Experimental studies using randomized controlled trials (RCTs) have long been a staple of medical research. In recent years, these methods have become increasingly popular in the social sciences. In development economics in particular, their use has attracted considerable debate with some scholars promoting them as the best means of identifying “what works” in terms of development interventions (Banerjee 2007; Glennerster and Kremer 2011), while others voice strong concerns and challenge their growing hegemony in the field (see e.g., Deaton 2009; Ravallion 2009).
This article, which also serves as the introduction to this symposium, focuses on the use of RCTs in identifying what works for one of the major topics in contemporary development policy: government performance. It asks two questions: (1) what have been the main contributions of RCTs to the study of government performance? and (2) what could be the contributions, and relatedly, the limits? Despite large separate literatures on government performance and on experimental methods, very little work has directly considered both together in this way. This article draws on review of both literatures, including a new systematic review of experimental and quasi-experimental studies on governance and government performance conducted by Gisselquist et al. (2013).
Broadly, we argue that RCTs have been and can be useful in studying the effects of some policy interventions to improve government performance, but that their use in testing hypotheses about the causes of change in government performance is limited in significant ways by the nature of the factors that we expect to matter most in this area. Theories of government and the state suggest that what might be most important in explaining variation in government performance are broad macro-structural shifts, national-level variation in institutions, and other not-easily-manipulable factors such as leadership. By contrast, RCTs have been used effectively and are best for studying targeted interventions to improve government performance (particularly in areas of public goods provision, voting behavior, and specific measures to address corruption and improve accountability), which are expected to have rapid results. In other words, the types of factors highlighted in some major theories of government performance are not amenable to study with RCTs, suggesting the continuing relevance of quasi-experimental and other techniques to hypothesis testing and policy design in this area, as well as the value of mixing experimental and non-experimental approaches.
As summarized in the final section of this article, the rest of this symposium builds from this discussion to consider both how experimental work might better address identified weaknesses and how experimental and non-experimental techniques might be combined most fruitfully, with particular attention to the study of government performance. Collectively, the studies in this symposium speak to debates over experimental methods in development studies more generally, pushing beyond what have become well-rehearsed and polarized positions towards a new middle ground. Contributors include scholars who regularly use experimental methods in their own work and those who have been more critical or reliant on other methods.
Discussion of experimental research, particularly in economics and political science, sometimes treats laboratory-type experiments, natural experiments, and RCTs or field experiments together, irrespective of their design features. Much of the discussion in this article is applicable to various experimental methods, but the focus is on RCTs, which imply a slightly different approach than the others to the testing of causal hypotheses: in the simplest experimental designs, causal effects are assessed by comparing measures “with” and “without” an intervention. This is most straightforward in a laboratory setting where other key variables can be held constant and measures can be taken before and after an intervention. In this setting, causal inference is relatively clear: the intervention causes the difference.
Many of the phenomena that we care about, however, are not amendable to this method. Outside of the laboratory setting, field experiments and RCTs study such phenomena using similar principles; because it is not always possible to hold constant all factors, in prospective experimental designs, with baseline and endline data, the identification of the counterfactual is achieved via random assignment to treatment, where measures from randomly selected “control” and “treatment” groups are taken before and after interventions, and the effect of the intervention then is the difference between “before” and “after” measures in the treatment as compared to the control groups. This basic and elegant logic underlies hypothesis testing and impact evaluation using RCTs, by ensuring that – in principle – any difference between the treatment and control is not systematic at the outset of the randomized experiment.
Quasi-experimental approaches by contrast rely on observational data that lack random treatment assignment. Relying on similar logic to experiments, they draw on quantitative techniques such as instrumental variables, regression discontinuity, or propensity score matching to estimate the impact of interventions. Our review of the literature in this study does not focus on other non-experimental methods, such as qualitative case studies, although the middle ground we explore in this symposium could also draw on these methods.
This article has three main sections following this introduction. The first briefly discusses theories of governance, highlighting several major ideas from the literature about the factors that cause variation in government performance. The second focuses on how RCTs have been used in the study of related topics, highlighting major findings from RCTs with respect to the provision of health, education, and other public goods; improvements in the performance of civil servants; and representation, participation, and deliberative democracy. The third part of the article brings these two sections together, exploring the limits and potential contributions of RCTs in the study of government performance and introducing the contributions of the studies in this symposium. A final section concludes.
2 Explaining Government Performance
Despite a wealth of literature on governance and government performance, even definitions remain contested (see Keefer 2009; Gisselquist 2012; Bratton 2013). Without delving too much into these debates, we adopt a basic definition of governance building on theories of government and the state. This work points to two major roles for public institutions in (1) providing public goods such as education, health care, water and sanitation facilities, and social protection, and (2) aggregating interests with respect inter alia to how and which public goods are provided. The later role can be achieved both through electoral and non-electoral forms of participation and is closely linked to discussion of accountability. As Putnam (1993: pp. 8–9) notes in Making Democracy Work, “public institutions are devices for achieving purposes, not just for achieving agreement. We want government to do things, not just decide things – to educate children, pay pensioners, stop crime, create jobs, hold down prices, encourage family values, and so on.” Individuals and groups within a polity have varying preferences about the type and manner of public goods provision and other collective issues, and a second key role of government is in “aggregating” and representing such interests to make collective decisions. In short, as Levi (2006: p. 5) summarizes, “Good governments are those that are (1) representative and accountable to the population they are meant to serve, and (2) effective – that is, capable of protecting the population from violence, ensuring security of property rights, and supplying other public goods that the populace needs and desires”. By extension, the quality of governance varies in the degree to which governments fulfil these two related roles.
A standard definition of government performance is the “capacity of governing authorities to provide public goods and services” (Bratton 2013: p. 2). As Bratton (2013: p. 2) elaborates:
Public goods and services are benefits intended for shared consumption from which nobody can be legally or effectively excluded. These benefits may be general political guarantees (such as national defense, law and order, or civil rights) or particular socio-economic commodities (such as transport infrastructure, electricity supply, or health and education services). Either way, government performance concerns the track record of state agencies at addressing collective social needs.
In these terms, the study of government performance is understood as an integral part of the broader literature on governance. It highlights effectiveness in public goods provision, but representativeness and accountability are also key because effectiveness may be judged in terms of the upholding of political guarantees that make possible representativeness and accountability (as well as the provision of socio-economic commodities), and because citizen assessment is relevant to the evaluation of how well collective social needs are addressed. Much more than in the literature on governance, the literature on government performance highlights performance management – “the planning, programming, implementation, measurement, monitoring, evaluation, and reporting of program results” (Bratton 2013: p. 2). A clear focus has been the range of ways in which better public sector management – including strategic management, financial management, human resource management, performance measurement, quality management, process management, and so on – influence government performance outcomes (see Bovaird and Loffler 2009). Work on “governance” by multilateral development banks, in particular the World Bank, can be understood in these terms as broadly indistinguishable from government performance. The World Bank (2012: p. 9) explicitly treats governance as “the processes by which authority is exercised in the management of a country’s economic and social resources” and “the capacity of governments to design, formulate, and implement policy and deliver goods and services.”
Theories of government and the state, however, push us beyond this focus on public sector management to suggest a number of additional explanations for variation in government performance, both across polities and over time, tied to a range of structural, institutional, and cultural factors, as well as individual agency. In general, this work deals with the two components of governance separately, offering explanations either for better representation and accountability (often framed in terms of the emergence of liberal democracy versus other forms of government), or for more effective public goods provision. Much work also addresses disaggregated government performance outcomes, such as the provision of effective policing, secure property rights, universal health care, or high quality state-funded education.
Far from having a single model of change in government performance, major theoretical traditions offer diverse and often contradictory explanations for key outcomes. For instance, what constitutes good government performance in terms of providing the institutional environment most conducive to economic growth? Friedman (1962) suggests that a “good” government serves as a “rule maker” and “umpire” to create and enforce minimal rules, such as property rights and a monetary framework. Disciples of Keynes (1964), by contrast, see a more extensive role for governments in fiscal and monetary policy. Similarly, major debates concern whether more or less regulation is most conducive to private sector development, and the form that regulation should take (see Kirkpatrick 2014).
One of the major structural factors highlighted in classical explanations of variation in the quality of governance is “modernization” or the level of development. Max Weber, for instance, suggested that modernization leads to fundamental changes in the nature of authority, from traditional and charismatic towards the sort of rational-legal approach embodied in contemporary public sector management principles (Weber 2009). Modernization theory of the 1950s and 1960s posited that economic growth led to fundamental structural changes in the economy and society, which would in turn lead to greater popular participation in government and the foundations of democracy (Lipset 1981). Other major structural arguments highlight factors ranging from the class structure (e.g., Moore 1966; Luebbert 1991) and ethnic structure (e.g., Horowitz 1985) to the role of the state and state-society relations (e.g., Skocpol 1979; Migdal 1988) and geography (e.g., Herbst 2000).
Institutional explanations for governance outcomes are among the most diverse, focusing on a range of forms and mechanisms. Indeed, social scientists often define institutions so broadly – as formal and informal rules, norms, and organizations – that some of the structural explanations noted above and the cultural explanations noted below are treated in this camp (see Steinmo 2008). Since the 1990s, new institutionalist economics inspired by North (1990) and others has been particularly important in the thinking on governance underlying work by the World Bank and other multilateral development banks, which has focused on how the “rules of the game” shape economic development (Grindle 2010). The World Bank, for instance, has focused on its role in helping countries to “put in place institutions and systems that can become the foundations of sustainable growth” (World Bank 2012). It points to the need both to strengthen the capacity of institutions to enforce regulations, provide public services, and manage resources effectively, and to adopt the “right” institutions (e.g., regulations favorable to private sector development).
Institutional explanations often focus our attention on the long-running impact of “lock in” or path dependence that makes changing institutions difficult and costly, even when they are inefficient. North argues, for instance, that it was rather haphazard institutional choices that put England on a path toward efficient market economy, with relatively strong property rights, an impartial judicial system, and a fiscal system with expenditures tied to tax revenues, where other countries adopted different (and ultimately less effective) institutions that placed them on different paths. Likewise, from a sociological perspective, Immergut (1992) argues that the structure of political institutions in Sweden, France, and Switzerland influenced whether each country developed comprehensive national health care or more fragmented insurance programs. Political institutions and procedures, rather than the demands of social groups, set the terms of political negotiations, leading to divergent outcomes.
Institutionalists have also studied constitutional engineering and electoral system reform as means of improving representation, accountability, and governance more generally, particularly in divided societies (see Sartori 1997; Reilly 2001). Consociational theory, for instance, proposes that governance in a state divided along ethnic, religious, or communal lines can be stable if democracy has four key institutional characteristics: a grand coalition formed by the political leaders of various factions; a mutual veto, necessitating consensus among groups for political decisions; proportionality, in that each group occupies a share of government posts proportional to its share of the population; and segmental autonomy, allowing autonomous rule for different groups (Lijphart 1977). Many favor for similar reasons the reorganization of the state along more decentralized or federal lines in order to defuse social divisions, increase public accountability and political participation, and improve public service delivery (see Seabright 1996; Bardhan 2005; Brancati 2006; Robinson 2007).
Cultural factors are also relevant to theories of variation in government performance. Tocqueville’s (2003) classic exploration of the role of political culture in explaining democracy in America is one example. In the Tocquevillian tradition, Putnam (1993) explored government performance across Italian regions, taking advantage of a unique situation in which 15 new regional governments were established simultaneously in 1970 with similar constitutional structures and mandates, but some performed better than others. In explaining this variation, he pointed to the role of social capital – rooted in Italy in early medieval history. The role of social capital in government performance and the factors that influence its variation likewise have been emphasized in multiple contexts (e.g., Boix and Posner 1998; Varshney 2001; Putnam 2007).
Finally, a significant body of work points to the decisive role of individuals, namely decision-makers and political leaders, in affecting government performance. Kingdon’s (1995) model of policy-making, for instance, posits three “streams of processes”: problems, policies, and politics. Policy windows, which may arise predictably (such as during a vote on legislation) or suddenly (when problems arise), are periods during which the three streams are combined and issues may rise on the policy agency. Policy entrepreneurs take advantage of policy windows to push their agendas and particular policy solutions. Grindle and Thomas (1991) similarly suggest a critical role played by individuals in agenda setting, decision making, and the implementation of public reforms that affect government performance. Leaders are often seen to play an especially decisive role during periods of institutional change and uncertainty (Samuels 2003). Scholarly attention has also focused on the role of particular individuals (e.g., Mukunda 2012) and leadership styles (e.g., Jackson and Rosberg 1982).
In summary, the study of how and how well governments perform is central to the study of politics, and work in the field – just a very few examples of which are cited here – suggests a number of structural, institutional, cultural, and agent-focused explanations for variations in government performance. This brief overview provides a quick glimpse into some of the major approaches in the literature.
3 Findings From Experimental Work
In one of the earliest reviews on the use of field experiments to study governance and government performance, Humphreys and Weinstein (2009) identify four major questions on which researchers have focused: (1) what is the role of political institutions in the process of decision-making and policy implementation? (2) How do social norms and informal institutions affect individual and collective action? (3) What is the impact of information and incentives on political behaviour, notably accountability? And (4) how can violence and conflict be prevented? The authors cover a limited number of studies and acknowledge that “there has not yet been a significant accumulation of knowledge from the use of field experiments in the political economy of development […] For this reason, we focus more on the promise of the field than on its achievements” (Humphreys and Weinstein 2009: p. 370).
In a subsequent review of experimental research on democratic governance, Moehler (2010: p. 30) addresses the related question of whether field experiments can “be productively employed to study the impact of development assistance on democracy and governance outcomes?” She highlights several key weaknesses of field experiments, but is generally sanguine about the possibilities: “The enterprise of DG field experiments,” she notes, “will be constrained more by mundane challenges to successful research design and implementation than by the inherent limitations of field experiments” (Moehler 2010: p. 42). Her review identifies 41 randomized field experiments of interest in the developing world, including eleven dealing with elections, ten with community-driven development, nine with government performance in public service delivery, three with the use of quotas, and seven with other topics. The majority of the reported studies (22) were conducted in Africa, and nine in India.
More recently, Olken and Pande (2011b) conducted a narrative, non-systematic review of the literature on governance, following a principal-agent approach. They include in the review 16 studies that adopt rigorous experimental and non-experimental methods to establish causality in the analysis of policies that aim to improve governance in developing countries, dividing the literature into two areas: (1) participation and participatory institutions to exercise greater control over politicians, and (2) the roots of corruption and the incentives and institutional features that can prevent rent-seeking behaviour and leakages.1
Recent studies also review findings from RCTs with respect to development interventions more generally. Banerjee and Duflo (2012), for instance, draw largely on the results of their work at the MIT Poverty Action Lab to propose new solutions to global poverty, highlighting the role of “ideology, ignorance, and inertia” in explaining why aid is not always effective. Many of their solutions point to how the poor lack critical information and hold incorrect beliefs (e.g., about the benefits to education) that help to perpetuate their poverty. They highlight findings from RCTs dealing with hunger, health, education, family planning, risk management, microfinance, and entrepreneurship. Karlan and Appel (2012) build a similar argument about solutions to global poverty, also drawing heavily on findings from RCTs. They highlight “seven ideas that work:” microsavings, reminders to save, prepaid fertilizer sales, deworming, remedial education in small groups, chlorine dispensers for clean water, and commitment devices (Karlan and Appel 2012: pp. 272–275).
Building on this earlier work and in order to address potential threats of publication bias, a systematic review of published and unpublished papers using experimental and quasi-experimental methods to study government performance was conducted as part of the research for the broader project of which this symposium is a part.2 From an initial sample of over 3000 papers on governance-related topics released between 1990 and 2012, the review identified and analyzed 139 papers judged relevant based on topic and methods employed. Building on the approach outlined above, the review includes both studies focused on the provision of public services and on aggregating interests in various ways. As summarized in Figure 1, studies are further classified based on the type of policy intervention, that is, whether it primarily aims to (1) improve supply-side capabilities of governments, and the social and political institutions that facilitate that process; (2) change individual behaviour through various devices, notably incentives, and/or (3) improve informational asymmetries. Interventions of the first type focus on the supply-side dimensions of policies, affecting how public institutions themselves provide public goods and services. Improving government performance in this context may involve changes both to what is provided (e.g., books, classrooms) and their quality (e.g., better books, better teachers). The second and third types of interventions directly influence the demand-side of government performance, i.e., how the population (usually individuals, households, and communities) interact with public institutions. Demand-side interventions are found either to provide incentives (often in cash) or to provide better information about the provision of goods and services, both with the objective of changing individual behavior. Finally, studies are grouped into commonly-referenced policy areas (e.g., social protection, health and education, democracy, accountability and corruption).
As shown in Figure 1, under the “provision of public goods” cluster, the largest number of studies (41) focuses on health care and education policies, with fewer addressing issues of employment, water and sanitation, housing, and other topics. Under “aggregating interests” particular attention is paid to institution building (26), particularly in the context of improving supply-side capabilities (17), and to electoral participation and voting behavior (24), particularly in the context of improving information asymmetries (19). The largest number of studies in our sample was conducted in the USA and India (see Table A1 in the Appendix)
It also is worth highlighting the challenge of drawing sharp distinctions in this body of work between experimental and quasi-experimental studies. In particular, our review of the literature reveals that a significant number of studies that adopted experimental designs had to resort to quasi-experimental regression techniques such as propensity score matching and instrumental variables techniques to address errors in implementation that caused endogeneity problems, spillovers, and sample contamination (see Table A1 in the Appendix).
A number of findings emerge from these studies that are relevant to explaining variations in government performance. As above, many of these relate to the ways in which governments (or donors) can improve the provision of basic public goods, particularly in the areas of health care and education and deal with the impact of projects providing specific goods or services. In their study of the Primary School Deworming Project in Kenya, for instance, Miguel and Kremer (2004) find that the program not only improved students’ health in both treatment schools and neighboring schools, but also reduced school absenteeism by a quarter (although there was no evidence of an effect on academic test scores). In demonstrating the impact of expanded insurance coverage on improved health outcomes among children, Quimbo et al. (2011) draw on the Quality Improvement Demonstration Study in the Philippines to show that zero co-payments and increased enrolment were associated after release from the hospital with reduced likelihood of wasting and of having an infection (9–12 and 4–9%, respectively). Kremer et al. (2009) evaluate the impact of a merit scholarship program in Kenya in which girls who scored well on exams had school fees paid and received a grant, finding that the program had an effect not only on improved student test scores, but also on teacher attendance.3
A number of studies explore the impact of public information campaigns on public goods provision. Pandey et al. (2009), for instance, evaluate the impact of a community-based information campaign across three Indian states consisting of eight or nine public meetings to disseminate information to communities about its state-mandated roles and responsibilities in school management. They find the largest impacts on teacher effort, and more modest improvements on student learning and the delivery of benefits to students (stipends, uniforms, and mid-day meal). Also in India, Pattanayak et al. (2009) explore the impact of the intensified “information, education, and communication” campaign carried out in Orissa as part of the nationwide Total Sanitation Campaign to change rural household attitudes about the use of latrines. The study found that latrine ownership rose significantly in treatment villages and remained the same in control villages. Pattanayak et al. (2009) further address the question of whether social and emotional costs (“shaming”) or financial incentives (“subsidies”) better influence behaviour. They find that although latrine ownership rose most among households below the poverty line and eligible for a government subsidy (5–36%), it also rose among wealthier households not eligible for the subsidy (7–26%), suggesting that shaming, even in the absence of subsidies, can work to change behaviour.
Conditional cash transfers as a strategy have received particular attention and been evaluated in several different contexts. A number of studies focus on Mexico’s Progresa/Oportunidades program (e.g., Stecklov et al. 2007; De La O 2013). Leroy et al. (2008) for instance, find the program to be associated with better growth in infants below 6 months of age (but to have no impact for babies 6–24 months). Other studies explore the impact of conditional cash penalty programs. One example is Dee’s (2011) study of the effects in ten counties of the state of Wisconsin’s Learnfare program, which sanctions a family’s welfare grant when teenagers in the family do not meet school attendance targets. Data suggest evidence in nine counties that Learnfare increased school enrolment by 3.5% and attendance by 4.5%.
Another set of studies focus on interventions to improve the performance of public sector employees such as teachers and nurses. Multiple studies highlight the impact of financial incentives. Duflo and Hanna (2005), for instance, find that a financial incentive program immediately reduced teacher absenteeism in rural India, which was also associated with an improvement in student test scores and achievement 1 year after the start of the program. Basinga et al. (2011) find in Rwanda that adoption of performance-based payment of health-care providers (“P4P”) was related to improvements in the use and quality of child and maternal care services, including a 23% increase in the number of institutional deliveries and increases in the number of preventive care visits by children (56% for those 23 months and younger, and 132% for those 24–59 months), and improvements in prenatal quality as measured by compliance with Rwandan prenatal care clinical practice guidelines.4 Other studies explore the impact of relatively minor administrative reforms: Banerjee et al. (2012), for instance, test the impact of four low-cost reforms across police stations in eleven districts in Rajasthan. Results suggest that two of these reforms – freezing staff transfers between police stations and providing in-service training in investigation skills and “soft” skills like communication and leadership – were effective in improving police effectiveness and public satisfaction, while the other two reforms – placing community observers in police stations and a weekly duty rotation – were not effective.
A growing body of experimental and quasi-experimental work also studies issues related to aggregating interests through electoral politics in new and emerging democracies. Fujiwara and Wantchekon (2013), for instance, explores whether public deliberation – in the form of town meetings – can overcome clientelism in Benin. The experimental data show a positive effect on perceived knowledge about policies and candidates and on voter turnout, as well as increased electoral support for the candidates participating in the intervention. Collier and Vicente (2008) evaluate the effect of a campaign against political violence run by an NGO in Nigeria, involving town meetings, popular theatres, and door-to-door distribution of material. They find that this intervention served to reduce the intensity of election-related violence. Hyde (2010a) shows that the presence of election observers had an effect on election quality in the 2004 Indonesian presidential elections, measured in terms of votes cast for the incumbent. Ichino and Schündeln (2012) study the effect of domestic observers on voter registration in Ghana in 2008. They find that because parties operate over large areas, observers in one registration center may displace irregularities to others, which suggests the need for some revisions to how such observers are deployed in many countries.
Finally, a number of studies explore topics at the intersection of representation and public service provision, with particular attention to the impact of community-based monitoring initiatives. Björkman and Svensson (2009), for instance, find in Uganda that holding meetings among community members and health workers to discuss health services and how to improve them, to compare citizen and health worker views of service provision, and to collectively discuss patient rights and provider responsibilities, led to improved health outcomes (reduced child mortality and increased child weight), as well as more community monitoring of health care a year after the intervention. Olken (2010) explores the relationship between direct democracy and local public goods provision in rural Indonesia, studying plebiscites introduced in some villages to replace a meeting-based process presumably dominated by elites. Plebiscites were associated with higher public satisfaction and perceived benefits from the project, greater willingness to contribute, and increased knowledge about the project. On the other hand, Olken’s (2007) study of “top down” versus grassroots participation in corruption monitoring in Indonesia suggests that government-led approaches may be the more effective on this issue. Increasing government audits had a significant effect on reducing corruption in term of reducing missing expenditures and discrepancies between official project costs and independent estimates of costs, while increasing grassroots participation had little impact.
In summary, findings using experimental data highlight an array of strategies that governments can adopt to improve public service provision and representation and accountability in particular areas. Innovations that have been explored in multiple contexts include public information campaigns, conditional cash transfers, financial incentives to improve the performance of public sector employees, community-based monitoring, and public deliberation at the local level.
4 The Limits of Experimental Methods in the Study of Government Performance
The elegance of RCT findings arguably has a tendency to promote method-driven, rather than theory-driven, research: it tends to encourage work that asks questions that can be addressed with experiments, rather than work that begins with questions that are seen as important to answer and then proposes hypotheses and assesses the methods most appropriate for testing them.
One key criticism levelled at experimental work is that it does not address “big” questions and “big” theories (Hyde 2010b). Indeed, if we compare the factors explored in the experimental studies reviewed in Section 3 with those identified in major theories of government performance, there is clearly a disconnect. Key factors in major theories of governance, such as modernization, social structure, and national institutions, for instance, are largely absent as an object of study in experimental evaluations.
Proponents of RCTs, however, make a compelling argument that their avoidance of grand theory can be a strength as well. Banerjee and Duflo (2012), for one, explicitly advocate for an incremental, micro approach. Their solutions posit that government functioning can be improved with small policy reforms that at the margin can lead to desirable improvements in policy, without major changes to social and political structures. Karlan and Appel (2012: p. 37) contend that development “needs to be on the ground,” “up in the realm of high-minded concepts…the air is thin and there are no poor people to be found.”
Still, despite such explicit rejection of grand theory, experimental approaches are not absent of theoretical underpinnings. Analysis often falls clearly within the rigor of behavioral economics, drawing implicitly on its theories of individual behaviour, (ir)rational choice, and information. Nevertheless, this micro focus exacerbates a second of the key weaknesses of experimental work: the low external validity of its findings. If findings from RCTs are to be used to identify generalizable impacts – to help predict the impact of similar interventions in other situations – experimental work must be able to say something about the broader context. Precisely because experimental researchers tend to adopt a micro approach to research enquiry, and eschew more high level theorizing about what within particular contexts might be unique or have influenced results, experimental studies tend to provide weak leverage on the question of whether similar outcomes might be expected in other contexts (see Pritchett and Sandefur 2013).
One strategy for improving external validity in experimental research is to speak more directly to broader theoretical propositions. Martel Garcia and Wantchekon (2015) in this Issue draw on structural theory to advance our understanding of external validity and generalization of causal problems. They show that external validity is a function of theoretical generalization, i.e., the ability of researchers to explain and predict outcomes across variations in treatments, outcomes, and settings. However, even when adopting such approaches, a degree of uncertainty remains with regard to the underlying mechanisms that explain, under a theoretical framework, the distribution of policy outcomes for a particular group (treatment and control) vis-à-vis the distribution for the entire population. This constraint forces us to look beyond experimental methods alone in the study of government performance.
Another strategy to improve the external validity in experimental research involves replicating the same type of intervention in different sets of conditions. For example, the Consultative Group to Assist the Poor (CGAP) and the Ford Foundation have funded the replication of Bangladesh’s BRAC’s Challenging the Frontiers of Poverty Reduction/Targeting the Ultra Poor program in India, Pakistan, Honduras, Peru, Ethiopia, Yemen, and Ghana, using experimental designs for their evaluations. The “Metaketa” projects, such as the new initiative on political information and electoral choices carried out in Benin, Brazil, Burkina Faso, India, Mexico, and Uganda are another example.5 Further, with the rapid accumulation of experimental and quasi-experimental evidence, systematic reviews and meta-regression analyses have been increasingly used to produce generalizations about specific policy interventions and their effects under different socio-economic settings. This process has been supported by a number of organizations, including the Cochrane Collaboration, Campbell Collaboration, and International Initiative for Impact Evaluation (3ie).
A third limit to RCTs in the study of government performance is in the type of causal factors that they can reasonably study. This constraint follows partly from the need for large numbers of units to be studied in order to gain precise estimates, which encourages researchers to focus on low level factors, rather than on factors held by higher level units, such as national institutions (Moehler 2010). Some traction on such factors can be gained by “scaling up” findings from low level factors. For instance, some findings from studies of public administrative reforms carried out at the municipal level may inform reform at the national level. However, municipal versus national politics are so different in other ways that this sort of scaling up clearly provides only suggestive evidence.
Experimental evaluation protocols have nonetheless proved to be relevant in contexts where incumbent governments and political elites need to be persuaded to scale up social policies. Barrientos and Villa (2015) in this Issue analyze a new database on antipoverty social protection programs in Latin America and sub-Saharan Africa, documenting the influence of political competition, as well as the rise of “evidence-based” development policy, in explaining the adoption of such evaluations. They find that while in Latin America impact evaluation of antipoverty programs has been driven by agency competition, design features of programs, and political factors, in sub-Saharan Africa, the interactions between donors and domestic elites have undermined the demand for evaluations.
The limits to the causal factors that RCTs can study also follow from the simple inability of researchers to manipulate some key variables identified in the literature, such as the level of development, national institutions, culture, or the quality of national leadership. Putnam’s (1993) study of new regional governments in Italy serendipitously gave him a natural experiment to exploit, but such situations of “natural” random assignment may be both rare and problematic to analyze (Sekhon and Titiunki 2012).
In other cases, ethical considerations may impede the study of particular factors and outcomes. For instance, interventions expected to foment electoral violence or to influence which candidate wins an election are not undertaken for obvious ethical reasons. As Humphreys (2015) in this Issue notes, more work remains to be done in this area by researchers in the social sciences as the main principles of research ethics currently employed were developed by and for medical research. Humphreys provides a detailed discussion and critique of ethics as currently understood in work in this area and of approaches to consent that can inform the further development of ethical guidelines.
A fourth issue that limits the utility of RCTs in the social sciences is their relatively short-term window of analysis. This is particularly problematic in the study of government performance because so many of our major theories focus on “non-linear” processes that evolve over decades or generations, while RCTs rarely look at impacts beyond the “linear” trajectory between two points in time, usually a few years. Take, for example, the hypothetical case of a J-shaped curve derived from the long-term relationship between economic liberalization and political stability: in the short-term, economic liberalization may lead to a sudden rupture between economic and political actors that cause an increase in political instability. An RCT may conclude that economic liberalization is bad for political stability. However, if theory’s predictions are correct, once markets and institutions are developed further, political stability would actually improve (Gans-Morse and Nichter 2008). Although the time horizons of RCTs could be extended somewhat, they would still not be long enough to explore many of the major theories of governance.
One approach that may help to extend the time horizon is suggested by the use of what Baldwin and Bhavnani (2015) in this Issue call “ancillary experiments.” As more RCTs become available in the social sciences, opportunities arise to use existing experimental data to investigate new questions. Researchers can collect new data on populations assigned to treatment and control groups in previously executed experiments, and then rely on the initial randomization to identify new effects. Ancillary experiments also provide many of the advantages of RCTs but at lower cost, since the intervention has already been undertaken. A classic example of ancillary experiments is the study by Angrist (1990), who took advantage of the Vietnam draft lottery to examine the effects of military service on lifetime earnings.
A fifth key concern with RCTs is that they are similarly limited in terms of the unit of analysis upon which they can evaluate impacts, which is generally the individual. Some studies focus on other units of analysis, such as voting constituencies or local regions, but no studies of which we are aware conduct experiments with representative samples at the national level. This is simply due to the fact that the treatment effects arising from policy interventions are often small, and therefore large sample sizes are needed to conclude, with enough statistical power, that the differences between the treatment and control groups are unlikely to be due to chance. This connects to a final issue: the cost of RCTs.
Some scholars have questioned whether RCTs are worth the cost, which often involve million dollar budgets (Heckman and Smith 1995). Randomization by group or cluster is often used in medical science to lower the cost of RCTs via phased implementation. This approach significantly decreases the cost of running studies, particularly in contexts where the outcomes of interest are easily assessed; however, even if they could be adapted to address some key theories of government, it is not necessarily clear whether they would be more cost-effective in testing these theories than observational methods.
Observational studies, in contrast, use econometric methods each with its own assumptions, strengths, and weaknesses to estimate the impact of policy innovations and use data ranging from small surveys to nationally representative surveys and censuses. From the perspective of experimental design, observational studies have been subject to criticism on the basis of their internal validity. Dehejia (2015) provides in this Issue a survey of observational methods, including regression analysis, instrumental variables, difference-in-difference estimators, regression discontinuity, and matching, and assessed their internal and external validity relative both to each other and to RCTs. He concludes that both experimental and non-experimental methods face important challenges, but can sometimes be fruitfully combined.
Indeed, although experimental and observational methods are often presented as mutually exclusive, they are in fact complementary in two senses. First, the principles of experimental design can improve the internal validity of observational studies. Survey-based initiatives to study governance such as the Afrobarometer project for instance have begun to explore such strategies (Bratton 2013). Second, they can be combined using non-experimental adjustments to deal with implementation failures in RCTs. As our review of the literature in Section 3 suggests, experimental studies in the fields of development economics and political science – arguably more so than in medicine – can pose significant logistical and methodological challenges that result in reliance on quasi-experimental regression techniques to tackle problems such as cofounding, selection bias, spillovers, and impact heterogeneity (Deaton 2009).
This study argues that RCTs have been and can be useful in studying the effects of some policy interventions and reforms, but that their use in the study of government performance is also limited in significant ways, particularly by the nature of the factors that we expect to matter most. RCTs are best for studying targeted interventions (particularly in areas of public goods provision, voting behaviour, and specific measures to address corruption and improve accountability), which are expected to have rapid results. However, our review of theories of government and the state suggests that what might be most important in explaining variation in outcomes are broad, macro-structural shifts, national level variation in institutions and political culture, and the actions of individual policy decision-makers.
The focus in this article has been on the use of RCTs to test hypotheses about the causes of variation in government performance. One might argue that in fact a narrower focus is warranted, highlighting precisely the individual, household, or community-level factors that policy makers might leverage to improve government performance. In other words, RCTs may have particular resonance for a more policy-focused audience as the sort of interventions best studied with RCTs may be precisely those most readily manipulable by donors and domestic policy makers.
However, given the limits of RCTs explored in this study, it is clear that even within a narrower policy-focused frame, analysts must go beyond experimental methods to address the questions of “what works” and “what could work.” Small-scale interventions may be most readily manipulable by policy makers, but policy makers can also focus attention on influencing change in a number of larger-scale factors identified in the literature as potentially important to improving government performance, such as national institutions and social capital, that RCTs are ill-equipped to study. Thus, policy makers might look beyond experimental studies to inform their reform strategies in this sense. In addition, even with respect to reforms and interventions that have been explored through RCTs, the weak external validity of RCTs raises major questions about whether policy makers in other contexts should expect to see the same results, that is, whether they can be transferred successfully.
Finally, replication, systematic reviews, and meta-analysis may help in accreting knowledge using an array of data sources and methods. The discussion presented in Section 3 relies on a growing literature which employs both experimental and quasi-experimental methods. These studies have been examined following systematic review methodologies to provide a rigorous synthesis on the impact of various interventions and reforms to improve government performance. This synthesis itself speaks to a middle ground between experimental and non-experimental methods that forms the common theme of this symposium.
This UNU-WIDER symposium was developed under the project on “Experimental and Non-experimental Methods to Study Government Performance: Contributions and Limits,” which was supported under WIDER’s Research and Communication on Foreign Aid (ReCom) program (2011–2013). We gratefully acknowledge specific program contributions from the governments of Denmark (Ministry of Foreign Affairs, Danida) and Sweden (Swedish International Development Cooperation Agency – Sida) for ReCom, as well as core financial support to the UNU-WIDER work program from the governments of Denmark, Finland, Sweden, and the UK. We are grateful to Armando Barrientos, Kate Baldwin, Rikhil Bhavnani, Michael Bratton, Rajeev Dehejia, Macartan Humphreys, Fernando Martel Garcia, Javier Sajuría, Juan M. Villa, Leonard Wantchekon, and the Editors for helpful comments and suggestions on early versions of this article. The errors are ours alone.
Angrist, Joshua D. (1990) “Lifetime Earnings and the Vietnam Era Draft Lottery: Evidence from Social Security Administrative Records,” The American Economic Review, 80(5):1284–1286.Google Scholar
Arndt, Christiane and Charles Oman (2006) Uses and Abuses of Governance Indicators. Paris: OECD Development Centre.Google Scholar
Baldwin, Kate and Rikhil R. Bhavnani (2015) “Ancillary Studies of Experiments: Opportunities and Challenges”. Journal of Globalization and Development, 6(1): 113–146.Google Scholar
Banerjee, Abhijit and Esther Duflo (2012) Poor Economics: A Radical Rethinking of the Way to Fight Global Poverty. New York: PublicAffairs.Google Scholar
Banerjee, Abhijit V., Raghabendra Chattopadhyay, Esther Duflo, Daniel Keniston and Nina Singh (2012) Can Institutions Be Reformed from Within? Evidence from a Randomized Experiment with the Rajasthan Police. Working Paper 17912. Cambridge, MA: NBER.Google Scholar
Banerjee, Abhijit Vinayak (2007) “Making Aid Work.” In: (A. V. Banerjee, ed.) Making Aid Work. Cambridge, MA: MIT Press.Google Scholar
Bardhan, Pranab (2005) Scarcity, Conflicts and Cooperation: Essays in the Political Economy and Institutional Economics of Development. Cambridge, MA: MIT Press.Google Scholar
Barrientos, Armando and Juan M. Villa (2015) “Evaluating Antipoverty Transfer Programmes in Latin America and Sub-Saharan Africa. Better Policies? Better Politics?,” Journal of Globalization and Development, 6(1): 147–179.Google Scholar
Basinga, Paulin, Paul J. Gertler, Agnes Binagwaho, Agnes L. B. Soucat, Jennifer Sturdy and Christel M. J. Vermeersch (2011) “Effect on Maternal and Child Health Services in Rwanda of Payment to Primary Health-Care Providers for Performance: An Impact Evaluation,” The Lancet, 377:1421–1428.CrossrefGoogle Scholar
Björkman, Martina and Jakob Svensson (2009) “Power to the People: Evidence from a Randomized Field Experiment on Community-Based Monitoring in Uganda,” The Quarterly Journal of Economics, 124:735–769.CrossrefGoogle Scholar
Bovaird, Tony and Elke Loffler, eds. (2009) Public Management and Governance, 2nd ed. Abingdon: Routledge.Google Scholar
Bratton, Michael (2013) Measuring Government Performance in Public Opinion Surveys in Africa: Towards Experiments? WIDER Working Paper. Helsinki: UNU-WIDER.Google Scholar
Collier, Paul and Pedro C. Vicente (2008) Votes and Violence: Evidence from a Field Experiment in Nigeria. CSAE WPS/2008-16. Oxford: University of Oxford.Google Scholar
Conn, K. (2014) Identifying Effective Education Interventions in Sub-Saharan Africa: A Meta-Analysis of Rigorous Impact Evaluations. Ph.D. dissertation, Columbia University.Google Scholar
De La O. Ana (2013) “Do Conditional Cash Transfers Affect Electoral Behavior? Evidence from a Randomized Experiment in Mexico.” American Journal of Political Science, 57(1):1–14.CrossrefGoogle Scholar
Deaton, Angus (2009) “Instruments of Development: Randomization in the Tropics, and the Search for the Elusive Keys to Economic Development.” Research Program in Development Studies, Center for Health and Wellbeing. Princeton, NJ: Princeton University.Google Scholar
Dehejia, Rajeev (2015) “Experimental and Non-Experimental Methods in Development Economics: A Porous Dialectic,” Journal of Globalization and Development, 6(1): 47–69.Google Scholar
Duflo, Esther and Rema Hanna (2005) Monitoring Works: Getting Teachers to Come to School. Working Paper Series 11880. Cambridge, MA: NBER.Google Scholar
Dunning, T., G. Grossman, M. Humphreys, S. Hyde, C. McIntosh, C. Adida, E. Arias, T. Boas, M. Buntaine, S. Bush, S. Chauchard, J. Gottlieb, F. D. Hidalgo, M. E. Holmlund, R. Jablonski, E. Kramon, H. Larreguy, M. Lierl, G. McClendon, J. Marshall, D. Nielson, M. Platas Izama, P. Querubin, P. Raffler and N. Sircar (2015). “Political Information and Electoral Choices: A Pre-meta-analysis Plan.” Available at: http://egap.org/research/metaketa/.
Friedman, Milton (1962) Capitalism and Freedom. Chicago: University Of Chicago Press.Google Scholar
Fujiwara, Thomas, and Leonard Wantchekon (2013) “Can Informed Public Deliberation Overcome Clientelism? Experimental Evidence from Benin.” American Economic Journal: Applied Economics, 5(4): 241–55.CrossrefGoogle Scholar
Gisselquist, Rachel M. (2012) Good Governance as a Concept, and Why this Matters for Development Policy. Working Paper 2012/30. Helsinki: UNU-WIDER.Google Scholar
Gisselquist, Rachel M., Miguel Niño-Zarazúa and Javier Sajuria. (2013) “Improving Government Performance in Developing Countries. A Systematic Review.” Unpublished paper, UNU-WIDER, Helsinki.Google Scholar
Glennerster, Rachel and Michael Kremer (2011) Small Changes, Big Results: Behavioral Economics at Work in Poor Countries. Boston Review: A Political and Literary Forum.Google Scholar
Grindle, Merilee S. (2010) Good Governance: The Inflation of an Idea. Faculty Research Working Paper Series RWP 10-023. Cambridge, MA: Harvard Kennedy School.Google Scholar
Grindle, Merilee S. and John W. Thomas (1991) Public Choices and Policy Change: The Political Economy of Reform in Developing Countries. Baltimore: John Hopkins University Press.Google Scholar
Herbst, Jeffrey (2000) States and Power in Africa: Comparative Lessons in Authority and Control. Princeton: Princeton University Press.Google Scholar
Horowitz, Donald L. (1985) Ethnic Groups in Conflict. Berkeley: University of California Press.Google Scholar
Humphreys, Marcatan (2015) “Reflections on the Ethics of Social Experimentation,” Journal of Globalization and Development, 6(1):87–112.Google Scholar
Hyde, Susan D. (2010a) “Experimenting in Democracy Promotion: International Observers and the 2004 Presidential Elections in Indonesia,” Perspectives on Politics, 8:511–527.Google Scholar
Hyde, Susan D. (2010b) “The Future of Field Experiments in International Relations,” The ANNALS of the American Academy of Political and Social Science, 628:72–84.Google Scholar
Ichino, Nahomi and Matthias Schündeln (2012) “Deterring or Displacing Electoral Irregularities? Spillover Effects of Observers in a Randomized Field Experiment in Ghana,” The Journal of Politics, 74:292–307.CrossrefGoogle Scholar
Immergut, Ellen M. (1992) Health Politics: Interests and Institutions in Western Europe. Cambridge: Cambridge University Press.Google Scholar
Jackson, R. H. and C. G. Rosberg (1982) Personal Rule in Black Africa. Los Angeles: UCLA Press.Google Scholar
Karlan, Dean and Jacob Appel (2012) More Than Good Intentions: Improving the Ways the World’s Poor Borrow, Save, Farm, Learn, and Stay Healthy. New York: Plume.Google Scholar
Kaufmann, Daniel, Aart Kraay and Pablo Zoido-Lobaton (1999) Governance Matters. Policy Research Working Paper 2196. Washington, DC: World Bank.Google Scholar
Keefer, Philip (2009) “Governance.” In: (T. Landman and N. Robinson, eds.) The Sage Handbook of Comparative Politics. London: Sage.Google Scholar
Keynes, John M. (1964) The General Theory of Employment, Interest, and Money. London: Harcourt Brace Jovanovich.Google Scholar
Kingdon, John W. (1995) Agendas, Alternatives, and Public Policies, 2nd ed. New York: HarperCollins.Google Scholar
Krishnaratne, S., H. White and E. Carpenter (2013) Quality Education for all Children? What Works in Education in Developing Countries. 3ie Working Paper. New Delhi: International Initiative for Impact Evaluation (3ie).Google Scholar
Leroy, Jef L., Armando García-Guerra, Raquel García, Clara Dominguez, Juan Rivera and Lynnette M. Neufeld (2008) “The Oportunidades Program Increases the Linear Growth of Children Enrolled at Young Ages in Urban Mexico,” The Journal of Nutrition, 138:793–798.Google Scholar
Levi, Margaret (2006) “Why We Need A New Theory of Government,” Perspectives on Politics, 4:5–19.Google Scholar
Lijphart, Arend (1977) Democracy in Plural Societies: A Comparative Exploration. New Haven: Yale University Press.Google Scholar
Lipset, S. M. (1981) Political Man: The Social Bases of Politics. Baltimore, MD: Johns Hopkins University Press.Google Scholar
Luebbert, Gregory M. (1991) Liberalism, Fascism, or Social Democracy: Social Classes and the Political Origins of Regimes in Interwar Europe. Oxford: Oxford University Press.Google Scholar
Masino, S. and M. Niño-Zarazúa (2015) What Works to Improve the Quality of Student Learning in Developing Countries? WIDER Working Paper 2015 (033).Google Scholar
Martel Garcia, Fernando and Leonard Wantchekon (2015) “A Graphical Approximation to Generalization: Definitions and Diagrams,” Journal of Globalization and Development, 6(1): 71–86.Google Scholar
McEwan, P. J. (in press) “Improving Learning in Primary Schools of Developing Countries: A Meta-Analysis of Randomized Experiments,” Review of Educational Research.Google Scholar
Migdal, Joel (1988) Strong Societies and Weak States: State-Society Relations and State Capabilities in the Third World. Princeton: Princeton University Press.Google Scholar
Moehler, Devra C. (2010) “Democracy, Governance, and Randomized Development Assistance,” The ANNALS of the American Academy of Political and Social Science, 628:30–46.Google Scholar
Moore, Barrington (1966) Social Origins of Dictatorship and Democracy: Lord and Peasant in the Making of the Modern World. New York: Beacon Press.Google Scholar
Mukunda, G. (2012) Indispensible: When Leaders Really Matter. Cambridge: Harvard Business Review Press.Google Scholar
North, Douglass C. (1990) Institutions, Institutional Change and Economic Performance. Cambridge, MA: Cambridge University Press.Google Scholar
Olken, Benjamin A. and Rohini Pande (2011a) Corruption in Developing Countries. Working Paper 17398. Cambridge, MA: NBER.Google Scholar
Olken, Benjamin A. and Rohini Pande (2011b) Governance Review Paper. J-PAL Governance Initiative: Abdul Latif Jameel Poverty Action Lab.Google Scholar
Pandey, Priyanka, Sangeeta Goyal and Venkatesh Sundararaman (2009) “Community Participation in Public Schools: Impact of Information Campaigns in Three Indian States,” Education Economics, 17:355–375.CrossrefGoogle Scholar
Pattanayak, Subhrendu K., Jui-Chen Yang, Katherine L. Dickinson, Christine Poulos, Sumeet R. Patil, Ranjan K. Mallick, Jonathan L. Blitstein and Purujit Praharaj (2009) “Shame or Subsidy Revisited: Social Mobilization for Sanitation in Orissa, India,” Bulletin of the World Health Organization, 87:580–587.CrossrefGoogle Scholar
Petrosino, A., C. Morgan, T. Fronius, E. Tanner-Smith and R. Boruch (2012) “Interventions in Developing Nations for Improving Primary and Secondary School Enrollment of Children: A Systematic Review,” Campbell Systematic Reviews, 19:1–191.Google Scholar
Pritchett, Lant and Justin Sandefur (2013) Context Matters for Size: Why External Validity Claims and Development Practice Don’t Mix. Working Paper 336. Washington, DC: Center for Global Development.Google Scholar
Przeworski, Adam, Michael E. Alvarez, Jose Antonio Cheibub and Fernando Limongi (2000) Democracy and Development: Political Institutions and Well-being in the World, 1950–1990. New York: Cambridge University Press.Google Scholar
Putnam, Robert (1993) Making Democracy Work: Civic Traditions in Modern Italy. New York: Princeton University Press.Google Scholar
Quimbo, Stella A., John W. Peabody, Riti Shimkhada, Jhiedon Florentino and Orville Solon (2011) “Evidence of a Causal Link Between Health Outcomes, Insurance Coverage, and a Policy to Expand Access: Experimental Data from Children in the Philippines,” Health Economics, 20:620–630.CrossrefGoogle Scholar
Ravallion, Martin (2009) “Should the Randomistas Rule?” Economists’ Voice, 6:1–5.Google Scholar
Reilly, Benjamin (2001) Democracy in Divided Societies: Electoral Engineering for Conflict Management. Cambridge: Cambridge University Press.Google Scholar
Robinson, Mark (2007) “Does Decentralisation Improve Equity and Efficiency in Public Service Delivery Provision?,” IDS Bulletin, 38(1):7–17.Google Scholar
Samuels, Richard J. (2003) Machiavelli’s Children: Leaders and Their Legacies in Italy and Japan. Ithaca: Cornell University Press.Google Scholar
Sartori, Giovanni (1997) Comparative Constitutional Engineering: An Inquiry into Structures, Incentives, and Outcomes. New York: New York University Press.Google Scholar
Seguin, M. and M. Niño Zarazúa (2015) “Non-clinical Interventions for Acute Respiratory Infections and Diarrhoeal Diseases Among Young Children in Developing Countries,” Tropical Medicine & International Health, 20(2):146–169.Google Scholar
Skocpol, Theda (1979) States and Social Revolutions: A Comparative Analysis of France, Russia and China. Cambridge: Cambridge University Press.Google Scholar
Stecklov, Guy, Paul Winters, Jessica Tood and Ferdinando Regalia (2007) “Unintended Effects of Poverty Programmes on Childbearing in Less Developed Countries: Experimental Evidence from Latin America,” Population Studies, 61:125–140.CrossrefGoogle Scholar
Steinmo, Sven (2008) “Historical Institutionalism.” In: (D. Della Porta and M. Keating, eds.) Approaches and Methodologies in the Social Sciences: A Pluralist Perspective. Cambridge: Cambridge University Press.Google Scholar
Thomas, M. A. (2009) “What Do the Worldwide Governance Indicators Measure?” European Journal of Development Research, 22:31–54.Google Scholar
Tocqueville, Alexis de (2003) Democracy in America and Two Essays on America. Translated by Gerald E. Bevan. London: Penguin Books.Google Scholar
Weber, Max (2009) “Politics as a Vocation.” In: (H. H. Gerth and C. Wright Mills, eds.) From Max Weber: Essays in Sociology. Abingdon: Routledge.Google Scholar
World Bank (2012) Strengthening Governance: Tackling Corruption. The World Bank Group’s Updated Strategy and Implementation Plan. Washington, DC: World Bank.Google Scholar
For a detailed description of the systematic review protocol, see Gisselquist et al. (2013). One issue worth highlighting is that some papers may be classified in multiple ways. For the purposes of our discussion here, we chose a single classification based on our assessment of the most relevant category.
About the article
Published Online: 2015-05-16
Published in Print: 2015-06-01
Citation Information: Journal of Globalization and Development, Volume 6, Issue 1, Pages 1–45, ISSN (Online) 1948-1837, ISSN (Print) 2194-6353, DOI: https://doi.org/10.1515/jgd-2014-0011.
©2015, Miguel Niño-Zarazúa et al., published by De Gruyter. This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 3.0 License. BY-NC-ND 3.0