Recently, the analysis of corruption in developing countries has undergone a radical revision. Specifically, scholars are increasingly emphasizing that we need to move away from studying corruption as individualistic acts of deviant behavior and instead acknowledge that corruption can become institutionalized as informal rules and routines, thus putting great pressure on individuals to perform according to these norms.Footnote 1 As succinctly put by Della Porta and Vannucci [ [14]: 230; emphasis in the original], ‘[t]he “basic norm” of this invisible legal system sanctions the unavoidability of bribes, the rule that recourse to hidden exchange cannot be avoided in return for any “resource” of value obtainable from the public structure within the corruption network’. Or in the words of Persson et al. [ [64]: 254; emphasis added], ‘in the majority of the world’s countries, corruption is the expected behavior rather than the exception’.

Moreover, not only has the study of corruption in the developing world seen a seismic shift in its underlying theoretical assumptions but scholars have also begun to question the ways in which we operationalize and measure corruption. In particular, a growing number of scholars stress that institutionalized corruption can take very different forms, which, in turn, means that our conceptual frameworks should reflect these differences. Thus, whereas previous comparative research primarily measured the extent of corruption – relying mostly on aggregated expert-based assessments (such as Transparency International’s Corruption Perceptions Index or the World Bank’s ‘control of corruption’ indicator) – there are now more and more frameworks that highlight qualitative differences in the organization of corruption [for example, [36, 41, 86]].

However, while this emerging neo-institutional approach to corruption has undoubtedly made a great contribution to furthering our understanding of corruption in the developing world, there remains a question that has received almost no attention: how can we explain differences in the institutionalization of corruption across developing countries?

To address this question, the paper here will apply Johnston’s typology of corruption to East Asia – a geographical region that features a wide variety of corruption types. The general argument that emerges from this analytical exercise is that historical sequencing matters. As will be shown, the different types of corruption that can be found in contemporary East Asia were institutionalized in the mid-twentieth century when universal suffrage was rolled out across the region. Specifically, where universal suffrage was introduced under the constraints of a ‘political marketplace’ characterized by a large number of significant ‘loyalty sellers’, corruption assumed ‘oligarchs and clans’ or ‘official moguls’ forms; where universal suffrage was introduced after the number of loyalty sellers had been significantly reduced, corruption came to be locked into the ‘elite cartel’ type.

To make this argument, the paper will begin by providing a comprehensive literature review that outlines the methodological shift in recent corruption research. Based on the emerging neo-institutional approach to corruption, the remainder of the paper will then, first, map different qualitative types of corruption across East Asia and, second, show how these different types can be traced back to the historical moment at which political elites were forced to make strategic decisions over how to mobilize the masses in electoral contests. To conclude, the paper will generalize these findings to the developing world as a whole.

Studying corruption: from institutional economics to neo-institutionalism

Ever since academic research began to take a stronger interest in causal explanations of corruption, the institutional economics approach has been the dominant inquiry paradigm. Based on a conventional rational choice model and set against the foil of the so-called principal-agent problem, the institutional economics approach makes the assumption that the agent is always interested in maximizing his or her own payoff from corrupt behavior, while the principal is interested in minimizing the welfare costs that come with corruption. For example, in a much cited study of political corruption framed through the institutional economics approach, Kunicová and Rose-Ackerman [45] take it as a given that voters (the principal) ‘prefer honest officials to elected ones who enrich themselves through payoffs’, while politicians (the agent) are driven by the goals of ‘individual wealth and reelection’. In other words, for the institutional economics approach, both the agent’s and the principal’s interests are exogenously determined.

Under these assumptions, the institutional economics approach predicts that corrupt acts occur when a rationally acting agent, after performing a risk-profit calculation, decides that the potential profit of engaging in corruption outweighs the risks [e.g., see [44, 72]]. Whereas the profit factor refers to the question of how much there is to steal – for example, proponents of the institutional economics approach have argued that the expected profits of corruption are larger under conditions of a heavily regulated market [e.g., [17, 21]] and a resource-rich economy [e.g. [83]] – the risk factor refers to the principal’s ability to monitor and punish the agent for corrupt behavior. In turn, the risk factor is a function of institutional design. For example, research following the institutional economic approach suggests that the monitoring capacity of the agent will be strengthened through a decentralization of government [13] and greater press freedom [9], while accountability is facilitated by presidential systems of government [60] and plurality electoral systems [45]. In short, the institutional economics approach puts forward the argument that corruption is largely the product of institutional incentive structures.

Recently, however, the institutional economics approach has come under heavy criticism from the neo-institutionalist approach to corruption. While still retaining a rational choice framework, the neo-institutionalist approach departs from the institutional economics approach by recognizing that only some preferences being pursued by individuals are exogenously given – such as the drive towards self-interest maximization – while other preferences may be endogenously determined by the behavior of other individuals. Specifically, the neo-institutionalist approach to corruption disagrees with the view that individuals’ honesty or dishonesty can be assumed to be fixed exogenously; instead, whether an individual decides to remain ‘clean’ or engage in corrupt behavior is a function of interactions between individuals. In particular, the two variants of the neo-institutionalist approach – the game-theoretic and the transaction costs version – argue that the risks of engaging in corruption behavior depends on whether interaction between individuals has led to an institutionalization of corruption, with corrupt transactions regulated by informal rules and practices.

According to the game-theoretic variant of the neo-institutionalist approach, the risk of engaging in corruption is lower in a high-corruption equilibrium – that is, in a context where corruption is systemic and widespread – because there will simply be no principals to ‘operate’ formal monitoring and punishment mechanisms, and hold agents accountable for their corrupt behavior [e.g. [64, 65, 73]]. This is due to a classic collective action problem: because corruption is the expected behavior, individuals cannot trust other individuals that they will act clean and play their part as principal. As a result – knowing that their individual contribution will make little, if any, contribution to holding the agent accountable for corruption – it is most rational for each individual to free ride on others’ anti-corruption efforts.

The transaction costs version of the neo-institutionalist approach, on the other hand, highlights two different mechanisms to explain how inter-individual interaction shapes the agent’s risk calculation. First, the institutional economics approach is criticized for modeling corrupt acts as one-off exchanges. Instead, the transaction costs approach stresses how, if corrupt acts are carried out repeatedly between the same partners, mutual trust will grow stronger, thus generating the expectation that the other side will abstain from fraud or betrayal [25]. Second, the transaction costs approach maintains that, in a high-corruption equilibrium, the risk of getting caught and punished for corrupt behavior is lower than in a low-corruption equilibrium because information and skills regarding the ‘logistics’ of corruption are more widespread, and because the large-scale reproduction of corrupt exchanges may have prompted the emergence of actors who specialize in the enforcement of exchange contracts [14].

Irrespective of the different causal mechanisms put forward for why systemic corruption may reproduce itself over time, the two sub-strands of the neo-institutional approach highlight the same phenomenon: corruption can become institutionalized as an informal system of norms and practices that shapes individuals’ strategic thinking and behavior. Under circumstances of institutionalized corruption, individuals see corruption as the standard process to ‘get things done’ rather than deviant behavior that violates prevailing norms. As a result, formal institutions will – in contradiction to what the institutional economics approach argues – lose their constraining effect on individual behavior.

Importantly, the neo-institutionalist assumption that individuals’ preferences are endogenously shaped by how other individuals behave has implications for the conceptualization of corruption. Whereas research based on the institutional economics approach commonly employs a quantitative measure of corruption – that is, the dependent variable is the extent of corruption in a political system – the neo-institutional argument calls attention to possible qualitative differences in the institutionalization of corruption. In fact, a growing body of research shows that such qualitative differences matter when it comes to analyzing the effects of corruption. For example, several studies have shown that whether corruption has a harmful effect on the quality of governance and economic development depends on its industrial organization, not on its sheer extent [for example, [41, 42, 68]]. Likewise, various authors have argued that anti-corruption measures need to be tailored to specific types of corruption; a ‘one size fits all’ approach will not work [e.g., [36, 78]].

However, despite these important theoretical and empirical contributions, the neo-institutionalist approach has paid very little attention to the question of how corruption becomes institutionalized in the first place. In other words, how do we explain differences in the institutionalization of corruption between countries? Specifically, why do corruption markets differ in their degree of organization, their risk profile, and their key actors?

Corruption in East Asia

To begin to tackle these questions, this section will apply one of the most widely cited typologies of corruption – Johnston’s [36–38] four-type framework – to the developing and newly industrialized countries of East Asia.Footnote 2 While this sample does not contain any cases of ‘influence market’ corruption, which are generally hard to find in the developing world, it includes Johnston’s three other types of corruption, thus allowing for a systematic inquiry into the factors that shape the institutionalization of corruption.

Elite cartels, ‘big men’, and moguls

Essentially, Johnston puts forward three questions to identify different types of corruption: Who are the principal actors around whom corruption is organized? What are these actors seeking to achieve through corruption? And how do they employ corrupt activities to attain their objectives? Framed through these questions, it becomes clear that East Asia’s developing and newly industrialized countries fall into three types: ‘elite cartel’ corruption (Singapore, Taiwan, South Korea, Malaysia), ‘oligarchs and clans’ corruption (Thailand, Philippines), and ‘official moguls’ corruption (Indonesia) (see Table 1).

Table 1 Corruption types in developing and newly industrialized East Asia
Full size table

Singapore, despite the fact that quantitative, expert-based measurements of corruption (such as Transparency International’s Corruption Perceptions Index or the World Bank’s ‘control of corruption’ indicator) tend to rank the country among the world’s ‘cleanest’, shows tendencies of ‘elite cartel’ corruption. The most obvious symptom of this is the politicization of the state bureaucracy by the dominant People’s Action Party (PAP) and anecdotal evidence that high-ranking party officials use state-owned enterprises (the so-called government-linked companies, GLC) as a source of rents [e.g. [28, 69]]. The ‘fusion of party and state’ [26] – and hence the collusion between political and bureaucratic elites – also becomes evident through the electoral process, as PAP-controlled constituencies are typically treated preferentially over opposition-held districts when it comes to the allocation of certain government programs (e.g. housing subsidies, infrastructural upgrades).

Even clearer cases of ‘elite cartel’ corruption can be found in Taiwan, South Korea and Malaysia. Here, collusive intra-elite networks extend beyond the institutions of the state to also include non-state actors. However, the operating logic is the same: corruption serves to strengthen the loyalty within intra-elite networks, with the aim to pre-empt or to co-opt rising competitors. In Taiwan, collusive networks tend to be highly localized and organized around individual politicians. Typically, patron-client networks connect politicians to so-called local factions (difang paixi), with the purpose of exchanging public resources and services (e.g. contracts, credit, protection for illegal business) for the delivery of votes and the provision of campaign funding [e.g. [22, 30]]. In South Korea, on the other hand, collusive intra-elite networks have historically been organized at the national level. A highly institutionalized system of kickbacks was first set up under military rule (1961–1987), whereby senior party officials collected fixed payments from business conglomerates (the so-called chaebol) in exchange for access to public resources (credit, import licenses etc.). Yet, Wad observes how ‘the structure of this ‘politics-economy’ collusion was not dismantled with the rise of democracy’ [ [84]: 210]. Rather, the introduction of free elections in the late 1980s meant that the number of collusive networks multiplied, with the chaebol channeling financial resources to those politicians with a good chance of winning the presidential office [ [61]: 115]. Similarly, in Malaysia, authoritarian rule under the United Malays National Organization (UMNO) has also bred institutionalized forms of bribery at the national level, as many businesses ‘retain, cultivate and ultimately rely on their connections with top UMNO leaders to secure continued patronage while, in turn, providing financial and other backing for their political patrons’ [ [39]: 296].

While ‘elite cartel’ corruption is thus organized around networks of colluding elites, the principal actors under the ‘oligarchs and clans’ type of corruption are ‘big man’ networks that specialize in the large-scale theft of public resources. The often violent scramble among these networks for access to public resources means that the boundaries between public and private become indistinct; meanwhile, the gains from corruption are constantly under threat from other networks.

In the Philippines, ‘big man’ networks usually take the form of political machines under the control of political clans. These machines are essentially ‘vehicles for raiding the state and distributing political and economic largesse’, based on institutionalized linkages with key actors in the local political economy, such as the bureaucracy, legal and illicit business, print and broadcast media, and civil society associations [ [79]: 207–208]. The dense networks of informal linkages that political clans maintain in their respective bailiwicks have essentially resulted in a ‘privatization of public resources’ at the local level [ [50]: 10]. Moreover, through temporary alliances with other clientelistic machines manifested in loosely structured political parties, clans gain almost unchecked access to resources at the national level, such as economic rents (cartels, monopolies), patronage, and pork barrel spending [ [32]].

In Thailand, the most effective ‘big man’ networks have evolved around so-called rural godfathers (chao pho in Thai), described by Anderson as ‘mafioso-like politician capitalists who, by the use of violence, political connections, and control of local markets and rackets, become feared provincial bosses’ [ [2]: 42]. Typically, chao pho networks will reach upwards to governmental officials (elected politicians, bureaucrats, police and military officers) and, through the mechanism of vote buying, downwards to the electorate [ [59]: 85]. Competition for access to economic rents controlled by the central government is even fiercer than in the Philippines. Critically, networks controlled by political actors not only compete against each other but also against networks controlled by military officers, with the latter having a hand in marauding practices such as budget and procurement fraud, and illicit business activities (e.g. narcotics trafficking, extortion rackets, illegal bookmaking) [63]. However, once a cabinet position is secured, the same operating logic kicks in as in the Philippines, as political actors ‘use access to corruption revenue (granting concessions, skimming off funds, or auctioning top positions in the bureaucracy) to recoup their elections expenses, accumulate financial resources for the next election and consolidate or build on the number of MPs under their control’ [ [88]: 266].

Finally, Indonesia is a case of ‘official moguls’ corruption – distinguished from ‘oligarchs and clans’ corruption by the fact that the ‘owners’ of corruption networks face very little legal or political restraint when fueling their networks with public and private assets. In Indonesia, “official mogul” networks tend to be controlled by key officials and party apparatchik of the former Suharto regime (1966–1998), who were able to reinvent themselves as parliamentarians and executive politicians, and have used their control over political office to forge informal links with business interests, bureaucrats, organized crime, and even military and police commands [ [27]]. Networks aim to ensure continued access to the state apparatus by recruiting vote brokers, who, in turn, mobilize voters through the delivery of material goods – either through vote buying or clientelistic exchange relations [4]. Moguls unilaterally abuse their position of political power to suck public resources out of the state or to extort assets from private entities. Examples include politicians diverting state funds directly into party coffers [ [53]: 247], senior bureaucrats running ‘semi-formalized’ systems whereby positions in the civil service are ‘sold’ to the highest bidders [7], and military officers being involved in a range of illicit businesses, from illegal logging and mining to protection rackets and smuggling [ [55]: 109].

Towards a sequential argument

To account for the differences in the institutionalization of corruption across East Asia, the remainder of this paper will put forward an ordered sequential argument, in which ‘the temporal order of the events in a sequence is causally consequential for the outcome of interest’ [ [20]: 218]. Specifically, it is argued that when a concentration of the ‘political marketplace’ preceded the introduction of mass suffrage, corruption became institutionalized as the ‘elite cartel’ type; when mass suffrage was adopted in a context of a fragmented marketplace, corruption became locked into either ‘oligarchs and clans’ or ‘official moguls’ forms.

Following de Waal [15], politics can be conceptualized as a set of marketplaces in which ‘sellers’ of loyalty offer their allegiance to the highest bidder. These marketplaces of loyalty operate at different levels of the political system – for example, marketplaces operate within the metropolitan elite but they also connect metropolitan elites to provincial elites. The price that loyalty sellers can extract for their allegiance depends on the value of politically relevant resources they control. In particular, the relative bargaining power of loyalty sellers depends on their stock of mobilizational and coercive resources – in other words, their social control over parts of the population and their ability to resist the enforcement of official policies through violent means.

When mass suffrage is implemented under conditions of a fragmented marketplace – that is, a marketplace in which politically relevant resources are distributed among a larger number of elites – a strong element of uncertainty is injected into the political system. Generally speaking, the introduction of elections always increases the level of uncertainty in elite interaction [see [66]] – mainly because the selection of leaders becomes less predictable and because it is difficult to foresee whether all relevant actors will commit themselves to the new rules of the game. In fragmented marketplaces, however, the uncertainty introduced by elections is disproportionately larger than in concentrated marketplaces. The reasons for this may vary. For example, the marketplace may feature powerful coercive entrepreneurs (such as warlords or militia leaders) who – due to the resources they control – have the ability to sabotage the electoral process and resist the implementation of laws passed by elected governments. Similarly, uncertainty introduced by elections can be higher in fragmented marketplaces as there may be provincial elites in control of significant mobilizational resources (such as landlords). Under circumstances where mobilizational resources are dispersed over a large number of autonomous vote banks, the predictability of election outcomes is very low, as elites in control of these vote banks may unforeseeably decide to change sides.

To lower the level of uncertainty, those elites with an interest in capturing national office will be forced to secure the loyalty of coercive entrepreneurs and vote bank owners. However, there are two issues here. To begin with, coercive entrepreneurs and vote bank owners are unlikely to provide loyalty in exchange for access to formally institutionalized career paths into public office, as it is precisely the weakness of formal public institutions that makes these actors powerful. Alternatively, loyalty buyers could buy off coercive entrepreneurs and vote bank owners one by one; yet the problem here is that such one-off payments do not prevent the loyalty seller from reneging on the agreement.

A more stable solution is to institutionalize corruption on a large scale, governed by informal rules for resource distribution and conflict resolution. As neo-institutional scholars argue [e.g. [14, 25]], the repetition of corruption interactions over time will increase trust between actors and, driven by actors’ anticipation of future exchanges, create incentives to honor agreements. Accordingly, when mass suffrage is introduced in a context of a fragmented political marketplace, the best option for electorally ambitious elites to secure the loyalty of coercive and mobilizational entrepreneurs is to institutionalize corruption in such a way so as to allow the latter to nurture autonomous ‘big men’ and ‘moguls’ networks.

As will be outlined in more detail in the next two sections, when universal suffrage was rolled out across East Asia in the mid-twentieth century, political marketplaces differed significantly from one another in terms of how mobilizational and coercive resources were distributed (see Fig. 1) – with important implications for the institutionalization of corruption.

Fig. 1
figure 1

Political marketplaces at the critical juncture

Full size image

In the Philippines, which saw universal suffrage being established under American colonial rule as early as 1935, mobilizational resources were largely in the hands of landholders who, through patron-client relationships with peasants, controlled large vote banks. Politicians seeking control of executive office at the national level bought the loyalty of these landowning clans by deliberately keeping state institutions weak and allowing ‘big men’ to plug their networks directly into the state for the extraction of public resources. This laid the foundation for the Philippines’ ‘oligarchs and clans’ corruption.

In Thailand, where the implementation of electoral competition was a more protracted affair, the key to understanding the institutionalization of ‘oligarchs and clans’ corruption is the distribution of coercive resources in the post-WWII political marketplace. Critically, the Thai military was highly factionalized, which resulted in a recurring pattern of coups and counter-coups. Hence, when competitive elections were gradually rolled out from the mid-1940s onwards, political elites – driven by the constant threat of military intervention – refrained from investing in political party organizations as mobilizational vehicles. Instead, they turned to the aforementioned chao pho, thus paving the way for corruption to become institutionalized around the latter’s informal networks.

In Indonesia, the political marketplace was also characterized by fragmented control over coercive resources. Specifically, a prolonged guerrilla war against the Dutch colonialists had facilitated the emergence of regional warlords. To incorporate these coercive entrepreneurs into the post-colonial democratic project, the national government saw itself forced to dismantle the bureaucratic state that the Dutch had left behind, thereby nurturing formidable ‘mogul’ networks that still dominate the corruption market in Indonesia today.

In contrast, in Singapore, Taiwan, South Korea and Malaysia, control over both the means of organizing violence and the means of mobilizing political support was much more concentrated. This meant that, when universal suffrage was implemented, political actors striving to gain or maintain control of the state did not have to buy cooperation from a fragmented base of loyalty sellers. The consequence of this was that rulers were able to invest in universalistic procedures and professional norms to govern the operation of state organizations. Stronger political institutions and a greater role for formal organizations, in turn, meant that corruption became institutionalized as the ‘elite cartel’ type.

Once institutionalized in the mid-twentieth century, these different forms of corruption reproduced themselves over time. Confirming the arguments developed by the neo-institutional approach to corruption, processes of reproduction have remained locked into a ‘path-dependent’ trajectory despite significant changes to the formal institutional set-up of political systems. Perhaps most remarkably, ‘oligarchs and clans’ and ‘official moguls’ forms of corruption do not seem to have been negatively affected by processes of political change that replaced ‘electoral authoritarian’ regimes with fully fledged democracies in the 1980s and 1990s. This is contrary to the theoretical expectations of the institutional economics approach, whose proponents argue that the introduction of democratic processes should increase the risks of engaging in corruption – mainly by strengthening both the principal’s monitoring abilities (through a free press and civil society organizations as well as the separation of judicial power) and punishment capacity (through the electoral accountability mechanism).

The path-dependency of ‘big men’ and ‘moguls’ corruption in the Philippines, Indonesia and Thailand can be explained by corruption networks’ ability to hijack and ‘switch off’ democracy’s formal monitoring and punishment institutions. For example, research on the new democracies in Southeast Asia has shown that law-enforcement and judicial authorities are often themselves embedded in complex corruption networks [e.g. [51, 62, 70]]. They thus have no incentive to effectively fight corruption, as this would deprive them of their own extortion and criminal rackets. Similarly, ‘big men’ and ‘moguls’ networks often also succeed in co-opting civil society and the media – either by threatening to deploy violence or by ‘buying’ their support with material incentives [e.g., [80]]. Finally, as already touched on above, corruption networks in the Philippines, Indonesia and Thailand typically extend into the electorate through patron-client linkages. This essentially removes voters’ incentives to punish corrupt politicians in elections.

The sequential argument about the path-dependency of corruption is again summarized in Fig. 2. The key components of this argument are the antecedent conditions (concentration of the political marketplace), the critical juncture (introduction of mass suffrage in the mid-twentieth century), and the reproduction of institutionalized forms of corruption after the end of the critical juncture.

Fig. 2
figure 2

The path-dependency of institutionalized corruption

Full size image

Institutionalizing corruption

The following section will develop the neo-institutional argument in more detail. In particular, it will be shown how – at the critical juncture when universal suffrage was implemented – particular historical processes had led to highly fragmented political marketplaces in some cases and more concentrated marketplaces in other cases. In a second step, it will be demonstrated how these varying degrees of market concentration created different incentives for the institutionalization of corruption.

… in fragmented marketplaces

When universal suffrage was implemented in the Philippines, Indonesia and Thailand, political marketplaces commanded high prices for loyalty and cooperation. In the Philippines, colonialism had resulted in decentralized control over resources for the mobilization of voters, while marketplaces in Indonesia and Thailand featured a significant number of political entrepreneurs who used their control over the organization of violence to sell their allegiance in return for access to state resources. In all three countries, high market rates for loyalty resulted in the institutionalization of either ‘oligarchs and clans’ or ‘official moguls’ corruption.

The Philippines – which was under Spanish possession between the mid-sixteenth century and 1898, and subsequently American control until 1946 – never experienced a significant investment in centralized political structures by colonial authorities. Reflecting its peripheral and unimportant status, during the period of Spanish colonialism the Philippines was administered through a system of indirect rule that relied on the Catholic Church and local strongmen for the exercise of political power [ [76]: 14–15]. The latter saw a significant boost in their power when, in the early nineteenth century, the Spaniards commercialized Filipino agriculture, thus giving strongmen the opportunity to amass vast tracts of land to establish hacienda-like plantations. Thus, by the time the Americans took over, the Philippines had seen the consolidation of provincial fiefdoms built on strongmen’s economic power and social control over local populations through patron-client networks.

The Americans, similar to the Spaniards, did not invest in territory-wide bureaucratic institutions. Instead, driven by the goal of preparing the Philippines for independence and self-government, US colonial authorities prioritized the implementation of elections as a mechanism of political control. And, significantly, the way in which elections were rolled out – starting at the local level and moving up to the national level – allowed provincial strongmen to turn their local fiefdoms into electoral bailiwicks [ [1]: 7–11]. Indirect colonial rule and the particular process of decolonization thus contributed to a political marketplace in which control over mobilizational resources was highly fragmented in the hands of provincial strongmen. As a consequence, politicians striving for national office saw themselves forced to ‘buy’ the loyalty of these provincial elites, which most visibly manifested itself in the emergence of loosely structured political parties that were merely ‘a working alliance of patron-client systems’ [ [46]: 75]. To facilitate the buy-in of provincial elites, state designers created a bureaucratic apparatus that would be only weakly insulated from politicians’ demands. To begin with, national legislators were given wide discretion over the disbursement of pork-barrel funds and patronage appointments in bureaucratic agencies. This not only allowed national politicians to fuel patron-client ties with provincial elites in control of significant vote banks – who, in turn, would use these payments to further strengthen the loyalty of their vote banks [ [34]: 271–272] – but, at the same time, the ‘systems of patronage overwhelmed the capacity of central agencies to supervise lower levels of government’ [ [33]: 296]. The result of the latter was that provincial elites effectively came to exercise ‘monopolistic personal control over coercive and economic resources in their territorial jurisdictions or bailiwicks’ [ [76]: 141]. Overall then, these design choices institutionalized a system of ‘oligarchs and clans’ corruption under which the state apparatus would repeatedly be choked by ‘an anarchy of particularistic demands’ [ [31]: 13] from vertically integrated networks seeking to fuel patron-client ties with public resources.

In Indonesia, unlike in the Philippines, the post-colonial marketplace featured a relatively high concentration of mobilizational resources. Colonialism was again an important factor that shaped these market dynamics. First, while initially making heavy use of indigenous political structures to sustain political power over the Indonesian archipelago, the Dutch colonialists increasingly moved towards a more direct form of administration. For one, during the second half of the nineteenth century, indigenous elites were, step by step, integrated into the colonial Beamtenstaat, which significantly undermined traditional modes of social control [77]. Moreover, although formally retaining a dualistic legal system that featured two sets of courts and laws – one based on European norms, the other one on traditional (adat) laws – in reality, the system became more and more subjected to control from the center, thus effectively stripping customary courts of their powers [48]. Second, the growing importance of the colonial state in managing economic activity produced a system of capitalist agriculture that was dominated by foreign-owned plantations and wage labor [ [67]: 10–15]. This meant that a large number of peasants were able to escape clientelistic practices, which made them available for alternative modes of mobilization by metropolitan elites [82].

However, while control over the organization of political mobilization was thus relatively centralized, the same cannot be said about the control over coercive resources. Most significantly, years of guerrilla warfare against the Dutch meant that local military units had become largely self-sufficient by developing an independent economic base – often in the illegal economy. As Anderson [3] explains, ‘[s]ome provincial military commanders, headed towards warlord status, began to create their own hidden budgets by protecting smugglers, controlling local export revenues and practising extortion’. When the central government sought to turn the revolutionary guerrilla forces into a disciplined and professional army – which would have resulted in reduced informal profit-making opportunities for provincial commanders – this triggered a series of local armed revolts [ [12]: 54–55]. Although successfully crushing the revolts, elites in Jakarta continued to perceive ‘danger of political adventurism by regional military leaders’ and thus largely shelved their reform plans [ [54]: 49]. Instead, they decided to institutionalize corruption as a mechanism to buy the loyalty of coercive entrepreneurs.

The first step in this direction was to abolish the dualistic legal structure inherited from the Dutch in favor of the weaker adat side of the structure, which ‘made it much easier for Indonesia’s military and political elites to gut the legal system of its autonomy and subordinate it to executive power’ [ [89]: 153]. Subsequently – first under Sukarno’s Guided Democracy regime (1957–1966), then to an even greater extent under Suharto’s New Order regime (1966–1998) – army officers were placed in key government and administration roles, and a massive expansion of military business activities was encouraged. Provincial commanders were thus given ample opportunities to fuel their personal networks through corrupt activities – either by ‘selling’ public licenses, permits and contracts to private market actors, siphoning off funds from military-controlled companies, or running illicit operations (such as illegal logging or mining, smuggling, or drug trafficking) [11]. And thus the seed for Indonesia’s “official moguls” corruption had been planted.

In Thailand, too, the fragmented control over coercive resources is key to understand the institutionalization of corruption in the second half of the twentieth century. The reasons for this fragmentation were very different from Indonesia’s historical experience. Starting in the late nineteenth century, Thai monarchs had implemented a number of bureaucratic reforms with the aim of strengthening the state and halting the advancement of Western colonial powers. The king retained the right to appoint and remove senior civil servants at his will, which, in 1932, provoked a coup from young officers, who set themselves the goal of opening up higher ranks in the military to non-aristocrats [ [87]: 173]. The 1932 coup and the abolishment of the absolute monarchy installed the military as the most powerful institution in Thai politics; yet, critically for the discussion here, the military was high factionalized along ideological lines.

Factionalism in the military fragmented the control over the organization of violence. As was reflected in the large number of (successful and unsuccessful) coups in the years before and after WWII, ‘the use of force became the ultimate arbiter of political disputes between rival factions within the military elite’ [ [5]: 106]. It was in this context of heightened insecurity that elections were gradually implemented.Footnote 3 The implications were profound. Most importantly, political elites refrained from setting up well-organized parties to compete in the newly created elections, as they feared that – in the event of a coup – party assets could be seized [ [58]: 255]. Instead, political elites invested in patron-client networks as the main organizational vehicles. ‘Intra-elite rivalry’, as Doner and Ramsay [18] note, ‘meant that networks were critical’. In particular, ‘big man’ networks grouped around senior military leaders who, in order to strengthen their networks, engaged in activities that are characteristic of ‘oligarchs and clans’ corruption: extracting resources directly out of the state to feed into the network and awarding corruption ‘franchises’ to network members [ [74]: 302–303].

With economic development giving rise to new social forces, the number of relevant ‘big man’ networks multiplied. Specifically, the emerging Bangkok-based business elite, in its ambitions to take over the institutions of (semi-democratic) representation, turned towards rural bosses – earlier referred to as ‘godfathers’ (or chao pho) – who, because of the monopolistic quality of their economic activities and/or their involvement in organized crime, were able to deliver large blocks of votes. This led to a system of ‘competitive clientelism’ [18] whereby different networks competed over access to corruption revenues with which to strengthen their client base.

… in concentrated marketplaces

In contrast to the Philippines, Indonesia and Thailand, political marketplaces in Singapore, Taiwan, South Korea and Malaysia had gone through processes of concentration before universal suffrage was established in the mid-twentieth century. As the political marketplaces did thus not contain significant sellers of loyalty when metropolitan elites prepared to organize vehicles for mass mobilization, corruption became institutionalized in the ‘elite cartel’ type. As will be shown in the following section, the ways in which processes of political party formation unfolded within these contexts of concentrated marketplaces can explain the more subtle differences in the institutionalization of ‘elite cartel’ corruption.

To begin with, it can be observed that political marketplaces at the critical juncture of mass mobilization did not feature powerful coercive entrepreneurs. Critically, in all four cases, independence from colonial rule had been achieved through peaceful means. Hence, unlike in Indonesia, post-independence rulers did not face pockets of guerrilla fighters who would have been able to extract a payment for their loyalty. Moreover, compared to Thailand, militaries were relatively cohesive and unified. In the case of Taiwan, this was facilitated by the fact that, when the Kuomintang (KMT) withdrew to the island in 1949 to escape the Communist forces on the mainland, only the most loyal elements of the army followed party leader Chiang Kai-shek [ [23]: 59]. For the case of Malaysia, effective civilian control over the military is frequently attributed to the role that the British played in creating a professional and well-disciplined unit under colonial rule [ [57]: 264]. Singapore, on the other hand, was left without a significant military force when it gained independence from Malaysia in 1965; rather, the military was a creation of the People’s Action Party (PAP) government, which had come to power in 1959. The PAP followed the Israeli model of a citizen army, which put an effective check on any fragmentation tendencies in the organization of violence, as it meant that there were simply ‘no clusters of politically motivated soldiers that could form a base for ambitious generals to exploit’ [ [90]: 161]. Finally, in South Korea, the military – which had been built up with US aid and technical assistance – did show signs of fragmentation. However, military factions were mainly the result of strategic meddling by the first president, Rhee Syngman (1948–1960). The factions did not, unlike in Thailand, reflect ideological differences. As a result, the subsequent regime around Park Chung-hee had no difficulties in destroying these groups and maintaining a centralized monopoly of violence [ [43]: ch. 6].

Not only had control over coercive means become centralized by the time universal suffrage was introduced, but mobilization resources, too, had come to be concentrated in the hands of metropolitan elites. In particular, unlike in the Philippines, political marketplaces did not contain landowners in control of significant clientelistic vote banks. In Taiwan and South Korea, landowning elites were considerably weakened through extensive land reform programs implemented after WWII. In South Korea, land reform was rolled out under intense pressure from US military authorities, driven by fears of peasant unrest and political instability [35]. In Taiwan, on the other hand, land reform was initiated by the KMT in order to consolidate local political support for its ‘alien’ rule. The KMT was able to carry out land reform without serious resistance because, having arrived from ‘outside’, the party enjoyed a great deal of autonomy from local landlords [ [85]: 324].

In contrast, Singapore and Malaysia had historically not been characterized by strong landholding classes. While in Singapore it was sheer geographical size that had prevented the emergence of a significant landed class, the mobilizational capacity of Malay landowners was weakened by the following two factors: first, similar to Indonesia, colonialism had resulted in a large share of agricultural land being controlled by foreign-owned plantations; second, arable land was still in abundant supply in the mid-twentieth century, which offered peasants ample opportunities for squatting and thus a way to escape clientelistic control by landowners [ [8]: 182].

It was in these concentrated marketplaces that political elites designed organizational vehicles for mass mobilization – in particular, political parties. And it was elites’ design decisions that explain why ‘elite cartel’ corruption became institutionalized in different ways.

In Taiwan, the KMT leaders’ design choices were shaped by the Chinese Civil War. Specifically, after being forced to retreat to Taiwan as Communist forces advanced on the mainland, the party leadership came to the assessment that ‘the weakness of the KMT’s own organization [had] contributed to its defeat’ [ [16]: 58]. In an effort to emulate the organizational structures of the victorious Communist Party of China (CPC), the KMT created a dense network of branches and cells to penetrate society and the state bureaucracy. However, as an organization that had ‘colonized’ a people who did not feel Chinese, the KMT faced a particular problem – that is, the question of how to root itself in the local population. The KMT addressed this challenge by developing clientelistic linkages with local factions, thereby laying the foundation of Taiwan’s ‘elite cartel’ corruption.

In Singapore, the PAP – rather than emulating the ‘mass party’ model – made moves to weaken its grassroots foundation after winning control of the government in 1959. This was because moderate party leaders feared that leftist factions could use the grassroots network as a launch pad for a power grab. Instead, PAP leaders worked towards fusing the party with the state – most notably, by assigning to local government units (such as Community Centres) the functions of party branches [ [52]: 95–96] and tying the provision of constituency-specific programs – such as infrastructure, social welfare provision, and housing subsidies – to electoral support for the PAP [ [75]: 389]. Thus Singapore’s particular mode of ‘elite cartel’ corruption, characterized by collusion between political and bureaucratic elites, was established.

Crucially, political elites in Taiwan and Singapore decided against nurturing a domestic capitalist class – for different reasons: while KMT leaders feared a strong Taiwanese business elite that could have developed into a threat to the party’s dominant position [ [10]: 150], the PAP leadership – driven by the reality that the domestic market was too small to support a program of import substitution industrialization – adopted a strategy of export-oriented industrialization based primarily on investment by multinational companies [ [69]: 141–142]. This contrasted starkly with South Korea and Malaysia, where state-led programs of industrialization gave rise to significant domestic business sectors in the 1960s and 1970s. Private business came to function as an important dispenser of illicit funding for the respective regime party, with significant implications for the institutionalization of ‘elite cartel’ corruption.

In South Korea, the Rhee regime planted the seeds for the emergence of large business conglomerates (the aforementioned chaebol) by selling Japanese-owned assets to loyal supporters for a fraction of their real worth and by gearing the US-funded program of import substitution industrialization towards cronies; however, it was the Park regime that – by paving the way for the inflow of commercial Japanese loans and by shifting to more capital-intensive export-oriented industrialization – really accelerated the growth of the chaebol. The government’s main tool for industrial coordination was its tight control over the allocation of credit, which had been made possible by nationalizing the banking sector. As briefly mentioned above, this quasi-monopoly on credit allowed the Park regime to institutionalize a system of bribery, whereby high-ranking government officials would provide the chaebol with access to loans in exchange for the payment of fixed kickback rates. The illicit payments extracted from private business were an important source of funding for the regime’s Democratic Republic Party (DRP) [ [29, 40]: 133–134] – and continue to characterize ‘elite cartel’ corruption in South Korea nowadays.

In Malaysia, the UMNO initially relied on its links with the cash-rich Malaysia Chinese Association (MCA) – a coalition that was formally registered as the Alliance Party – for the funding of its operations. The UMNO became more financially independent after the introduction of the New Economic Policy (NEP) in 1969 – a program aimed at promoting ethnic Malay entrepreneurs to help reduce the dominance of ethnic Chinese over the country’s business sector. For one, the UMNO set up a number of party-owned enterprises. Second, and more important for the discussion here, party leaders used their control over NEP resources to tie the newly nurtured Malay business elite into informal reciprocity networks. Over time, corrupt exchanges between the two sides took on an increasingly institutionalized nature, thus breeding strong ‘elite cartel’ networks that continue to serve as the backbone of corruption in Malaysia today [see [24, 39]].

Comparative perspectives and conclusion

While the emerging neo-institutionalist approach has considerably contributed to our understanding of corruption in the developing world, the question of how we can explain differences in the institutionalization of corruption has so far received very little attention. To address this gap in the literature, the paper here developed a systematic comparison of seven developing and newly industrialized countries in East Asia. The argument that emerged through the analysis is that different corruption types were institutionalized during critical junctures around the middle of the twentieth century when political elites, following the introduction of universal suffrage, had to make strategic choices regarding the organizational design of vehicles for mass mobilization. These choices were conditioned by the political marketplace: where the bargaining power of loyalty sellers was high, cooperation in mass mobilization had to be ‘bought’ through ‘oligarchs and clans’ and ‘official moguls’ corruption; where political marketplaces did not feature significant loyalty sellers, corruption became institutionalized in ‘elite cartel’ forms. In other words, historical sequencing mattered: countries in which the political marketplace had gone through a process of concentration before universal suffrage was introduced (Singapore, Taiwan, South Korea, Malaysia) are now marked by less harmful types of corruption than countries where mass voting rights were rolled out in a context of fragmented political marketplaces (Philippines, Indonesia, Thailand).

Even though this argument was developed within the context of a limited geographical sample, there are good reasons to suggest that the paper’s findings can be generalized to the developing world as a whole. To begin with, we can establish that ‘oligarchs and clans’ and ‘official moguls’ corruption are the most prevalent types of corruption in the developing world. This is not only based on Johnston’s [36] own research but also on the work of regional studies experts that highlights how corruption in the developing worlds tends to be structured around fierce competition over public resources between powerful networks, such as ‘big man’ networks in Sub-Sahara Africa [e.g., [81]] or camarillas in parts of Latin America [e.g., [56]]. Moreover, we know that, typically, mass suffrage was rolled out under conditions of highly fragmented political marketplaces. In Latin America, most countries had adopted universal male suffrage at the end of the nineteenth century, at a time when powerful landlords continued to control the vote of the rural peasantry and, in some cases, caudillo warlords remained a powerful force [see [19]]. In Sub-Sahara Africa, elections were, broadly speaking, introduced with decolonization in the mid-twentieth century – yet again, political marketplaces were generally characterized by high levels of fragmentation. This was primarily due to colonialism, which – as Berman [6] explains – ‘rested largely on the institutionalization of “Big Man-Small Boy” politics in rural society, built on the hierarchies of personal rule of the “decentralized despotism” of chiefs and headmen.’ In other words, colonial rule had – by sponsoring local strongmen – led to a large number of powerful loyalty sellers whose support in electoral contests could only be secured by keeping formal institutions weak and providing ample opportunities for corruption.

In addition, not only do these broad patterns support the generalizability of the sequencing argument developed in the preceding paper, but further evidence can be produced through a brief comparative sketch of four countries that are frequently referred to as anti-corruption overachievers – Chile, Uruguay, Costa Rica and Botswana.

To begin with, at the time when universal suffrage was adopted in these four countries, control over mobilizational resources was relatively concentrated. In the cases of Uruguay, Costa Rica and Botswana, this was largely due to the fact that, unlike in the case of the Philippines discussed in much detail earlier, the social fabric in rural areas did not provide a fertile seedbed for patron-client relations: while in both Uruguay and Botswana agriculture was dominated by labor-light cattle ranching, most farmers in Costa Rica were small holders. In Chile, clientelism did structure social relations in the countryside. However, universal suffrage arrived comparatively late (1925). Crucially, male universal suffrage was introduced after industrialization had given rise to a large working class, which, in turn, meant that – by the time more extensive voting rights took effect – urban-based mass parties had built up considerable mobilizational resources. These parties then extended their organizational structures into the countryside and thus broke landowners’ hold over the peasantry [49].

Moreover, in all four cases, control over coercive resources was, unlike in the cases of Indonesia and Thailand, not fragmented when universal voting rights were implemented. Both Chile and Costa Rica are usually seen as exceptional in the sense that they largely avoided the problem of caudillismo that gripped most of Latin America after decolonization; similarly, independence in Botswana did not, unlike in many other countries in sub-Saharan Africa, trigger civil conflict or a succession of military coups. Uruguay did go through a long period of warlordism after the end of colonial rule; however, the central state established a firm monopoly on the use of violence by the end of the nineteenth century – that is, at least two decades before universal voting rights for men were adopted (1918).

In short, this paper has presented considerable evidence that differences in the institutionalization of corruption between developing countries have historical origins. More generally, this supports the neo-institutional argument that corruption – once institutionalized – is subject to path-dependent effects.

The lessons for the analysis of corruption are manifold. Most importantly, combining Johnston’s [36] corruption typology with neo-institutionalist theories of corruption – in particular, the transaction costs variant [e.g., [14, 25]] – highlights that, in contexts were corruption is systemic, corruption tends to be organized collectively by informal networks whose dense links reach deep into the state to extract resources for the strengthening of particularistic loyalties. This network logic means that the individualistic risk-benefit calculation on which the institutional economics approach is based has little analytical value . Specifically, networks significantly lower the risk of corruption by harboring certain resources for the facilitation of corruption (such as know-how and information) and by generating social capital between network members. Once the organization of corruption has become institutionalized around ‘oligarchs and clans’ or ‘official moguls’ networks, political reforms designed on the basis of methodological individualism (as put forward by the institutional economics approach) will do very little to combat corruption. This is because corruption networks, for example, possess the ability to capture democracy’s key institutions, thereby severely undermining these institutions’ monitoring and punishment capacities.

Future research on the effectiveness of anti-corruption measures in the developing world is thus well advised to not only replace the focus on the individual with a greater concern for the role of networks, but also to pay more attention to the qualitative differences in how corruption markets are organized, rather than relying on quantitative indicators derived from expert-based assessments.