Publication | U4 Issue

How change happens in anti-corruption

A map of policy perspectives

Connecting practice with policy paradigms

Debates about the nature and consequences of corruption, why it persists, and how it can be managed have thrived over the past couple of decades. Much less in evidence has been systematic thinking about the most effective ways that countries, sectors, or organisations can achieve sustainable transitions away from endemic corruption and towards consistent integrity. This is perhaps surprising, given the vast sums of money that have been invested in anti-corruption efforts and the preponderance of evidence showing that mainstream approaches have not fully delivered.

One possible reason for the disappointing results of many interventions is that their design did not fully reflect an understanding of how change happens.ac5fd4df09c8 Anti-corruption programmes may have had theories of action, which define planning processes related to a single intervention, but these are not equivalent to theories of change. A true theory of change – a pathway for how reforms will induce changes in complex internal dynamics, leading to larger endogenous change – has been largely lacking from anti-corruption policy design.3a3a590ebe9f The failure to see anti-corruption as a process involving a complex sequence of steps may reflect the tendency of the donor community to undervalue the complex, highly contingent, and contentious processes by which their own countries transitioned (often only recently) away from widespread corruption.97ec6e110b2b

Nonetheless, in recent years there has been some deliberate thinking applied to these questions about the most effective pathways out of widespread corruption, and some of this has challenged the tenets of conventional approaches. As much of this innovative thinking remains somewhat fragmented, the aim of this U4 Issue is to draw out connections between the various strands in order to present the key contours of anti-corruption policy thought that relates to theories of change.

This paper therefore zooms out to look at the larger picture. At the same time, of course, improving anti-corruption requires us to zoom in on specific interventions in an attempt to evaluate which ones are effective and under what circumstances. But these micro-level perspectives need to be situated within a broader framework of change. Seeing the contours of the forest may be just as important as understanding the intricacies of the trees.

I reviewed the literature to identify strands of thought that form a more or less coherent perspective, one that is generally based on the same set of assumptions around how change happens and so gives rise to a distinct set of policies or approaches. I focus on perspectives that seek to explain pathways out of widespread corruption in domestic contexts, rather than on international-level agreements or strengthening of internal safeguards for development aid.

The first part of this paper shows that anti-corruption policy over the past two decades has been largely influenced by a ‘state modernisation’ approach. While this has led to some advances, it is now regarded as insufficient and as burdening anti-corruption programmes with increasing opportunity costs. Most fundamentally, the dominant approach implies theories of change that, according to critics, do not hold up when confronted with reality. Moreover, insights from collective action theory and other perspectives suggest that the state modernisation approach is plagued by implementation challenges. This does not necessarily imply that all existing anti-corruption practice must be abandoned. But it does point to the need for a better understanding of how anti-corruption change happens, as well as new road maps that may require shifts in how anti-corruption is designed and practised.

The second part of the paper outlines five emerging policy perspectives that I have labelled ‘indirect anti-corruption,’ ‘localisation,’ ‘nurturing norms,’ ‘big bang,’ and ‘transnational (state to state).’ They represent new directions because they rest on alternative theories of change that diverge in some way from the dominant state modernisation paradigm. They do not necessarily aim to supersede this paradigm; some are complementary. They are also not mutually exclusive. Yet taken together, they offer some clarity on possible alternative directions that anti-corruption practitioners and donors may want to pursue.

Table 1. Taxonomy of policy perspectives in anti-corruption

Dominant paradigm
Policy perspective How change happens Types of interventions
State modernisation
Strengthening rules, transparency and enforcement to dis-incentivise corrupt acts
Legal reform, enforcement capacity, transparency and monitoring, social accountability
Alternative perspectives
Policy perspective How change happens Types of interventions
Indirect Deeper reforms and structural change Broad accountability reforms, welfare systems, education reform, gender parity, economic growth
Localisation Local agency and solutions, feasible spaces

Deep contextual analysis, targeted reforms, support for reform coalitions, convening local agency

Nurturing norms Informal institutions and norms: society-wide constraints against corruption Strengthening civil society spaces, changing social norms, collective action infrastructure
Big bang Shifting equilibrium: executive power and accelerated implementation Comprehensive reform packages
(state to state)
Policies beyond and between nation-states Reforms in aid-giving states, disincentives for elites, political leverage

The dominant paradigm: State modernisation

The main paradigm influencing policy choices in anti-corruption over the past three decades can be described as state modernisation.6cbceb40e1d5 It is rooted in rational choice theory, a perspective that conceives of public officials, politicians, and citizens as acting rationally with the aim of maximising their own preferences. This theory about human motivation is at the basis of the ‘principal-agent’ model that has underpinned policy design within the state modernisation paradigm.b237d419f42d A principal-agent model suggests that the persistence of corruption is rooted in insufficient monitoring and sanctioning of public agents by their principals, whether these are politicians or society at large. Anti-corruption, from this perspective, is about shaping institutions and building capacity to alter the personal calculations made by potentially corrupt actors.

In developing countries, these institutions and capacity are often understood to be lacking or deficient; the need to create or improve them gives rise to the state modernisation paradigm.9c04834cda9f This in turn calls for the state and associated social actors to reduce individual discretion, increase punishments for misconduct, and enhance monitoring. Modernisation also implies hastening the development of specific laws and procedures, enforcement organisations, and monitoring capacity to fashion an anti-corruption system. Often an implicit assumption is that when governance systems in more-corrupt countries come to look like those in less-corrupt countries, it becomes harder for corruption to prosper.

Anti-corruption tools that can be applied towards this end include public financial management reforms, audit institutions and anti-corruption agencies, national anti-corruption strategies, whistle-blower protections, procurement controls, community monitoring, and transparency measures, all of which can be developed or strengthened.adb1d3178c8b Capacity building involves a familiar set of related approaches aimed at improving corruption risk management, law enforcement capabilities, and national strategy making. A key aim has been to strengthen the capacity of important state institutions such as the police, judiciary, oversight institutions, parliament, and local government, but attention has also been given to bottom-up approaches that involve support to civil society groups and external watchdogs to enable them to monitor corruption.43669a13a3a5

It is clear that anti-corruption reforms rooted in state modernisation have contributed to advances, especially for some contemporary achievers like Singapore and Hong Kong.0f8d29a1d90e Georgia’s transition also relied, in part, on know-how and institution building. New managerial and coordinating bodies, oversight mechanisms such as the improved monitoring of local governments, and stricter application of sanctions for misconduct were successful in sweeping away organised corruption.1c7de6562ca0 Progress also relied on know-how from the outside. For example, the Georgian police used US and Italian legal expertise to develop new measures to target the organised criminal groups that had been driving corruption.ebc04e8f622c

While some scholars point out that the policy repertoire over the past three decades has remained broadly standardised, revolving around influential ‘anti-corruption packages’,7566c19ea744 there has undeniably been some evolution within this paradigm. This includes a focus on sectoral mainstreaming, increased specialisation and refinement of legal tools, enhanced methods of external or social accountability, and the incorporation of technology and e-governance tools. An increasingly sophisticated research base has been constructed to assess the effects of some of these reforms. As an illustration, Avis, Ferraz, and Finand9a635f39252 show that audits of Brazilian municipalities enhance accountability, increasing by 20 percent the likelihood of judicial action against a municipality to rectify malfeasance.

The most significant contribution of the dominant paradigm has been to trigger reforms and to construct an ever-improving repertoire of tools and practices that government and societies can use to deal with corruption. These are most likely indispensable; indeed, it is hard to imagine an anti-corruption policy approach that does not take into account the need for some degree of state modernisation aimed at developing institutions for enforcement, transparency, and accountability of state actors.

Inefficiencies of the dominant approach

However, the refinement of tools seemingly has not led to any great impact. In fact, empirical research has provided little evidence of sustainable positive change resulting from specific instruments forged under this paradigm.1953203df6ed It is not that there have been no improvements; rather, these approaches, whether in a particular sector or a national context, have been found to be generally inefficient insofar as the investments in anti-corruption have not resulted in the expected payoffs. This pattern of inefficiency is also captured by a raft of macro-empirical evidence showing that in the short to medium term, few countries receiving this kind of anti-corruption treatment achieve sustainable transitions away from corruption.b6b6405643d1

We still do not know the exact reasons why these varied tools have exhibited limited effectiveness in so many cases. Understanding, at the micro level, under what conditions and in which combinations anti-corruption interventions are productive – or not – remains an important area of research. But there is mounting evidence that one factor is failure to follow through with steadfast implementation. As a result, the policies and measures adopted do not lead to a sufficient internalisation of anti-corruption norms and behaviour that could underpin sustainable change. New anti-corruption laws do not exert a grip on behaviour; anti-corruption agencies lack autonomy and struggle to fulfil their mandates; trainings for bureaucracies do not lead to change in organisational norms; and ‘strengthened’ oversight institutions are instead weakened.

It is not surprising that implementation deficits tend to be a recurring issue in evaluations of this anti-corruption approach. The problem is that this challenge is often framed as another principal-agent or technical problem, rather than, as we shall see, one of collective action, power asymmetry, social norms, or the political economy of reforms. Reviewing the European Union’s external anti-corruption approach, Mungiu-Pippidi43b735b272c9 identifies a pattern resulting from the dominant paradigm: integrity-related laws and regulations are introduced, and then, once the inevitable ‘implementation deficit’ arises, a further push is made to enforce those laws via autonomous agencies. If they do not deliver, then new implementation agencies are created. While this may be a generalisation to some extent, the description highlights the mismatch between the presumed understanding of how change happens and the reality of what occurs. This has led to a number of critiques that focus on flaws in the theories of change underlying the state modernisation paradigm.

Critiques of the theories of change

One of the most influential explanations for why the dominant approach has faced difficulties in its application is grounded in collective action theory. This theory emphasises that individual rationalisations and calculations are shaped by expectations about how others may act: such expectations may apply to, for example, colleagues within a public administration or competing politicians or members of a community.a951a07c17a7 In contexts where corruption is not widespread and the rule of law is strong, the collective action problem is less relevant: actors mostly do not expect that others are corrupt, so there is less pressure on each person to act corruptly. The interests of the individual and the collective coincide, so those in authority, whether police commissioners or ministers of government or directors of hospitals, are more likely to make an effort to detect and punish corruption.b43cb2aa2ce8 In such settings, state modernisation perspectives have a clear path to effectiveness.

By contrast, in contexts of widespread corruption, people expect that most others will take opportunities to act corruptly or at least will refrain from active anti-corruption behaviour (such as whistle-blowing). Individuals in such societies are likely to rationalise that it is in their own interest to act corruptly as well, irrespective of whether that action damages the good of the group. This applies to both principals and agents, so we can expect to find few actors (‘principled principals’) willing to proactively pursue anti-corruption measures.c7f16ea45ac3 A minister may reason that there’s no gain from implementing an anti-corruption agenda if he cannot trust that rival politicians will refrain from using corruption to gain a competitive advantage; a doctor may find it difficult to refrain from taking a bribe when she knows or suspects that other doctors are already doing so. Hence, in these contexts of a high-corruption equilibrium, the very solutions proposed by principal-agent theory ‘are subject to the same incentive problems they are meant to solve’.5687cb22e470

Collective action theory has not been the only source of critique. Another line of argument is that state modernisation fails to take into account how power and interests shape the implementation and impacts of reforms.064b551d43ce The neglect of social norms and informal systems has also been cited as key in explaining why mainstream approaches have had a limited impact.a7fe44b2d9e3 And finally, critics have pointed out that state modernisation approaches do not take into account the transnational nature of corruption.e003d95d6e5d

The central insight of these critiques is not that state modernisation interventions are faulty per se, but that they are unlikely to be carried out with sufficient commitment or efficiency in contexts of high corruption. By putting questions around limited agency front and centre, these emerging perspectives have prompted new policy thinking that investigates how anti-corruption interventions can overcome fundamental implementation challenges and lead to more sustainable change. The five policy perspectives outlined in the next section in one way or another take up this issue of agency. At the heart of this effort is an attempt to provide stronger theories of change.

Mapping alternative policy perspectives

Indirect anti-corruption

This policy perspective is summed up by the admonition: ‘You don’t fight corruption by fighting corruption’.632a7c2f5fac The argument is that the state modernisation approach focuses too narrowly on reverse-engineering anti-corruption measures from those found in low-corruption countries, an approach that assumes that direct anti-corruption reforms are the main cause of transitions to integrity. Proponents of an indirect approach argue that this is a fundamental flaw, as we know from empirical research that the causes of change are not specific reforms, such as national anti-corruption strategies or new regulations, that aim to change individual incentives for potentially corrupt actors. Rather, it is deeper changes to governance or society that often allow for broad and collective progress.929fa21eb5ab Those countries that have sustainably transitioned to a less-corrupt equilibrium have done so mostly without recourse to specific anti-corruption policies and institutions.93f0d06f234f

Studies of past cases of transition have illustrated this logic of indirect change. Estonia, the ‘cleanest’ of the post-enlargement European Union states, never officially prioritised anti-corruption.5c3ff25075b3 In South Korea, the turn away from a highly clientelistic state was due in large part to property and education reforms. Extensive land reform in the 1950s led to a burgeoning middle class whose members insisted on better education for their children. Those children then became students who demanded that the civil service open up to all who were qualified; this in turn led to integrity-promoting reforms, such as entrance exams, that gradually entrenched meritocracy.d12ab91cca73

Understanding which policy reforms matter for the indirect ‘squeezing out’ of corruption is at the heart of the research agenda for this perspective. One line of thinking emphasises deep-rooted governance efforts, based on the view that transitions away from corruption have not sprung from confined debates around specific corrupt practices but have emerged through more fundamental contestations about who is to govern whom, and with what justification, and whose voices will be heard.7f8eb9618225 Hence, some have argued for anti-corruption efforts to focus less on legal reform and transparency and more on shifting fundamental relations of accountability by promoting ‘deep democratisation’8708400af935 or stronger administration through rationalisation and e-governance.

Still, much of the thinking about indirect approaches has focused more broadly on the kind of feasible social and political reforms that may bring about control of corruption as a by-product. Researchers have used regression analysis and case studies of past transitions to discover what these policy variables may be. Universal social welfare policies, for example, have in recent years helped generate a change in corruption levels, whether by weakening old patronage practices or by reducing inequality and thus enhancing trust.ac634a8c911f Rothstein7c4195ef381c suggests that a free and universal education system and a trusted tax system in the public sector are likely to diminish corruption. Uberti83a0c116aac2 points to economic reforms and subsequent growth as driving down corruption levels in Albania, with vast and expensive donor-led efforts exerting no more than a marginal effect. Increasing gender parity in the public sector has also been shown to reduce corruption.25a6af8f62af

This perspective offers an empirically grounded theory of change by pinpointing causal links between specific reforms and actual transitions. It broadens the scope for anti-corruption beyond typical measures, potentially liberating these efforts from obstruction by political elites: it may often be more politically feasible to help develop welfare institutions than to oppose corruption directly, for example. Reforms can be redirected to areas where donors have more agency and leverage, such as economic development, health, and education.

Still, this approach faces challenges as a realistic policy vehicle for anti-corruption. As Taylor sums up: ‘The time horizon for these structural improvements is usually at least several decades long – hardly the stuff for today’s results-oriented reformers’.fefd40ceb6ca Moreover, the way structural changes play out is highly contingent on local circumstances. The policy changes in South Korea, for example, were undertaken simultaneously in the Philippines, but to little avail.ab69521dba76

The indirect approach can also be challenged on grounds that it minimises the importance of agency, specifically that of political actors working within the system to push through reforms. Throughout Chile’s transition, structural changes remained marginal: the reduction in corruption cannot be explained by the growth of the middle class, the rising gross domestic product (GDP), or the copper. Key instead was the ‘extraordinary agency’ of ‘patrician politicians, radical challengers, libertarian economists and bureaucratic auditors’ who conjured a ‘hard-built equilibrium’ even prior to the country’s democratisation.1add773946cf Even in South Korea, ‘human agency was essential’.40a21dad5a3a Mungiu-Pippidi and Johnston argue that ‘societal modernisation’ may play a role but cannot determine shifts, demonstrating that increases in the Human Development Index (a proxy for modernisation) do not correlate with control of corruption.2f6d5334a95c

Nevertheless, this perspective both clarifies and creates dilemmas for anti-corruption practitioners. Could anti-corruption funding be spent more efficiently on programmes that seek to accelerate structural transformation, such as the promotion of employment, educational reforms, or social insurance policies? At the very least it points to an important policy research agenda, as an understanding of how structural change is connected to anti-corruption could make anti-corruption choices smarter and more sustainable. For example, if we know that gender parity in the public sector is an important predictor of the control of corruption, might it make sense to channel more anti-corruption efforts into countries that have greater parity?


This policy perspective pushes back against both the universalistic conceptions of anti-corruption implied by the state modernisation paradigm. Localisation, however, signifies much more than ‘tailoring to context’ or ‘local ownership,’ vague terms that often obscure the need for deeper shifts in the relationship between international donors and local actors.f2db3471fb95 It means that the choice of anti-corruption targets should be determined by local conditions and not by preconceived templates of what anti-corruption should look like or by donor preferences. In particular, localisation means targeting anti-corruption efforts to specific, discrete, and manageable spaces where sustainable transitions to impartial governance are realistically achievable and will have a high local impact.22fb653afde4 This is in contrast to state modernisation approaches that often seek ‘systemic anti-corruption,’ across-the-board approaches embodied in national strategies or integrity systems that are implemented without much consideration for local fit.f3e09a0353c4

Localising anti-corruption could mean taking advantage of potential ‘pockets of effectiveness’ within specific ministries, public sector organisations, courts, local governments, or delivery units (such as a cluster of hospitals), so that transitions to integrity can develop even in highly corrupt settings.0b6b14024f32 Clearly, alignment with local political agency is key to making this approach work. Localisation thus requires careful scrutiny of the underlying conditions of the context in question.59ec75268f27

Indeed, localisation also draws on broader theories that suggest ‘doing development differently’ by emphasising local agency and endogenous change.c00a0c0b6fc0 For anti-corruption, the implication is that methods of change should be locally designed rather than imported, a view that challenges a general trend in anti-corruption that elevates ‘best practice’ templates and external expertise.10689139f5fd With localisation, the emphasis is on local leadership and action rather than on donor agency and external knowledge. This is due in part to a recognition that ‘locals (broadly defined) are more likely than outsiders to have the motivation, credibility, knowledge and networks to mobilise support, leverage relationships and seize opportunities in ways that qualify as “politically astute”’.16a257438ca8 It also reflects a recognition that outsiders may have some leverage but ultimately face limitations in their ability to bring about genuine change.a5ab9b1e10fe

Under the localisation approach, donors become ‘enablers’ rather than active agents of change. The starting point should be ‘a genuine effort to seek out existing capacities, perceptions of problems and ideas about solutions’ that may exist locally.6b834133d547 This may mean supporting specific reform coalitions. Khan, Andreoni, and Roy1f28db90d46b argue that feasible anti-corruption strategies should include mobilising support for anti-corruption reforms so that these become ‘self-enforcing.’ It can also entail entering into deliberate partnerships with local centres of informal power, such as forms of customary authority.df7f70dc4284 Localising can also mean the ‘de-colonising’ of typical anti-corruption project management tools, reframing logframes and indicators in ways that put ‘local values, knowledge and experience at the heart of [the] work’.5af1cea4369c

Localising solutions also requires unpacking corruption into very specific problem-types and then designing agendas around those dynamics. For example, Jackson et al. show that tackling corruption in Jordan would mean going beyond typical enforcement frameworks to support efforts that tap into local norms and forms of agency.68a89c20ac6a Working with the grain of social norms is also important. In Rwanda, decentralised public service agreements rooted in traditional cultural practices and norms have achieved notable successes.b0c361c73972

The localisation perspective also suggests that the anti-corruption field should open itself to diverse approaches, as appropriate interventions to support feasible anti-corruption strategies will vary across contexts. The main drawback of a localisation strategy is that it will face constraints tied to the politics of development aid.

Nurturing norms

This policy perspective sees the effectiveness of anti-corruption efforts as being determined by the balance of social forces either for or against corruption.d262f21554e7 It is based on research that shows agency in successful transitions as emanating not from the implementation or development of formal institutions, but from informal institutions. These cases demonstrate the generation and maintenance of anti-corruption as an effective social norm – a society-wide ‘rule of the game’ that can move reforms forward and strengthen resilience against pushback.b2cd6cf32477 In Estonia, for example, the coherent and sustained social pressure of public opinion, released by the ending of communist constraints, was the most important cause of that the country’s transition.dd3359ea5be8

Proponents of this perspective argue that state modernisation approaches based on introducing formal institutions are at risk of being subverted by countervailing social forces embodied in underlying informal institutions and norms.e65bc6520583 These in turn are shaped by self-serving forms of collective action, such as patronage networks. As a result, prevailing social norms may condone and even encourage corruption.111da9ece59a Citizens may find themselves in the role of ‘clients’ who cannot hold political actors to account;b1636f09e818 private firms may relate to the state through established rules of cronyism.1ed1cec6f98b Such negative dynamics can easily undermine the potential for constructive collective action and agency oriented towards integrity.338625e7df19 In Georgia, for example, the newly won autonomy of public institutions was eroded from within as reformist elites began to seek their own rents as part of serving political backers. This suggests that norms against state capture and cronyism never sufficiently developed in Georgia.3e3e4fb2d16c

Transitions therefore require a far-reaching focus on pro-actively nurturing normative constraints on corruption. Anti-corruption must become ‘a widespread norm endorsed by … a majority of active public opinion’.22d1d728d14f Developing such normative constraints increases the chances that society – media, groups of citizens, nongovernmental organisations (NGOs), voters, public officials, politicians – will strike back quickly and effectively against attempts to undermine anti-corruption reform. Prioritising interventions that nurture normative constraints is in line with collective action theory, which emphasises that social challenges can be solved by developing expectations and trust that other social actors – colleagues or fellow citizens – won’t free-ride in the production of a public good, in this case public integrity.

The policy means to achieve normative constraints against corruption will require interventions that go beyond or entail modifications to conventional approaches based on civil society. Such interventions, such as funding NGOs, social accountability measures, or awareness campaigns, have rarely shown decisive effectiveness in shifting social norms or building up informal constraints against corruption. Social accountability efforts, for example, generally focus on generating information or holding state actors to account rather than on nurturing society-wide constraints.eb014ae3bbb7

The research agenda is concerned with understanding what kinds of policies may be relevant to norm building. So far, no easy answers have emerged, although thinking has settled on a number of strands. The first relates to going beyond funding NGOs to fostering a richer civil society in the de Tocquevillian sense, one where people come together in spaces between the state and the market to collectively solve problems.1e719f6d1fc1 The evidence indicates that the more a society has this collective action capacity – as measured, for example, by density of membership and number of civil society associations – the stronger will be its control of corruption.465a05864e94 The means to generate this capacity is unclear, but the nature of support to civil society may need to change. A review of two decades of programming in Eastern Europe suggests that ‘few donor programs [are] designed to foster collective action, leadership, and media activity’.8b06f60c5620 Helping to organise new coalitions that combine collective interests with the capacity to pressure rule enforcement agencies has been suggested as a key anti-corruption strategy.e7452c4d1931

Nurturing normative constraints against corruption also requires more than changing personal attitudes through awareness-raising or education campaigns: it must include a coordination mechanism whereby individuals can be sure that others are also moving in the same direction. Though the direction of a social norms policy agenda is somewhat unclear, some tentative suggestions include working with social networks, constructing social spaces for norms changes, and reaching people through ‘edutainment,’ or educational entertainment.dfc5398b6235

The contribution of this perspective is that it locates agency away from formal authority and institutions and concentrates on the informal influences on behaviour that shape the balance of forces in society. Focusing on these opens the possibility of a broader menu of reforms. Yet at the same time it throws down a more intimidating challenge. It is far from clear what kinds of policies are needed for the genuine development of collective action capacity, or whether reformers even have the patience to engage in the slow, painstaking development of social norms and constraints. The approach also raises legitimate questions around intrusiveness, especially if the reforms are to be donor-led.

Big bang

This perspective relates agency to the pace and extent of reforms. Proponents suggest that the incremental building-block approach typically pursued in anti-corruption is inadequate, and therefore a more sweeping approach is needed. This is because societies can become stuck in what economists refer to as a high-corruption equilibrium. According to this view, individuals’ choices around corruption are contingent on their ‘mental models’ of other people’s behaviour.cd50d5cd2485 For people in societies where corruption is widespread, this mental model can be summed up as follows: since other people are corrupt, I might as well engage in corruption myself; and if I myself am untrustworthy, then everyone else must be untrustworthy too.2eaf2bec3114

In such settings, the pay-off for being corrupt is higher, as dishonest behaviour incurs fewer social or physical costs than in settings where corruption is low. Corruption becomes self-reinforcing, meaning that shifts away from this steady state can only work if the equilibrium is ‘popped’ – that is, if all actors change their behaviour at the same time.42b43b0de6cb The policy implication is that gradual reforms are highly susceptible to backsliding into the old equilibrium. When that happens, despite huge efforts to curb it, corruption successfully ‘fights back,’ an outcome we see even in circumstances where anti-corruption efforts were initially deemed successful.e5f9634a7f2d

Reforms must be not only rapid but also sufficiently extensive, resulting in behavioural change across the public sector. Rothstein argues that impressions of public institutions are key conduits for change because public officials send strong signals about what kind of game is being played in society and about the rules of the game. People may change their mental models in response to the signalling of political and public actors who send the message that institutions now work in a different way.c6a7b65532b0 Single or step-by-step reforms to public institutions do not send a strong enough signal that society is working under a different logic than before and so are highly susceptible to pushback.

Sweden, for example, became free from corruption at the end of the nineteenth century precisely because an onslaught of reforms between 1855 and 1875 prompted a fundamental change in Swedes’ mental model of their public institutions.520dfaf6696c Georgia constitutes a contemporary example of reform according to the big bang principle. Changing how people perceived the state was an explicit aim. The mayor of Tbilisi said that the country had to ‘attack the symbols of corruption,’ which included making high-profile arrests, some widely publicised, of corrupt actors.b5c4d8beeab3

Rothsteincafab64e8f6d refers to this quick and comprehensive route to reform as the ‘big bang’ approach. Stephensonc28eaaebd2e4 writes that ‘many of the leading figures in anticorruption and development studies have advanced some version of [this] argument,’ encouraging policy makers to favour rapid change over incremental measures. This approach implies a return to top-down policies driven by the executive organs of the state. It also requires taking advantage of windows of opportunity – economic crisis, revolution, and so on – when an anti-corruption agenda can be implemented with least resistance.

Stephenson has recently pulled back, arguing that neither theory nor empirical research supports claims of the superior effectiveness of the big bang approach. He maintains that even if corruption is self-reinforcing, this does not mean that a high-corruption equilibrium is necessarily less responsive to incremental reforms than to rapid, comprehensive reforms.d18c36e6a553 In fact, it is a theoretical possibility that high-corruption equilibria may be amenable to sustained, cumulative interventions, with small shifts creating a spiral effect that can lead over time to a new equilibrium. He argues that an emphasis on the magnitude of reforms may come at the expense of their sustainability.ff1ab4136f0a

Ultimately, Stephenson argues that choices about the pace and extent of reforms should be based on empirical rather than theoretical reasoning. For example, the big bang approach may be better than other approaches in contexts where there are clear signs that corrupt actors would be able to circumvent or subvert more gradual reforms. Or events within the country may open up windows of opportunity to push for a quick and deep transformation.

There are potential downsides to aggressive reforms, not least the risks of sparking instability, inflexibility, and creeping authoritarianism.16eb0d814c61 Nevertheless, this debate pushes policy questions around pacing, comprehensiveness, and complementarities to the fore. This is surely an important research agenda, given that current strategies tend to be like ‘laundry-lists … provided without much guidance for implementation, sequencing, or concern for the systemic whole, which at best will only correct topical maladies’.95410a6f8a48 These questions also focus attention on the sustainability of donor approaches that tend to be atomised and partitioned into segments, with little coordination.

Transnational (state to state)

This perspective emphasises that globalisation has upended the system of nation-states operating as discrete entities bound by neat conceptions of national sovereignty. As Heywood207d4931a075 notes, the contemporary reality is one of ‘post-modern states,’ where national sovereignty is eroded by the blurring of boundaries between private and public and between domestic and transnational. Because national economies are integrated into transnational dynamics through trade, capital flows, and foreign direct investment, and because corruption thrives within this nexus, state-level corruption cannot be resolved only though anti-corruption efforts located in states.a9960682d2de The policy response, too, must be transnational, which means seeing the task of reducing corruption in one state as linked to policies in other states. While efforts have been made to improve the international architecture of anti-corruption through the United Nations Convention against Corruption and international bribery standards, for example, this state-to-state perspective has received less attention.

The transnational approach emphasises two aspects. The first has to do with breaking down the boundaries between ‘us’ and ‘them’ inherent in donor-beneficiary dynamics, prompting donors to ask: How are we contributing to corruption in other societies and what internal actions can address this? There are potentially a number of weak points within donor countries’ domestic legal, accountancy, and banking systems that can facilitate money laundering, fraud, and other forms corruption in developing countries. The transnational perspective emphasises a research and policy agenda to identify and craft solutions to these sources of corruption.

Over the past couple of decades, for example, the Department for International Development (DFID) has sought to actively repair weak points in the United Kingdom’s own regulatory and financial systems. This has resulted in DFID, a development agency, inducing changes in the policies of other domestic departments, ‘often against their inclinations’.9d4ab4c4b2d6 Stronger action on illicit financial flows and the enactment of the UK Bribery Act 2010 were fruits of this approach. The UK government also set up the International Corruption Unit, which remains the world’s only aid-funded law enforcement unit in a developed country that is dedicated to tackling developing-country corruption. ‘It remains something of a mystery why [other donors] have not followed suit,’ comments Mason.8d5f8c49eb21

A second aspect of transnational approaches involves a more self-conscious use of power and politics, representing a departure from technical approaches to anti-corruption. As explained by Mason,4c93f88accf7 aid-giving states can use ‘micro-levers’ to create disincentives for national elites in aid-receiving states to engage in corruption.67a2a2a4788b These measures target specific individuals or groups, frequently those associated with a country’s interior and financial ministries. Examples include decisions on granting or withholding visas to visit the donor country; exclusion orders; financial vetting checks on government representatives (‘know your partner’ checks); and restricting engagement when corrupt practice is suspected. Less direct measures may seek to increase the reputational damage of corruption through actions geared towards the business community, credit rating agencies, and regional peer groups, with a view to deterring inward investment. The point is that donors begin consciously to assess what matters to the elite in a country and find ways to influence them. This approach requires strong collaboration between a donor agency and other parts of the same government, as control over these ‘non-aid’ levers resides largely outside the remit of development agencies. While aid is, of course, inherently political, these levers are more explicitly so, involving the political branches of government and explicitly political choices.

The use of political leverage against corruption, however, needs to be improved. Research points to the under-use of political levers in relation to corruption and to the weakness of donor coordination mechanisms in-country.8d2f175d6e2b Political conditionality to induce anti-corruption reform has shown disappointing results.b44a19ada00c There is certainly scope for deliberate policies and initiatives to compel donors to improve their coordination. One review suggests that donors ‘simply cannot afford to respond to corruption cases in the haphazard and poorly planned fashion that is common practice currently.’ These authors argue that more energy needs to be invested in responses that have a strategic focus beyond ‘getting the money back’ and that stimulate domestic accountability.d47d71278cd8 A step change is needed: coordination should be seen as an adjunct to programming, a way to manage efficiency and facilitate planning, rather than as a political act to deter corruption.

Overviews of anti-corruption options typically do not consider state-to-state policies as part of the anti-corruption repertoire.224cb6393abd Moreover, they are not really included within donors’ own understanding of what anti-corruption comprises, and so are rarely afforded priority (though there are exceptions, such as the UK government’s 2017 anti-corruption strategy). While the transnational perspective on its own cannot prompt transitions at the national level, proponents argue that the cumulative impact of transnational policy instruments could be significant and, at the very least, can work in combination with technical approaches. The policy-research agenda can identify promising mechanisms to improve ‘introspective’ policies, but it’s an agenda that faces constraints. Development practitioners will need to develop new skills to advocate across government, and structural constraints around the delivery of aid may prevent progress. Moreover, for critical scholars, these policies may be deemed intrusive, less about addressing corruption and more about powerful states finding ways to extend their influence and authority in developing countries.ce6e904147cf


Based on a review of the corruption literature, this U4 Issue identifies emerging policy perspectives in anti-corruption that diverge from the dominant (state modernisation) paradigm. Five are highlighted, though others may exist. In general, these perspectives base their arguments on questions of agency, that is, who and what is most likely to bring about sustainable changes to the domestic control of corruption. In doing so, they adopt a lens that opens up the scope of anti-corruption, viewing it not as an exogenous technical intervention aimed at tackling discrete bundles of corrupt practice, but as a process that relies on complex endogenous change. This change involves triggering of a set of political and social processes and dynamics that will eventually lead to the internalisation of integrity, whether in a particular organisation, in a sector, or across society. Each perspective relies on a distinct theory of change.

These perspectives point to ways in which the specific interventions of the state modernisation approach could be improved. For example, the indirect perspective suggests that anti-corruption may thrive when certain structural conditions are in place, while the big bang perspective holds that one way to enhance current approaches is by taking advantage of windows of opportunity to drastically increase the rate of reform. The localisation perspective outlines how current anti-corruption approaches can be better contextualised. The nurturing norms perspective calls for new kinds of models to change social norms and informal institutions that undergird corruption. The transnational or state-to-state perspective suggests that the repertoire of anti-corruption measures should include non-aid, explicitly political levers. The various perspectives can be complementary: for example, a localisation paradigm may facilitate the nurturing of norms, while the big bang and transnational approaches may go together.

At the same time, there are tensions between these perspectives. The transnational paradigm may imply increased involvement of donors in domestic anti-corruption agendas, an approach that is anathema to proponents of the localisation perspective. It is time for a robust debate on the merits of the various approaches, with an understanding that diversity may be the hallmark of future approaches to corruption. For those who lament the stagnation in anti-corruption efforts, this debate can point the way towards possible new directions in policy.

  1. Fukuyama 2014.
  2. Green 2017.
  3. Mungiu-Pippidi 2015.
  4. Klitgaard 1988.
  5. Mungiu-Pippidi and Johnston 2017.
  6. Mungiu-Pippidi and Johnston 2017.
  7. Mungiu-Pippidi and Johnston 2017.
  8. Johnsøn, Taxell, and Zaum 2012.
  9. Kupatadze 2017.
  10. Taylor 2018.
  11. Mungiu-Pippidi 2015.
  12. 2018.
  13. Mungiu-Pippidi 2015; Taylor 2018; Mason 2018.
  14. Mungiu-Pippidi 2015, 2020.
  15. Johnsøn, Taxell, and Zaum 2012; Mungiu-Pippidi 2015; Poate and Vaillant 2011; Disch, Vigeland, and Sundet 2009.
  16. 2020.
  17. Persson, Rothstein, and Teorell 2013; Klitgaard 1988.
  18. Schelling 1960; Olson 1971.
  19. Persson, Rothstein, and Teorell 2019.
  20. Persson, Rothstein, and Teorell 2013, 456.
  21. Heywood 2017.
  22. Baez Camargo and Gatwa 2018; Scharbatke-Church, Barnard-Webster, and Woodrow 2017; Jackson and Köbis 2018.
  23. Khan, Andreoni, and Roy 2019; Booth and Unsworth 2014.
  24. Rothstein 2018; Mungiu-Pippidi and Johnston 2017.
  25. Rothstein 2018.
  26. Kaufmann 2017.
  27. You 2017.
  28. Kalniņš 2017.
  29. Johnston 2018.
  30. Mungiu-Pippidi and Johnston 2017, 253.
  31. Bauhr, Charron, and Wängnerud 2018.
  32. 2020.
  33. 2018.
  34. Taylor 2018.
  35. Mungiu-Pippidi 2020.
  36. Taylor 2018, 64.
  37. Mungiu-Pippidi and Johnston 2017, 240.
  38. You 2017, 154.
  39. Navia, Mungiu-Pippidi, and Martini 2017, 232.
  40. Michael 2004; Pope 2000.
  41. Khan, Andreoni, and Roy 2019.
  42. Booth and Unsworth 2014.
  43. Roll 2014; Khan, Andreoni, and Roy 2019; Walton 2013.
  44. Roll 2014; Khan, Andreoni, and Roy 2019.
  45. Mungiu-Pippidi 2020; Chandler and Sisk 2013.
  46. Booth and Unsworth 2014, 4.
  47. Michael 2004; Mungiu-Pippidi 2020.
  48. Green 2017; Booth and Unsworth 2014; Denney and Mallett 2017.
  49. Baguios 2019.
  50. Murtazashvili and Jackson 2018.
  51. 2019, 6.
  52. Booth and Unsworth 2014, 27.
  53. ADB 2012.
  54. Jackson, Tobin, and Eggert 2019.
  55. Kalniņš 2017.
  56. Fukuyama 2014; Mungiu-Pippidi 2015.
  57. Mungiu-Pippidi 2006; Mungiu-Pippidi 2015.
  58. Kupatadze 2017.
  59. Mungiu-Pippidi and Hartmann 2019.
  60. Keefer 2007; Barnett, Yandle, and Naufal 2013.
  61. Eisenstadt and Roniger 1984; Fox 1994; Auyero 2000.
  62. Banfield 1958; Olivier de Sardan and Blundo 2006; Jackson and Köbis 2018.
  63. This perspective also pushes back against the notion that transitions can be imposed on society through top-down executive authority alone. In general, executive action must coincide with the development of underlying social constraints. Enlightened despotism is unlikely. Only four ‘not free’ countries are found among the top 50 best-governed countries (Mungiu-Pippidi and Johnston 2017).
  64. Mungiu-Pippidi 2015, 163.
  65. Johnsøn, Taxell, and Zaum 2012.
  66. Khan, Andreoni, and Roy 2019.
  67. Mungiu-Pippidi 2015, 172.
  68. Mungiu-Pippidi 2015.
  69. Booth 2012.
  70. Jackson and Köbis 2018; Jackson, Tobin, and Eggert 2019; Hoffmann and Patel 2017.
  71. Rothstein 2005, 121.
  72. Rothstein 2005, 161.
  73. Baniamin 2015; Kupatadze 2017.
  74. Fisman and Golden 2017.
  75. Rothstein 2005, 166.
  76. Sundell 2016.
  77. Rothstein and Teorell 2015.
  78. 2019, 4.
  79. 2011.
  80. Stephenson 2019, 35–37.
  81. Stephenson 2019, 23.
  82. Taylor 2018, 64.
  83. Stephenson 2019, 37.
  84. Heywood 2017.
  85. 2017.
  86. 2020.
  87. Mason 2020.
  88. This paragraph draws on Pressure to change: A new donor approach to anti-corruption? (Mason 2018).
  89. 2018.
  90. de Vibe et al. 2013.
  91. Mungiu-Pippidi 2015, 191; 2020.
  92. Johnsøn 2016.
  93. Chandler 2010.
  94. Johnsøn, Taxell, and Zaum 2012.



The author would like to thank Inge Amundsen, Monica Kirya, Cecilie Wathne, and Aled Williams for reviewing early versions. He is also grateful for the very helpful comments from Claudia Baez Camargo, Nils Köbis, and Grant Walton.