that no country in the 20th century had a life expectancy of less than 20 years (relative approach), while the upper bound of one of the indicators linked to the education dimension, expected years of schooling, is set to 18 years, which is the maximum numbers of education years needed to reach upper tertiary education in most countries (absolute approach). 3.2 Density The density of the state is de�ined at the local level as the overall level of state effectiveness in all policy domains. De�inition 3.3 (Density). ∈ [0,1] is the density of the state in region j, and it is an increasing function of the effectiveness of the state in each intervention in region j: 1 ̃ � � = �� � (2) =1 where wi is the weight attributed to dimension i, s.t. ∑ =1 = 1 , and θ ∈ ℝ regulates the + elasticity of substitution among the different policy domains. Let us consider wi = 1/n for each i = 1,2,...,n: for simplicity of exposition, every dimension counts equally in de�ining the state density. If θ = 1, the state’s density boils down to the unweighted arithmetic mean of the levels of ef�iciency in the different policy areas, which will be considered perfect substitute. For values of θ > 1, high effectiveness in one sector will compensate more and more for low ef�iciency in other sectors. On the other hand, if θ < 1, we will introduce some complementarities between the different areas of intervention, and we will give more and more weight to the area with the least effectiveness. Finally, if θ = 0, dimensions will be considered as perfect complements, and hence the level of density will be determined by the lowest level of ef�iciency in each region. Example 3.1. Take for example a country with two regions, A and B, and two observed policy ̃ 2 = 0 ), (̃ ̃ 1 = 0.4, areas 1 and 2, with state reach ( 1 2 = 0.2,̃ = 0.2), as summarized in the following matrix: 8 Table 1: Example of state reach in two policy areas and two regions If w1 = w2 = 0.5, and θ = 1 two regions would have the same density, since the different policy dimension would be considered perfect substitute, and hence region A would be allowed to compensate inef�iciency in dimension 2 with higher ef�iciency in dimension 1. If w1 = w2 = 0.5, and θ = 2, instead, region A will be considered denser than region B, because the index is giving more weight to the level of ef�iciency reached by A in dimension 1, which is higher than any level of ef�iciency reached by region B. Further, if θ = 0.5, region A will be considered less dense than region B because the index gives more weight to the lowest level of ef�iciency reached by A in dimension 2, which is smaller than any level of ef�iciency reached by region B. The following table summarize the results for different selections of parameters. Table 2: Example of state density in two regions for different choices of parameters 3.3 Discontinuity We can then summarize the information on the state’s density at the sub-national level in an overall index of discontinuity of the state. If the state’s density is the same in all regions, it means that the state has the same level of reach in all the territory, and no region is neglected at the expense of others. On the other hand, if we �ind that the state is denser in some regions than in others, this may be ultimately symptomatic of the political will to exclude speci�ic groups of the population. 9 De�inition 3.4 (Discontinuity). The discontinuity of the state is de�ined as the unequal distribution of the state’s density on the territory, and measured as the average relative density differences among regions, normalized by the average density in the country (µd): 1 � − � = � � (3) ( − 1) 2 =1 =1 Therefore, if all regions have the same density, Dθ = 0; if the state is absent in all regions except one, discontinuity is maximum and Dθ = 1. The discontinuity index de�ined in equation (3) may be understood as a Gini index computed on the density distribution across regions. It can be interpreted as the degree of relative state neglect on the territory. 1 4 An Empirical Application 4.1 Data To assess the discontinuity of the state on the territory, it is essential to use data that is representative at the sub-national level, preferably at the lowest available administrative level. To provide the largest possible coverage of developing countries, we sourced data from the International Integrated Public Use Microdata Series (IPUMS international), and the Armed Con�lict Location and Event Data Project (ACLED). IPUMS International offers the largest collection of publicly available harmonized census samples. It spans 103 countries, with historical data dating back to 1700 (Iceland, 1703). Available data are randomly extracted from original samples, with sample densities ranging from 1 percent to 10 percent of national populations. ACLED collects information on the dates, actors, locations, fatalities, and types of all reported political violence and protest events around the world. We restricted the analysis to the following dimensions: basic services; education; health; and security. These dimensions are frequently referenced in opinion surveys aimed at collecting information on respondents’ perceptions regarding the most pressing issues confronting their country. These challenges are what respondents believe the state should prioritize, thereby offering a reliable indication of the perceived normative functions of the state. For instance, in the International Social Survey Programme, Role of Government V (ISSP 1Similarly to the relative deprivation interpretation of the Gini index computed on the distribution of income in Yitzhaki (1979). 10 Research Group, 2018), 2and Global Barometer Survey (GBS, 2020), 3 the four dimensions are consistently reported within the top �ifteen priorities, even with some order difference across countries. In particular, among African countries, the top reported priority is poverty, followed by taxes; wages, incomes, and salaries; orphans, street children, homeless children; aids; housing; education; drought; corruption; unemployment; political violence; food shortage; water supply; and electricity. Among Asian and South Asian countries, the management of the economy is in the �irst place, followed by unemployment; corruption; wages, incomes, and salaries; poverty; in�lation; crime and security; taxes; housing; social problems (divorce, gambling, drugs); health; education. In Latin America the most pressing issues are crime and public security, followed by unemployment; the status of the economy; corruption; political crisis; in�lation; education; violence, gangs; health; poverty; shortages; drug consumption; low salaries (Authors’ elaboration on data from GBS, 2020). Among ISSP respondents, 77% indicated that the government should allocate more or signi�icantly more funding towards healthcare, while 73% expressed the same sentiment regarding education expenditure, and 52% that the government should spend more or much more on police and law enforcement (Authors’ elaboration on data from ISSP, 2018). 2 ISSP Research Group (2018) focuses on questions about political attitudes and the role of government, and provides nationally representative samples for a set of 35 countries (Australia, Belgium, Chile, Croatia, Czechia, Denmark, Finland, France, Georgia, Germany, Great Britain, Hungary, Iceland, India, Israel, Japan, the Republic of Korea, Latvia, Lithuania, New Zealand, Norway, the Philippines, the Russian Federation, the Slovak Republic, Slovenia, South Africa, Spain, Suriname, Sweden, Switzerland, Thailand, Tü rkiye, the United States, and the Repú blica Bolivariana de Venezuela). It collects information on the preference for more or less government spending in various areas (the environment, health, the police and law enforcement, education, the military and defense, old age pensions, unemployment bene�its, culture and the arts); government’s responsibility (provide a job for everyone, keep prices under control, provide health care for the sick, provide a decent standard of living for the old, provide industry with the help it needs to grow, provide a decent standard of living for the unemployed, reduce income differences between the rich and the poor, give �inancial help to university students from low-income families, provide decent housing for those who cannot afford it, impose strict laws to make industry do less damage to the environment, promote equality between men and women); responsibility for the provision of health care for the sick, care for older people, and school education (government, private companies/for-pro�it organizations, non-pro�it organizations/charities/cooperatives, religious organizations, family relatives or friends). 3 GBS (2020) is the �irst comprehensive effort to measure, at a mass level, the current social, political, and economic climate around the world, and is a collaborative research project comprising six regional Barometers. The last wave available for analysis (wave 3), used in this study, includes information collected from four regions: Africa (Afrobarometer), Asia (Asian and South Asia Barometer), Central and South America (Latinobaró metro), the Middle East (Arab Barometer). In particular, the country coverage is as follows: Japan; Hong Kong SAR, China; the Republic of Korea; Mainland China; Mongolia; the Philippines; Thailand; Indonesia; Singapore; Viet Nam; Cambodia; Malaysia; Myanmar (Asian and South Asia Barometer); Algeria, Benin, Botswana, Burkina Faso, Burundi, Cameroon, Cabo Verde, Cô te d’Ivoire, the Arab Republic of Egypt, Gabon, Ghana, Guinea, Kenya, Lesotho, Liberia, Madagascar, Malawi, Mali, Mauritius, Morocco, Mozambique, Namibia, Niger, Nigeria, Sã o Tomé and Prı́ncipe, Senegal, Sierra Leone, South Africa, Sudan, Eswatini, Tanzania, Togo, Tunisia, Uganda, Zambia, Zimbabwe (Afrobarometer); Algeria, the Arab Republic of Egypt, Jordan, Lebanon, Morocco, the West Bank and Gaza, Tunisia (Arab Barometer); Argentina, Bolivia, Brazil, Chile, Colombia, Costa Rica, Dominican Republic, Ecuador, El Salvador, Guatemala, Honduras, Mexico, Nicaragua, Panama, Paraguay, Peru, Uruguay, the Repú blica Bolivariana de Venezuela (Latinobaró metro). 11 To assess the effectiveness of the state in each of the dimensions in each region, we used the following indicators: (i) the share of households with piped water and (ii) the share of households with electricity, as indicators for the provision of basic services; (iii) the share of literate people (those who can both read and write), as an indicator of education; (iv) the number of children surviving per 1,000 live births, as an indicator of health and (v) the number of days in a year without episodes of violence, as an indicator for security. The �irst four indicators are found in IPUMS, and the last one in ACLED. The upper and lower bounds for the normalization are set at the maximum and minimum level of effectiveness observed in the pooled sample of all regions. Table A1 in the Appendix summarizes the variables used in the analysis. The �inal sample includes 32 countries, and a total of 545 regions, as detailed in Table 3. The majority of administrative divisions included in the analysis belong to Latin American countries (53%), followed by Sub-Saharan African countries (29%), East Asia and Paci�ic (8%), Middle East and North Africa (6%), South Asia (3%), and Europe and Central Asia (2%). The granularity of geographical disaggregation varies by country. In general, it corresponds to the level that would be de�ined as NUTS 2 for European countries (see Eurostat, 2024). For example, in Armenia, the smallest geographical unit analyzed corresponds to a province or the capital city; in Brazil, we use federal districts; in the Islamic Republic of Iran, we analyze provinces; in Peru, departments are considered. In Mali and Togo, the analysis is conducted at the regional level. Finally, it is worth noting that the administrative subdivisions used in this paper re�lect those in place at the time the census data were collected and may not align with current de�initions. Any modi�ication in the de�inition of administrative subdivisions used in the analysis will inevitably affect the overall discontinuity index. Dividing a region into smaller subdivisions is unlikely to reduce discontinuity; instead, it will typically increase variability. On the other hand, merging existing subdivisions into larger aggregations is unlikely to increase overall discontinuity, as it replaces individual densities with an average. However, since changes in subdivisions affect both the numerator and denominator of the index, the direction of the change cannot be predicted in advance. 4.2 Results 4.2.1 Effectiveness The distribution of normalized effectiveness across the 545 regions is summarized in Figure 1. Each region corresponds to one link (the width of the links is proportional to the number of regions �lowing through nodes), and it is colored according to the decile in the overall 12 distribution it belongs to, with higher deciles (in shades of blue) corresponding to higher effectiveness, and lower deciles (in shades of red) corresponding to lower effectiveness. Figure 1 displays two main results. First, the ranking of subdivisions belonging to each world region with respect to each single dimension. Second, the ranking pro�ile of regions across ranks dimensions, which can be stable (a region’s effectiveness belongs to the same decile in all dimensions considered), or erratic (the ranking of a region in terms of effectiveness varies according to the dimension considered). When looking at the distribution of normalized effectiveness across the 545 regions, some patterns emerge (see Figure 1). Sub-Saharan African (SSA) countries are disproportionately represented at the lower end of the effectiveness spectrum. For four of the �ive observed indicators, a region in an SSA country occupies the lowest position: Rwanda in health (in fact, the �ive Rwanda regions are the worst �ive out of the entire sample with respect to child mortality), Mali in education, Sudan in security, and Liberia in basic services, electricity. With respect to basic services, water, South Kordofan in Sudan ranks as the second worst, just ahead of �ive regions in Cambodia. Although regions in Sub-Saharan Africa make up only 29% of total observations, they constitute 91% of those in the lowest effectiveness decile for education, 89% for basic service delivery, electricity, 80% for health, 35% for basic service delivery, water. 13 Table 3: Country included in the analysis and corresponding number of administrative Divisions Number of Share of total Region Code Country Name administrative observations subdivisions Cambodia 24 4.4 EAP Lao PDR 18 3.3 ECA Armenia 11 2 Bolivia 9 1.7 Brazil 27 5 Colombia 29 5.3 Costa Rica 7 1.3 Dominican Republic 32 5.9 Ecuador 24 4.4 El Salvador 14 2.6 LAC Guatemala 22 4 Haiti 10 1.8 Honduras 18 3.3 Mexico 32 5.9 Nicaragua 17 3.1 Peru 25 4.6 Venezuela, RB 24 4.4 MNA Iran, Islamic Republic 30 5.5 SAS Nepal 14 2.6 Benin 12 2.2 Ghana 10 1.8 Lesotho 10 1.8 Liberia 15 2.8 Mali 9 1.7 Mozambique 11 2 SSA Rwanda 6 1.1 Senegal 11 2 Sierra Leone 14 2.6 Sudan 15 2.8 Tanzania 30 5.5 Togo 6 1.1 Zambia 9 1.7 Overall 545 100 Source: Authors’ elaboration on data from the Minnesota Population Center (2020) via IPUMS International, and Raleigh, Kishi, and Linke (2023) via ACLED Regions in Latin American countries (LAC), on the other hand, are overrepresented at the bottom of the effectiveness of states in curbing violence. They are 53% of the total regions in our sample, but make up 80% of the regions in the bottom effectiveness decile for security. In 2023, the regions of Alagoas, Amazonas, Bahia, Minas Gerais, Pará , Pernambuco, Rio de Janeiro, Sã o Paulo in Brazil; Cauca in Colombia; West in Haiti; and Guanajuato, Guerrero, Michoacá n de Ocampo, Nuevo Leó n, Veracruz de Ignacio de la Llave in Mexico experienced at least one episode of violence each day. 14 East Asia and Paci�ic (EAP) regions are the majority (49%) of observations found in the �irst decile of the effectiveness distribution for basic service delivery, water, in spite of representing just 7.7% of total regions in the sample. Some countries display quite homogeneous effectiveness across regions and dimensions. The extreme and only example of its kind is the region of Heredia, in Costa Rica (LAC) which belongs to the 10th decile (best effectiveness) in all dimensions considered (see Appendix, Figure A.1). On the opposite side of the spectrum, the regions of Sé gou in Mali (SSA), and Cabo Delgado in Mozambique (SSA) belong to the worse decile in all dimensions but, respectively, basic services, electricity and basic services, water, where they belong to the second decile (see Appendix, Figures A.2 and A.3). The region of Phnom Penh in Cambodia (EAP), performs above average in all �ive dimensions, being in the 9th decile of the distribution of health and basic services, electricity, in the 7-th decile of the distribution of education and water, and the 6-th decile for security. Similarly, the regions of Aragatsotn, Armavir, and Vayots Dzor in Armenia (ECA) are in the top 30% of the distribution for all �ive indicators (see Appendix, Figure A.4). Other regions display extreme trends. For instance, in Latin America, Rio Grande do Sul, Rio de Janeiro, and Sã o Paulo in Brazil, and Baja California, Distrito Federal, and Nuevo Leó n in Mexico all perform in the top decile with respect to all indicators, with the exception of security, where they belong to the bottom decile (see Appendix, Figure A.5). In the Middle East and North Africa (MNA), North Khorasan and South Khorasan in the Islamic Republic of Iran belong to the bottom decile for health, but are well above average in the other four indicators, and even in the top one with respect to security. 4.2.2 Density As displayed in Figure 2 and in Figure A.6, the density of the state can be quite different across regions in a country. The extreme case is Benin (SSA), which encompasses both a region in the top 20% of the density scale (Littoral) and two in the bottom 10% (Alibori and Atacora). 15 16 Colombia, Honduras, and Nicaragua in LAC and Lao PDR, and Cambodia in EAS all display regions as far apart as seven deciles of the density distribution. In particular, both Colombia and Honduras have regions in the top 10% density distribution (Quindı́o and Bogotá ; and Islas de la Bahı́a, respectively), and some in the bottom third decile (Chocó and Gracias a Dios, respectively); while Nicaragua and Cambodia have a region in the second highest decile (Managua; and Phnom Penh, respectively), and some in the bottom second decile (Atlantico Norte; and Preah Vihear, Kampong Thom, Ratanak Kiri, Mondul Kiri, Kampong Chhnang, respectively). On the opposite side of the spectrum, a few countries display quite a homogeneous density across countries. In Costa Rica (LAC), all seven regions are in the top 10% of the density distribution. In Armenia (ECA), all regions fall in the top decile of the density distribution, except Gegharkunik, which is in the 9th decile. In Haiti (LAC), all regions fall at the bottom of the density scale, except the regions of North West and South, which are in the second bottom decile. As shown in Figure 3, adjusting the parameter θ primarily leads to only marginal shifts between adjacent deciles. 84 percent of regions do not change rank when θ changes from 0.5 to 1. Similarly, the rank of 80 percent of regions remain unchanged when the parameter is adjusted from 1 to 2. Occasionally, changes span across two deciles of the density distribution when θ is modi�ied. Speci�ically, in Mexico (LAC), the States of Guanajuato, and Nuevo Leó n, and in Brazil (LAC), the States of Minas Gerais, Pará , Rio de Janeiro, and Sã o Paulo, move up by 2 deciles when θ is increased from 0.5 to 1. Similarly, Jalisco, Michoacá n de Ocampo, Nuevo Leó n, and Puebla in Mexico (LAC), as well as Rio de Janeiro and Sã o Paulo in Brazil (LAC), rise by 2 deciles when θ is increased from 1 to 2. Notably, in all these instances, the dimension of state effectiveness driving the changes is security, where the normalized effectiveness score is zero or nearly zero in all regions. When the parameter θ is set to 2, allowing for greater substitution between dimensions, the poor performance in security becomes less signi�icant, and the good performance in other dimensions carries more weight in determining the overall density. Conversely, when θ is decreased to 0.5, reducing substitutability across dimensions, strong performance in other areas cannot compensate for the severe lack of security in these regions. 4.2.3 Discontinuity As shown in Table 4 and Figure 4, Mali and Mozambique stand out as the two countries with the highest discontinuity. The precise ranking depends on the degree of substitutability across dimensions. When allowing for some complementarity (θ = 0.5), Mali exhibits the greatest discontinuity, followed by Mozambique and Zambia. However, as substitutability increases (θ = 1 and θ = 2), Mali and Mozambique remain the top two, while Sudan becomes the third highest. 17 18 Figure 3: Churning across deciles of the density distribution, for different values of θ Source: Authors’ elaboration on data from the Minnesota Population Center (2020) via IPUMS International, and Raleigh, Kishi, and Linke (2023) via ACLED The ten countries with the highest discontinuity are all in Sub-Saharan Africa, with the exception of Nicaragua, when θ = 2. Costa Rica, Armenia, and the Repú blica Bolivariana de Venezuela are instead the countries with the least discontinuity, and ranking is not affected by the choice of the parameter θ. The choice of the degree of substitutability/complementarity among dimensions is not trivial, as the ranking across countries may change substantially, as it is displayed in Figure 4. For instance, in Sierra Leone improves 10 positions when increasing the degree of substitutability among dimensions. Similarly, Brazil, Mexico, and Zambia show some improvement with higher substitutability. When θ = 0.5, the better performance in some of the dimensions cannot easily compensate for a very bad performance in other dimensions. Therefore, a country like Brazil, where many regions display a very low degree of security, despite relatively good performance under different aspects, looks overall worse when considering dimensions as complements (θ = 0.5) rather than substitutes (θ = 1 or θ = 2). 19 Figure 4: Rank of countries in order of Discontinuity, different values of θ Source: Authors’ elaboration on data from the Minnesota Population Center (2020) via IPUMS International, and Raleigh, Kishi, and Linke (2023) via ACLED. Note: countries are ranked from 1-maximum discontinuity to 32-minimum discontinuity. The choice of dimensions of state intervention may have a great in�luence in determining the overall discontinuity index of a country. As a robustness check, we re-computed the index by removing one dimension at the time (see Figure 5, and Appendix, Table A.3). For a few countries the set of dimensions included in the analysis does not have an impact on the overall rank. Costa Rica and Armenia show the lowest discontinuity index irrespective of the excluded dimension and degree of substitutability. For other countries, some dimensions have a larger impact on their overall rank. Mexico and Brazil, for instance, would appear much better in the cross-country comparison if security were not included in the analysis. 20 Figure 5: Rank of countries in order of Discontinuity, de�ined on different sets of dimensions, different values of θ θ =0.5 θ =1 θ =2 Source: Authors’ elaboration on data from the Minnesota Population Center (2020) via IPUMS International, and Raleigh, Kishi, and Linke (2023) via ACLED. Note: countries are ranked from 1-maximum discontinuity to 32-minimum discontinuity 21 Source: Authors’ elaboration on data from the Minnesota Population Center (2020) via IPUMS International, and Raleigh, Kishi, and Linke (2023) via ACLED. 22 5 Navigating Territorial Presence: Unpacking State-Density Correlates In this section, we investigate deeper into potential factors associated to the state density index, based on an exploration of the literature interested in the mechanisms explaining states’ presence in a territory. According to the literature, factors associated and direction of the associations are: A1. The extension of the territory is negatively associated to state density. Olsson and Hansson (2011) show there is a negative relationship between the size of a country’s territory and the strength of the rule of law for a large cross-section of countries. Acemoglu, Garcı́a-Jimeno, and Robinson (2015) argue that in less-developed parts of the world, there is a lack of capacity of central and local states to enforce law and order, regulate economic activity, and provide public goods. They cite Migdal (1988) to introduce their analysis and motivation: “In parts of the Third World, the inability of State leaders to achieve predominance in large areas of their countries has been striking...” This inability represents a conundrum for developing economies in which the state presence seems to be concentrated in speci�ic regions. In addition, there is evidence showing that country-speci�ic factors like the geographic expanse of a country affect both the incentives for engaging in corrupt activities and the governance of such acts (Goel & Nelson, 2007, 2010). Therefore, we would expect a lower state density in larger regions because it might be relatively hard to monitor government of�icials in geographically dispersed locations. This could translate into inef�iciencies in public service delivery. In short, building on these results, a �irst hypothesis emerges about the relationship between State density and geographic expanse. A2. State density tends to be lower when the proportion of foreign-born people in a region increases. A second subject of interest is whether foreign-born citizens’ presence within a territory correlates with higher state density. Easterly and Levine (1997) show that ethnolinguistic diversity helps explain cross-country differences in public policies and other economic indicators. In particular, they argue that low schooling, political instability, and insuf�icient infrastructure, among others, are explained by Africa’s high ethnic fragmentation. A growing body of research connects diversity to anti-welfare attitudes and lower levels of social welfare expenditure. Following the foundational work of Alesina and Glaeser (2004), where they claim that ethnic heterogeneity and fractionalization are some of the main reasons explaining why the United States has a weaker welfare state than Western Europe, other authors have claimed that immigration could be a potential challenge for Western European welfare states also (Eger, 2010; Larsen, 2011). Eger and Breznau (2017) utilizing multilevel 23 modeling, �ind a negative relationship between regional percent foreign-born and support for redistribution alongside between regional percent foreign-born and support for a comprehensive welfare state across 13 European countries. These �indings are key to our purposes because the welfare state is the modern institution responsible for the distribution of social services and bene�its to members of society like education, health care, and childcare (Eger & Breznau, 2017; Roosma, Gelissen, & van Oorschot, 2013). If people do not support social policy, the presence of the state, by our measures, could be lower. Interestingly, not all the works converge to the same point. Brady and Finnigan (2014) use the foreign- born percent of the population, net migration, and the 10-year change in the percent foreign born to explore the relationship between immigration and the welfare state in 17 af�luent economies. They show these variables fail to have robust signi�icant negative effects on welfare attitudes. Notably, most research on immigration and attitudes toward the welfare State comes from analyses of US states or comparisons of the United States to Europe. We inform these results by examining the relationship between the percentage of foreign-born people and state density in developing economies. Given the growing bulk of the literature, we expect to �ind a negative relationship between both variables. A3. State density is lower as the percentage of people belonging to an indigenous group increases. The literature has systematically shown that indigenous institutions matter for current economic development (Michalopoulos & Papaioannou, 2013; Dippel, 2014; Alsan, 2015), countries’ institutional development (Bentzen, Hariri, & Robinson, 2019), and public goods provision (Gennaioli & Rainer, 2007). Historically, however, some indigenous institutions have been ignored by the states. Take, for instance, the case of the United States, where the founders ignored not only the Native American institutions but the Native American people. Indigenous people were considered citizens of their separate nations, not citizens of the United States for an extended period (Bentzen et al., 2019). Moreover, some evidence indicates that indigenous populations’ incidence could be negatively correlated with state density. The World Bank (2017) shows that in Bolivia, the least dense region (Potosı́) is also the region with the highest percentage of the indigenous population, who historically have been underrepresented in state institutions and policy making. We take a more systematic approach to explore the relationship between the state’s presence and the percentage of the indigenous population. A4. State density tends to be lower in rural sectors. In developing economies, the access and quality of public services differ in rural and urban settings (World Bank, 2017), where basic services in rural areas are less accessible and of lower quality than those in the urban counterparts (Brinkerhoff, Wetterberg, & Wibbels, 2018). Brinkerhoff et al. (2018) show how remoteness is strongly related to access to and satisfaction with public services in rural Africa. Fafchamps and Moser (2003), using 24 data from Madagascar, show that crime increases with distance from urban centers and decreases with population density. Headey, Stifel, You, and Guo (2018) �ind that children in Sub-Saharan Africa’s rural communities have much worse linear growth and dietary outcomes than urban children. Passarelli-Araujo and de Souza (2023) show that Brazilian adults living in rural areas are more likely to perceive their health as poor than their urban counterparts, and the urban-rural health disparities are signi�icant and in�luenced by sociodemographic attributes. Fu, Sun, and Fang (2024) calculate the urban and rural water footprint of prefectures in China and �ind large urban and rural water footprint disparities with per capita urban water footprint on average over two times per capita rural water footprint. Energy consumption in Chinese urban areas is also found to be higher than that in rural areas (Wu, Geng, Zhang, & Wei, 2022; Zhao & Zhang, 2018). Given this evidence, we should expect a negative correlation between state density and the rural sector. A5. Finally, the association between state density and population density follows a U- shaped pattern. “Population density increases the variance of what can happen in the territory…” said Nobel Prize winner Paul Romer. The role of governance is to increase the likelihood of good things happening, such as productivity gains due to agglomeration, social cohesion, and so on, while preventing bad things from happening –such as crime, segregation, and the like. In that sense, it is not surprise that population density is closely associated to state’s performance. Goel and Nelson (2010) argue that there might be a greater chance of corrupt practices being caught or exposed in areas with dense population concentrations. In terms of our analysis, those regions with lower population densities (presumably) would have a lower state density because it could be more dif�icult for people to monitor the activities of state servants and this could mean de�iciencies in public service delivery. Dispersion increases the cost of the state’s presence, while excess agglomeration makes certain types of quality service delivery more complex. Ladd (1992) proposes a U-shaped relationship between density and spending on public services, implying that after an optimal density, expenditures and density would rise. More recently, de Duren and Compeá n (2016), in a study of about 8,600 municipalities of Brazil, Chile, Ecuador, and Mexico, found that per capita municipal spending on public services is strongly and non-linearly correlated to urban population density and optimal expenditure levels for municipal services are achieved with densities close to 9,000 residents per square kilometer. Based on these �indings we propose the �inal hypothesis. 25 5.1 Using the state density index to test the associations 5.1.1 Available data sources Data come from IPUMS International. IPUMS International collects and distributes census microdata from around the world. The main associated variable is the state density index. 4 The sample includes 181 �irst-level administrative regions, across 9 countries: Bolivia, Brazil, Colombia, Costa Rica, Ecuador, El Salvador, Guatemala, Mexico, and Nicaragua. 5.1.2 A simple model speci�ication In order to verify the empirical relationships, we use a country-�ixed effect model as our main speci�ication. This approach accounts for individual-speci�ic effects that capture unobserved heterogeneity among countries. We primarily use OLS to estimate the following equation: yij = Xijβ + αj + εij (4) where i indexes �irst-level administrative regions and j countries. In addition, αj are country-�ixed effects and X is a vector of region-level covariates. Our dependent variable is the state density index for each �irst-level administrative region in each country (yij). Our covariates are as follows: i) the area in square kilometers of the region, in thousands (K); ii) the percentage (%) of foreign-born people; iii) the % of people belonging to an indigenous group; iv) the square of the population density per square kilometer (in K); and v) and the % of households located in a place designated as urban. 5 5.1.3 Results Table 5 presents our main results. The �irst column lists the covariates. The next three columns show results for different θ values under the �ixed-effects speci�ication. The last three columns display results using pooled regressions, an alternative speci�ication. The �ixed effect model accounts for unobserved heterogeneity by allowing each country to have its own intercept, effectively controlling for time-invariant characteristics that may affect density. On the other hand, in the pooled speci�ication all regions are treated as if they come from a single country, ignoring potential differences across countries by assuming a common intercept for all units. We �ind robust evidence supporting A1. Holding other variables constant, the geographic extension is negatively correlated with states’ presence when θ = 0.5 (p = 0.005) and θ = 1 (p = 0.023). Interestingly, the variable loses signi�icance when θ = 2 (p = 0.113). We also �ind robust evidence supporting A4. State density is higher in areas with a greater percentage of 4 Our methodology involved integrating data sets from two separate sources, IPUMS International and the Armed Con�lict Location & Event Data Project (ACLED), to create the index. 5Table A.4 in the Appendix presents pairwise correlations in a correlation matrix for these covariates. We found no strong correlations among the variables, suggesting no multicollinearity issues in our model. 26 households located in places designated as urban, ceteris paribus, for all selected values of θ (p < 0.001). Our results then corroborate that the size of a country’s territory is negatively associated with the state’s presence, which is a contribution because of the composite nature of our indicator. This goes beyond individual dimensions such as rule of law, as conducted by Olsson and Hansson (2011). In addition, our results contribute to the literature on urban-rural disparities regarding the access and quality of public services (World Bank, 2017; Brinkerhoff et al., 2018; Headey et al., 2018; Passarelli, Araujo & de Souza, 2023) by showing evidence of the lower states’ presence in the rural sector. Conversely, we �ind no statistical evidence supporting A2, A3, or A5. To further explore, we use pooled OLS models, where data from different countries are aggregated and analyzed together, as shown in Table 5, right panel. These models con�irm support for A1 and A4, and interestingly reveal a positive, signi�icant relationship between our state density index and the percentage of foreign-born, contrary to A2. This may be due to the fact that more af�luent and better served regions are attracting more foreigners who are seeking employment opportunities. Table 5: Correlates to State’s Density, different values of In Table 6, we conduct a similar analysis using one dimension of normalized effectiveness at a time as the dependent variable, rather than the overall density index. The results largely support A4, with one caveat. Regardless of the speci�ication, we �ind that the state tends to provide better basic services, education, and health in urban areas compared to rural ones. However, in terms of security, urban settlements appear to be at a disadvantage. Security is 27 also negatively correlated with the geographic size of regions, which further supports A1. Additionally, the proportion of the indigenous population is strongly and negatively correlated with the literacy rate of adults, lending support to A3. Table 6: Correlates of State’s Effectiveness, θ = 1 6 Concluding Remarks: A New Approach to Measure State Presence In this paper, we developed an index to measure the discontinuity of the state, which we de�ine as the inconsistent presence and responsiveness of the state across its territory. Our approach consists of three steps: (i) evaluating the effectiveness of the state in selected dimensions within each region; (ii) summarizing this effectiveness into an index re�lecting the density of state intervention in each region; and (iii) calculating the discontinuity of state intervention across different regions. We then applied the index to representative data for a total of 545 regions in 32 developing countries. By focusing on the dimensions of basic services, education, health, and security, we aligned our analysis with the most commonly identi�ied priorities in international opinion surveys, including the International Social Survey Programme (ISSP) and the Global Barometer Survey (GBS). The analysis utilized speci�ic indicators to measure state effectiveness: the provision of piped water and electricity for basic services, literacy rates for education, child survival rates for health, and the frequency of peaceful days for security. These indicators were derived from IPUMS and ACLED data, ensuring a robust and representative evaluation. Our �indings reveal signi�icant variations in state effectiveness across the regions analyzed, underscoring the uneven distribution of state resources and interventions within 28 countries. In particular, regions in Latin American countries—from which a large portion of our sample is drawn—exhibited notable challenges in maintaining security, despite performing well in other key indicators such as education, health, and basic services. This suggests that while states in Latin America may have succeeded in certain social service sectors, they still face substantial obstacles in ensuring public safety and maintaining law and order, which remains a pressing issue in many of these regions. In contrast, Sub-Saharan African countries predominantly face challenges in providing basic services, particularly in areas such as water supply and electricity. This re�lects a persistent gap in infrastructure development, in spite of recent government efforts to close the gaps. The analysis also found that East Asia and Paci�ic countries are overrepresented among those struggling with providing water access, indicating that even relatively developed regions still face signi�icant service delivery gaps that need to be addressed. Interestingly, a few countries exhibit a relatively homogeneous state presence across their regions. For example, Haiti stands out as a case where all regions are equally under- served across the dimensions we analyzed, from education to healthcare and basic infrastructure. Conversely, countries like Costa Rica and Armenia display a more even distribution of state interventions, where all regions are similarly well-served. This homogeneity might indicate strong commitment to equitable service delivery across the country, which, in turn, reinforces the social compact, and overall cooperation (see World Bank, 2017). However, the majority of the countries in our study exhibit large discontinuities in state presence. In these countries, some regions are highly served, with well-established public services and infrastructure, while others remain neglected, suffering from low levels of state intervention and access to basic services. This disparity can contribute to uneven development within a country, potentially exacerbating social inequality and stoking regional grievances. The fact that the ten countries with the highest levels of discontinuity are all located in Sub-Saharan Africa is noteworthy, associated also with the fact that it is the region with the highest number of extreme poor people globally. Overall, our analysis underscores the importance of considering sub-national variations in state effectiveness when designing development policies. Large regional disparities in state intervention can undermine national efforts to promote inclusive growth and stability. Addressing these discontinuities requires a more nuanced understanding of local contexts and a tailored approach that accounts for the speci�ic needs and challenges of each region. Ensuring that state services are more evenly distributed across regions could help reduce inequalities and improve overall governance outcomes. This highlights the need for targeted policies that focus on the most underserved regions, particularly in countries where large discrepancies in service delivery persist. The study highlights the crucial role of sub-national data in assessing state effectiveness. By examining regional disparities within countries, the 29 paper reveals how the effectiveness of the state can vary drastically even within the same national borders. This is particularly important in developing countries, where national-level indicators may mask signi�icant inequalities in state presence and service delivery at the local level. The focus on key indicators—basic services, education, health, and security—re�lects the importance of these areas in the daily lives of citizens. These are the foundational pillars on which broader development is built, and any discontinuities in their provision can have far- reaching consequences for overall national progress. Policy makers should prioritize closing gaps in these basic areas, as addressing these fundamental needs can have a multiplier effect on other development goals. For example, ensuring better access to healthcare and education can improve long-term economic outcomes, while improving security can create a more stable environment for investment and growth. Efforts should focus on regions with the most pressing needs, while also ensuring that improvements in one sector do not come at the expense of others. While the study focused on 32 developing countries and four key dimensions of state effectiveness, the �indings suggest that the research could be expanded to include a wider range of countries and other indicators that may be relevant for assessing state effectiveness. Future research could incorporate additional indicators, such as infrastructure quality, economic development, or environmental sustainability, to offer a more comprehensive picture of state functionality. Moreover, expanding the geographic scope to include more countries from different regions could provide a richer understanding of how state effectiveness varies globally and help identify cross-national trends and patterns. By incorporating more dimensions and expanding the data set, researchers could generate more nuanced insights into how the state affects development outcomes. In conclusion, this study offers a valuable framework for assessing the discontinuity of state intervention, emphasizing the need for localized data and targeted interventions to address regional disparities in governance. The �indings underscore the importance of understanding the complex and varied challenges that different regions face, while also highlighting opportunities for improving state effectiveness through more tailored, region- speci�ic policies. Future research should expand the scope of analysis and include additional dimensions of state functionality to provide a more complete understanding of governance and its implications for sustainable development. 30 References Acemoglu, D., Garcı́a-Jimeno, C., & Robinson, J. A. (2015). State capacity and economic development: A network approach. American Economic Review, 105, 2364-2409. doi: 10.1257/AER.20140044 Alesina, A., & Glaeser, E. (2004). Fighting Poverty in the US and Europe: A World of Difference. Oxford University Press. doi: 10.1093/0199267669.001.0001 Alsan, M. (2015). The effect of the tsetse �ly on african development. American Economic Review, 105, 382-410. doi: 10.1257/AER.20130604 Barrera-Osorio, F., Blakeslee, D. S., Hoover, M., Linden, L., Raju, D., & Ryan, S. P. (2022). Delivering Education to the Underserved through a Public-Private Partnership Program in Pakistan. The Review of Economics and Statistics, 104(3), 399–416. doi: https://doi.org/10.1162/rest_a_01002 Beath, A., Christia, F., & Enikolopov, R. (2012). Winning hearts and minds through development? evidence from a �ield experiment in afghanistan. Policy Research Working Papers. The World Bank. doi: 10.1596/1813-9450-6129 Bentzen, J. S., Hariri, J. G., & Robinson, J. A. (2019). Power and persistence: The indigenous roots of representative democracy. The Economic Journal, 129, 678-714. doi: 10.1111/ECOJ.12568 Bereketeab, R. (2020). Education as an instrument of nation-building in postcolonial africa. Studies in Ethnicity and Nationalism, 20, 71-90. doi: 10.1111/SENA.12317 Besley, T., & Persson, T. (2009). The origins of State capacity: Property rights, taxation, and politics. American Economic Review, 99(4), 1218-44. doi: 10.1257/aer.99.4 .1218 Bondarenko, N., Krasil'nikova, M., & Shishkin, S. (2009). Public opinion about equal access to medical services. Sociological Research, 48, 3-22. doi: 10.2753/SOR1061 -0154480301 Brady, D., & Finnigan, R. (2014). Does immigration undermine public support for social policy? American Sociological Review, 79(1), 17-42. doi: 10.1177/ 0003122413513022 Brinkerhoff, D. W., Wetterberg, A., & Wibbels, E. (2018). Distance, services, and citizen perceptions of the State in rural africa. Governance, 31(1), 103-124. doi: 10.1111/gove.12271 Cuadrado-Ballesteros, B., Garcı́a-Sá nchez, I. M., & Prado-Lorenzo, J. M. (2012). Effects of different modes of local public services delivery on quality of life in spain. Journal of Cleaner Production, 37, 68-81. doi: 10.1016/J.JCLEPRO.2012.06.008 de Duren, N. L., & Compeá n, R. G. (2016). Growing resources for growing cities: Density and the cost of municipal public services in latin america. Urban Studies, 53(14), 3082– 3107. Dippel, C. (2014). Forced coexistence and economic development: Evidence from native american reservations. Econometrica, 82, 2131-2165. doi: 10.3982/ECTA11423 Dittmar, J. E., & Meisenzahl, R. R. (2020). Public goods institutions, human capital, and growth: Evidence from german history. The Review of Economic Studies, 87, 959-996. doi: 10.1093/RESTUD/RDZ002 31 Easterly, W., & Levine, R. (1997). Africa’s Growth Tragedy: Policies and Ethnic Divisions. The Quarterly Journal of Economics, 112(4), 1203-1250. doi: 10.1162/ 003355300555466 Eger, M. A. (2010). Even in sweden: The effect of immigration on support for welfare State spending. European Sociological Review, 26, 203-217. doi: 10.1093/ESR/JCP017 Eger, M. A., & Breznau, N. (2017). Immigration and the welfare State: A cross-regional analysis of european welfare attitudes. International Journal of Comparative Sociology, 58(5), 440-463. doi: 10.1177/0020715217690796 Eurostat. (2024). Regions in the European Union Nomenclature of territorial units for statistics (NUTS), 2024 Edition, Manuals and Guidelines. Fafchamps, M., & Moser, C. (2003). Crime, Isolation and Law Enforcement. Journal of African Economies, 12(4), 625-671. doi: 10.1093/jae/12.4.625 Flores-Macı́as, G. A. (2018). Building support for taxation in developing countries: Experimental evidence from mexico. World Development, 105, 13-24. doi: 10.1016/ j.worlddev.2017.12.014 Fu, Z., Sun, S., & Fang, C. (2024). Unequal prefecture-level water footprints in china: The urban-rural divide. Science of The Total Environment, 912, 169089. doi: 10.1016/ J.SCITOTENV.2023.169089 GBS. (2020). Global barometer survey, pooled data�ile. Taipei: Hu Fu Center for East Asia Democratic Studies, NTU. Retrieved from https://www.globalbarometer.net/ Gennaioli, N., & Rainer, I. (2007). The modern impact of precolonial centralization in africa. Journal of Economic Growth, 12, 185-234. doi: 10.1007/S10887-007-9017-Z/ METRICS Goel, R. K., & Nelson, M. A. (2007). Are corrupt acts contagious?: Evidence from the united States. Journal of Policy Modeling, 29, 839-850. doi: 10.1016/J.JPOLMOD .2007.09.002 Goel, R. K., & Nelson, M. A. (2010). Causes of corruption: History, geography and government. Journal of Policy Modeling, 32, 433-447. doi: 10.1016/J.JPOLMOD .2010.05.004 Headey, D., Stifel, D., You, L., & Guo, Z. (2018). Remoteness, urbanization, and child nutrition in sub-saharan africa. Agricultural Economics, 49, 765-775. doi: 10.1111/ AGEC.12458 ISSP Research Group. (2018). International social survey programme: Role of government v - issp 2016. GESIS Data Archive, Cologne. ZA6900 Data �ile Version 2.0.0. doi: 10.4232/1.13052 Justino, P. (2015). Civil unrest and government transfers in india. IDS Evidence Report 108, Brighton: IDS. Khanna, G., & Zimmermann, L. (2017). Guns and butter? �ighting violence with the promise of development. Journal of Development Economics, 124, 120-141. doi: 10.1016/J.JDEVECO.2016.09.006 Kubota, Y., Hidayat, U. K., Kurosaki, T., Obayashi, K., & Ohmura, H., (2024). Wartime service provision and post-con�lict State legitimacy: Perception-based foundation of sustainable development in Northwestern Pakistan. World Development Perspectives, 34, 100594. doi: https://doi.org/10.1016/j.wdp.2024.100594 Ladd, H. F. (1992). Population growth, density and the costs of providing public services. Urban Studies, 29(2), 273-295. doi: 10.1080/00420989220080321 32 Larsen, C. A. (2011). Ethnic heterogeneity and public support for welfare: Is the american experience replicated in britain, sweden and denmark? Scandinavian Political Studies, 34, 332-353. doi: 10.1111/J.1467-9477.2011.00276.X Latinobaró metro. (2018). Latinobarómetro 2018. Corporació n Latinobaró metro, Santiago de Chile. Mcloughlin, C. (2024). Public services as carriers of ideas that (de-) legitimise the State: The illustrative case of free education in sri lanka. World Development, 173, 106439. doi: 10.1016/J.WORLDDEV.2023.106439 Michalopoulos, S., & Papaioannou, E. (2013). Pre-colonial ethnic institutions and contemporary african development. Econometrica, 81, 113-152. doi: 10.3982/ECTA9613 Migdal, J. S. (1988). Strong societies and weak States: State-society relations and State capabilities in the third world. Princeton University Press. Minnesota Population Center. (2020). Ipums International: D020.V7.3. Integrated Public Use Microdata Series, International: Version 7.3 [dataset]. Minneapolis, MN: IPUMS, 2020. doi: 0.18128/D020.V7.3 Mobarak, A. M., Rajkumar, A. S., & Cropper, M. (2011). The political economy of health services provision in brazil. Economic Development and Cultural Change, 59, 723-751. doi: 10.1086/660001/0 O’Donnell, G. (2010). Democracy, agency and the State: Theory with comparative intent. Oxford University Press. Olsson, O., & Hansson, G. (2011, 6). Country size and the rule of law: Resuscitating montesquieu. European Economic Review, 55, 613-629. doi: 10.1016/ J.EUROECOREV.2010.09.008 Passarelli-Araujo, H., & de Souza, G. M. (2023). Urban-rural health disparities in brazil: Do sociodemographic attributes play a role? Preventive Medicine, 175, 107679. doi: https://doi.org/10.1016/j.ypmed.2023.107679 Poutvaara, P., & Jordahl, H. (2020). Public sector outsourcing. IZA World of Labor, 65. doi: https://doi.org/10.15185/izawol.65.v2 Raleigh, C., Kishi, R. & Linke, A. (2023). Political instability patterns are obscured by con�lict dataset scope conditions, sources, and coding choices. Humanities and Social Sciences Communications, 10(1), 74-17. doi: https://doi.org/10.1057/s41599-023-01559-4 Ratigan, K. (2022). Riding the tiger of performance legitimacy? chinese villagers’ satisfaction with State healthcare provision. International Political Science Review, 43(2), 259-278. doi: 10.1177/0192512120927482 Roosma, F., Gelissen, J., & van Oorschot, W. (2013). The multidimensionality of welfare State attitudes: A european cross-national study. Social Indicators Research, 113, 235-255. doi: 10.1007/S11205-012-0099-4/TABLES/8 Sá nchez de la Sierra, R. (2020). On the origins of the State: Stationary bandits and taxation in eastern congo. Journal of Political Economy, 128(1), 32-74. doi: 10.1086/703989 33 Scott, J. C. (1998). Seeing like a State: How certain schemes to improve the human condition have failed. Yale University Press. Stiglitz, J., Sen, A., & Fitoussi, J. (2009). Report of the commission on the measurement of economic performance and social progress. INSEE. Tilly, C. (2009). Foreword. In I. W. Martin, A. K. Mehrotra, & M. Prasad (Eds.), The new �iscal sociology: Taxation in comparative and historical perspective (pp. xi–xiv). Cambridge University Press. UNDP. (2017). Human development report 2016. United Nations. doi: 10.18356/b6186701- en World Bank. (2017). World Development Report 2017: Governance and the Law. Washington, DC: World Bank. doi: 10.1596/978-1-4648-0950-7 Wu, D., Geng, Y., Zhang, Y., & Wei, W. (2022). Features and drivers of china’s urbanrural household electricity consumption: Evidence from residential survey. Journal of Cleaner Production, 365, 132837. doi: 10.1016/J.JCLEPRO.2022.132837 Yitzhaki, S. (1979). Relative Deprivation and the Gini Coef�icient, The Quarterly Journal of Economics, 93(2), 321-324. doi: https://doi.org/10.2307/1885095 Zhao, P., & Zhang, M. (2018). The impact of urbanisation on energy consumption: A 30-year review in china. Urban Climate, 24, 940-953. doi: 10.1016/J.UCLIM.2017 .11.005 34 Territorial Inequalities: A Note on State Discontinuity Appendix List of Tables Table A. 1. Indicators for the construction of the State Density Index Dimension Indicator Source Service delivery Households with piped water, per cent IPUMS Households with electricity, per cent IPUMS Education Literate people in the region,1 per cent IPUMS Health Children alive, per 1,000 live births IPUMS Security Number of days in the year without episodes of violence ACLED 1 A person is typically considered literate if he or she can both read and write. All other persons are illiterate, including those who can either read or write but cannot do both Table A. 2. Countries included in the analysis, and respective data year World Region Country year East Asia and Paci�ic Cambodia 2013 Lao PDR 2005 Europe and Central Asia Armenia 2001 Latin America and the Caribbean Bolivia 2012 Brazil 2010 Colombia 2005 Costa Rica 2011 Dominican Republic 2010 Ecuador 2010 El Salvador 2007 Guatemala 2002 Haiti 2003 Honduras 2001 Mexico 2015 Nicaragua 2005 Peru 2007 Venezuela 2001 Middle East and North Africa Iran 2006 South Asia Nepal 2011 Sub-Saharan Africa Benin 2013 Ghana 2010 Lesotho 2006 Liberia 2008 Mali 2009 Mozambique 2007 Rwanda 2002 Senegal 2002 Sierra Leone 2015 Sudan 2008 Tanzania 2012 Togo 2010 Zambia 2000 Table A. 3. Discontinuity index, for different values of parameter , and different speci�ications of State interventions = 0.5 = 1 = 2 excluding: excluding: excluding: security water electricity education health security water electricity education health security water electricity education health Armenia 0.015 0.012 0.019 0.019 0.015 0.014 0.012 0.018 0.019 0.015 0.013 0.011 0.017 0.018 0.014 Benin 0.222 0.131 0.169 0.150 0.209 0.199 0.111 0.137 0.125 0.176 0.162 0.086 0.099 0.095 0.125 Bolivia 0.060 0.045 0.040 0.052 0.035 0.054 0.037 0.033 0.045 0.032 0.047 0.027 0.024 0.035 0.027 Brazil 0.041 0.131 0.137 0.129 0.128 0.040 0.074 0.081 0.072 0.071 0.038 0.044 0.056 0.042 0.042 Cambodia 0.177 0.084 0.108 0.138 0.156 0.138 0.071 0.064 0.090 0.108 0.108 0.051 0.037 0.053 0.063 Colombia 0.065 0.079 0.087 0.087 0.095 0.063 0.067 0.075 0.075 0.082 0.061 0.057 0.064 0.064 0.071 Costa Rica 0.011 0.008 0.010 0.009 0.008 0.011 0.008 0.009 0.009 0.008 0.011 0.008 0.009 0.009 0.008 Dominican Republic 0.042 0.029 0.032 0.032 0.037 0.040 0.028 0.029 0.030 0.034 0.038 0.025 0.025 0.026 0.029 Ecuador 0.042 0.016 0.032 0.037 0.039 0.041 0.016 0.030 0.035 0.036 0.038 0.015 0.025 0.031 0.031 El Salvador 0.040 0.024 0.029 0.026 0.036 0.040 0.023 0.027 0.025 0.034 0.040 0.021 0.024 0.023 0.030 Ghana 0.185 0.121 0.119 0.119 0.146 0.183 0.107 0.101 0.102 0.126 0.177 0.079 0.069 0.069 0.087 Guatemala 0.070 0.049 0.043 0.052 0.062 0.069 0.044 0.039 0.047 0.056 0.068 0.036 0.033 0.041 0.048 Haiti 0.198 0.074 0.081 0.085 0.087 0.149 0.042 0.059 0.052 0.050 0.117 0.040 0.052 0.059 0.049 Honduras 0.114 0.074 0.039 0.076 0.092 0.101 0.061 0.035 0.060 0.076 0.082 0.047 0.028 0.039 0.054 Iran, Islamic Rep 0.036 0.034 0.040 0.034 0.026 0.032 0.030 0.037 0.030 0.025 0.027 0.024 0.032 0.025 0.024 Lao PDR 0.139 0.096 0.063 0.094 0.124 0.126 0.080 0.054 0.070 0.103 0.102 0.056 0.041 0.045 0.067 Lesotho 0.076 0.060 0.028 0.059 0.068 0.052 0.036 0.026 0.035 0.040 0.041 0.024 0.022 0.022 0.025 Liberia 0.166 0.094 0.099 0.095 0.129 0.143 0.067 0.078 0.064 0.087 0.123 0.038 0.047 0.037 0.041 Mali 0.335 0.371 0.357 0.340 0.460 0.273 0.308 0.298 0.299 0.430 0.215 0.245 0.240 0.247 0.406 Mexico 0.018 0.115 0.118 0.114 0.112 0.018 0.070 0.074 0.069 0.067 0.017 0.041 0.046 0.040 0.039 Mozambique 0.365 0.227 0.235 0.242 0.311 0.326 0.180 0.192 0.182 0.240 0.268 0.123 0.135 0.120 0.150 Nepal 0.094 0.055 0.073 0.068 0.087 0.088 0.051 0.064 0.061 0.079 0.080 0.045 0.052 0.050 0.065 Nicaragua 0.147 0.090 0.085 0.110 0.130 0.143 0.082 0.076 0.098 0.118 0.135 0.066 0.059 0.076 0.092 Peru 0.089 0.052 0.058 0.074 0.071 0.086 0.049 0.053 0.069 0.064 0.080 0.043 0.044 0.057 0.053 Rwanda 0.398 0.236 0.183 0.286 0.197 0.296 0.114 0.116 0.145 0.126 0.226 0.048 0.059 0.048 0.060 Senegal 0.221 0.115 0.122 0.168 0.160 0.204 0.089 0.096 0.132 0.121 0.178 0.051 0.057 0.078 0.066 Sierra Leone 0.276 0.158 0.123 0.173 0.203 0.238 0.097 0.091 0.101 0.124 0.211 0.043 0.043 0.038 0.048 Sudan 0.277 0.183 0.203 0.245 0.295 0.235 0.152 0.164 0.193 0.261 0.177 0.116 0.123 0.131 0.215 Tanzania 0.176 0.088 0.105 0.135 0.146 0.157 0.068 0.085 0.101 0.112 0.133 0.044 0.056 0.057 0.065 Togo 0.289 0.176 0.171 0.200 0.222 0.277 0.147 0.140 0.160 0.177 0.255 0.096 0.088 0.098 0.104 Venezuela, RB 0.029 0.018 0.024 0.022 0.026 0.029 0.018 0.024 0.021 0.025 0.028 0.017 0.022 0.020 0.024 Zambia 0.363 0.184 0.193 0.264 0.259 0.321 0.123 0.142 0.171 0.167 0.268 0.055 0.072 0.069 0.070 Source: Authors’ elaboration on data from the Minnesota Population Center (2020) via IPUMS International, and Raleigh, Kishi, and Linke (2023) via ACLED. Table A. 4. Correlation matrix for our covariates Variables Area Foreign Indigenous Population Urban Area 1.000 Foreign -0.102 1.000 (0.172) Indigenous -0.132 -0.105 1.000 (0.076) (0.161) Population -0.038 -0.006 -0.028 1.000 (0.608) (0.938) (0.708) Urban 0.260 0.060 -0.316 0.153 1.000 (0.000) (0.419) (0.000) (0.040) Source: Authors’ elaboration on data from the Minnesota Population Center (2020) via IPUMS International. Notes: p-values in parentheses. List of Figures Figure A. 1. Normalized Effectiveness in different State Domains, Costa Rica Source: Authors’ elaboration on data from the Minnesota Population Center (2020) via IPUMS International, and Raleigh, Kishi, and Linke (2023) via ACLED. Notes: Darker colors indicate higher normalized effectiveness. Figure A. 2. Normalized Effectiveness in different State Domains, Mali Source: Authors’ elaboration on data from the Minnesota Population Center (2020) via IPUMS International, and Raleigh, Kishi, and Linke (2023) via ACLED. Notes: Darker colors indicate higher normalized effectiveness. Figure A. 3. Normalized Effectiveness in different State Domains, Mozambique Source: Authors’ elaboration on data from the Minnesota Population Center (2020) via IPUMS International, and Raleigh, Kishi, and Linke (2023) via ACLED. Notes: Darker colors indicate higher normalized effectiveness. Figure A. 4. Normalized Effectiveness in different State Domains, Armenia Source: Authors’ elaboration on data from the Minnesota Population Center (2020) via IPUMS International, and Raleigh, Kishi, and Linke (2023) via ACLED. Notes: Darker colors indicate higher normalized effectiveness. Figure A. 5. Normalized Effectiveness in different State Domains, Brazil Source: Authors’ elaboration on data from the Minnesota Population Center (2020) via IPUMS International, and Raleigh, Kishi, and Linke (2023) via ACLED. Notes: Darker colors indicate higher normalized effectiveness. Figure A. 6. Density of the State, = 1 Source: Authors’ elaboration on data from the Minnesota Population Center (2020) via IPUMS International, and Raleigh, Kishi, and Linke (2023) via ACLED. Notes: Darker colors indicate higher density.