Corruption: An Uneven Field of Research—Between State and Private Topics

Abstract

Research on state corruption has flourished since the 1990s; however, studies focused on corruption within non-state organizations are still limited. In this study, we conducted a systematic review of 18,435 articles from the Web of Science database, covering the years 2002 to 2020. Using topic modeling Latent Dirichlet Allocation (LDA), we analyzed the field of corruption research. Our analysis identified four main dimensions: state corruption as the predominant field, private-to-public corruption, private-to-private corruption, and technological–biological corruption. Our findings indicate that state corruption has a well-established research tradition, whereas private corruption remains underexplored. We highlight key conceptual limitations in understanding the mechanisms of non-state corruption and propose the idea of operational deviation from regular procedures to address these issues. This article concludes that further empirical research is needed on non-state corruption to develop a conceptual framework specific to this area, which features distinct characteristics from state corruption. Finally, we suggest implications for policy and practice based on our findings.

1. Introduction

Over the past 30 years, there has been significant advancement in studies and interest regarding the fight against state corruption, following the challenges outlined by Transparency International in 1993. One notable indicator of this growing interest is that the organization has established more than 100 national chapters since its founding in Germany. This increased focus is also evident in the rise in academic productivity. In the late 1990s, there was a surge in publications addressing public and political corruption [1]. During this time, influential studies emerged that investigated the causes and effects of corruption, with a particular emphasis on its consequences [2,3] and its impact on economic growth and development [4,5,6].

However, while research on state corruption has developed relevant conceptual frameworks, studying corruption in non-state contexts (private and non-profit organizations) remains notably undertheorized. Corruption studies in private contexts tend to focus narrowly on the corporate setting, analyzing the mechanisms by which executives exploit their positions to gain advantages in the private [7] or in markets through asymmetric information [8]. In the case of non-governmental organizations or non-profit entities, research has been relatively low [9,10,11]. Studies on organizations such as the UN, UNICEF, Greenpeace, WWF, or even FIFA are primarily found in books, essays, or journalistic chronicles, and the information available in the leading scientific repositories shows insufficient development in the academic literature. These approaches often assume that corruption in private contexts occurs through the same causes and mechanisms as those investigated in the state sphere, namely, an abuse of public office or a position of power to obtain private gains.

Considering the rise of transnational corporations and foundations in contemporary society, many of which have budgets and profits surpassing those of several nation-states, the increase in private global transactions, and the way both impact the public interest, in this article we argue that private and non-state corruption presents an analytical subordination to the widely developed state corruption research that hinders the further development of investigation on private and non-state corruption. This analytical subordination is characterized by (a) the lack of an applicable definition that addresses corruption when it involves only private interests and (b) the absence of a specific inquiry into the mechanism that produces corruption in this domain.

Consequently, we propose two related conceptual innovations. First, we define corruption as an operational deviation from regular procedures—whether in public or private sectors—a comprehensive concept applicable across various institutional frameworks. Second, this deviation suggests a mechanism that underlies the act of corruption. It is expected that the actions and behaviors of organizations and individuals will generally align with recognized legal, ethical, or technical norms within a specific social context, whether local, national, or transnational. However, situations such as the exchange of favors, the use of money or power to gain particular returns or benefits, and private actions like corporate collusion aimed at undermining free competition or exploiting uninformed customers all represent deviations from regular organizational operations and procedures. Other examples include unilateral contract renegotiations in the retail sector without informing the public or diverting foundation funds to support hidden causes or sustain internal bureaucracies. These instances illustrate how actions that deviate from established norms can lead to disappointment in expectations. This approach provides a nuanced understanding of corruption, acknowledging the differences between the public and private sectors while offering a common framework for analyzing corruption events and their interrelations. We also argue that the mechanism underlying private and non-state corruption emerges from a regulatory void, which private organizations attempt to address through self-regulation, such as establishing compliance rules. However, this regulatory void tends to prioritize strategic and instrumental interests over adherence to self-regulatory procedures, resulting in corruption as a deviation from these standards.

Building on this argument and proposed innovations, we systematically examine the academic literature on corruption using the topic modeling algorithm known as Latent Dirichlet Allocation (LDA). We analyzed 18,435 scientific articles indexed in Web of Science from 2002 to 2020 that include the concept of “corruption” in their title, abstract, or keywords. Through this method, we empirically reveal the thematic structure of the field of corruption and its subcomponents. This analysis allows us to identify and characterize the dominance of a state-centered approach and the conceptual gaps in understanding non-state corruption, as well as support the concept and mechanism of operational deviation as a general mechanism of corruption.

The overall findings indicate that the literature on corruption can be categorized into four dimensions, each with varying relevance. The first dimension is state corruption, which comprises 55% of the papers analyzed. Well-known cases in this regard involve the misuse of public power for personal gain. The second dimension is private-to-public corruption, accounting for 21% of the total. In this case, private companies bribe public officials to secure contracts or obtain regulatory licenses. The third dimension is private-to-private corruption, making up 11% of the literature. This involves companies or employees colluding to defraud the public or violate private or governmental regulations. A key example that combines both private-to-private and private-to-public corruption is the FIFA scandal. Exposed in 2015, this scandal revealed a systemic network of corruption involving FIFA officials, commercial partners, sponsors, national federations, politicians, and even states. Finally, the fourth dimension is technical–biological corruption, which constitutes 13% of the literature and is comparable in significance to the previous dimension, though it operates in a more technical domain. We highlight this topic for two reasons: it emerges naturally from the reviewed data, and the mechanisms used in this technical–biological area to manage or anticipate critical events—such as redundancy, modularity, and early warning signals—are also applied in organizational risk management. Everyday examples of data corruption can escalate into catastrophic failures in critical systems. This includes degraded data in financial trading systems that can lead to market crashes, compromised aviation control software that can result in accidents, and major socio-technical disasters at nuclear power plants. Notable examples include the incidents at Three Mile Island and Chernobyl, where deviations from established procedures had devastating consequences. Across all identified topics, including the technical–biological category, corruption can be understood as an operational deviation from regular procedures that undermines expectations of reliability and consistent performance.

To unfold this argument, we conduct a thorough review of the academic literature on corruption in three steps. First, we present our materials and methods. Next, we describe and analyze the four approaches and their subcomponents that arise from the LDA topic modeling of the literature on corruption. Finally, we discuss our findings, placing particular emphasis on the literature concerning non-state corruption and our definition of corruption as a deviation from regular procedures. We draw the main conclusions from our analysis.

2. Material and Methods

2.1. Data

To map the field of corruption research, we conducted an exhaustive analysis of scientific production using the Web of Science (WoS) database. The indexed articles that address corruption in the main collection of the Web of Science (WoS) between 2002 and 2020 include 18,435 observations.1 To select them, we used the term “corruption” in the platform’s search engine, including papers containing this term in any of their key data (i.e., title, abstract, keywords). These results were obtained by filtering only scientific articles, excluding other types of documents such as book reviews or editorial material. This approach allowed us to focus on high-quality, research-based contributions to the field.

The corpus of the literature on corruption includes studies from both the social sciences and humanities, as well as from disciplines not commonly associated with this concept, such as informatics or biology. Figure 1 shows the top ten disciplines (WoS categories) containing publications on corruption.2

Articles from economics show prevalence, with 2173 articles representing about 12%. They are followed by studies from political science, with 7%, and law, with 5%. Studies from public administration do not far outnumber those from business, finance, or management. In fact, there are two separate business categories, and when combined, they exceed the number of publications in public administration.

2.2. LDA Topic Model

Text mining, specifically topic modeling, was used to identify and categorize the main literature topics. Silge & Robinson [12] summarize it as a method for analyzing documents and classifying them in an unsupervised way. There are different methods to perform this exercise.3 Here we used the Latent Dirichlet Allocation (LDA) algorithm. Given the conceptual breadth of the analyzed field, LDA was particularly suitable due to its clustering mechanism and probabilistic approach that effectively captures the underlying thematic structure of documents. Each document is processed as a probabilistic mixture of topics, with each topic representing a distribution of conceptually related terms.4

LDA is designed to systematically map large textual corpora. It enables the identification of emerging topics in the literature without imposing predefined categories, which is particularly valuable for mapping a wide and conceptually diverse field such as corruption studies. This methodological approach allows us to recognize common patterns in the literature, uncover thematic structures, detailed subcomponents, and distinctive approaches across disciplines [18,19,20]. For our case, LDA provided a clear empirical distinction of the main conceptual dimensions that structure the literature and its subcomponents, revealing predominant research topics and underdeveloped areas. Figure 2 shows the flow of this process.

The implementation of our LDA analysis began with the data collection and a comprehensive pre-processing of the corpus. The entire corpus was converted to lowercase to ensure consistency in term recognition. We then removed punctuation marks and numerical values that could interfere with the semantic analysis. Common English stop words were eliminated to focus on meaningful content. Additionally, we expanded our analysis beyond single words by incorporating both unigrams and bigrams, allowing us to capture meaningful two-word phrases that might carry specific technical or conceptual significance. A Document-Term Matrix was then constructed to represent the frequency of terms across all documents in the corpus. As a final pre-processing step, we conducted additional cleaning to remove publisher-specific terms that could potentially bias the analysis.

While LDA can categorize and group the abstracts of articles, the algorithm needs to be informed of the number of topics to identify. This can be performed by an expert or by probabilistic coherence [21]. To determine the optimal number of topics for our corpus, we implemented an iterative modeling approach. This process involved testing multiple LDA models with varying numbers of topics. The final model parameters were calibrated to ensure robust results, employing 500 iterations for model convergence, a 180 iteration burn-in period, and optimized alpha and beta parameters of 0.1 and 0.05, respectively. The quality of each model iteration was assessed through multiple metrics, including probabilistic coherence scores, log-likelihood plots for convergence evaluation, and topic prevalence analysis. Each model calculated by the LDA had its coherence statistic for the number of topics indicated, so it was possible to identify the best fit. In Figure 3, we present the exercise to estimate the best number of topics for this corpus. We calculated the models from 10 topics up to 100 (i.e., the calculation was performed for 10, 20, 30, and so on up to 100), with the grouping into 50 topics being the most coherent.

Once the 50 topics were identified, we conducted an exhaustive analysis of the concepts included in each of them (see Appendix A). Two topics were eliminated, as they were made up of concepts typical for the structure of scientific articles and did not provide relevant information. We obtained 48 topics and provided names for them. To facilitate this process, it is possible to estimate the most relevant bigrams (dyads of words) in the main articles of the topic, allowing the names given to the topics to reflect the articles’ content.

2.3. Clustering

Finally, to interpret the topics coherently, we grouped them according to the closeness of themes, identifying the main trends and approaches. While we began with a term frequency–inverse document frequency (TF-IDF) method, which measures the importance of a term by comparing how frequently it appears in a specific document versus its frequency across the entire corpus, this quantitative analysis served primarily as an initial framework. This initial analysis using Ward’s method [22] suggested an optimal organization of 3–4 clusters. However, after an extensive qualitative review of the papers and their theoretical contributions, we expanded our analysis to examine 12 distinct clusters to achieve a more granular understanding of the field. With this measure of distance and similarity calculation between concepts, we constructed a dendrogram (hierarchical clustering classification) (see Figure 4).

This hierarchical visualization shows how the 50 topics identified through LDA (t_1 to t_50) are grouped according to their conceptual similarity. The height of the connections indicates the conceptual distance or proximity between topics, allowing the technical identification of 12 main clusters (Figure 4). This representation is a standard in bibliometric studies that apply similar methods [23,24,25].

While the 12 clusters represent a significant reduction from the 50 originally identified, it is still a large number for a comprehensive qualitative analysis of the field. The development of our final clustering framework involved several interconnected analytical stages that prioritized qualitative understanding while being supported by quantitative methods. We applied Ward’s hierarchical clustering method to organize topics into meaningful groups [26,27]. The key step in our analysis was the detailed qualitative review of the most relevant papers within each topic. This in-depth review of conceptual contributions and empirical findings was crucial for understanding the relationships between topics.

After carefully examining and labeling these 12 clusters based on their keywords and thematic content and, most importantly, after a thorough analysis of the theoretical frameworks and findings presented in the key papers of each topic, we were able to reorganize these into four comprehensive conceptual categories: (a) state corruption, (b) private-to-public, (c) private-to-private, and (d) technical–biological topics. This categorization emerged primarily from our qualitative analysis of the literature, with the correlation matrix and silhouette scores serving to validate our theoretically driven organization. This organization aligns with the initial Ward method suggestion of 3–4 optimal clusters, while incorporating the deeper theoretical insights gained from our qualitative review. Figure 5 summarizes the topics’ nature and meaning. This is the depicted visualization of how these 12 clusters are organized by content into 4 main clusters (different colors), offering a visual representation of the dimensions that structure the thematic landscape of corruption research.

Figure 5 shows the 48 topics (represented by colored circles containing specific topic IDs, from t_01 to t_50, excluding t_03 and t_47 which were removed as they contained non-substantive information) grouped into 12 combinations of clusters—e.g., in light green (center bottom), t_11, t_29, and t_49 form a combination of clusters called ‘Business Corruption’, and t_04, t_08, and t_34, form another combination called ‘Crime and Finance’. The size of each topic circle is proportional to its prevalence in the corpus. The 12 combinations of clusters are organized into 4 main dimensions identified through Ward’s method. The bar chart at the bottom displays the percentage distribution of articles across the four dimensions.

3. Results

As a result of the method applied, we organized our findings into four conceptual dimensions to provide a comprehensive description of the field. As said, they are (a) state corruption, (b) private-to-public, (c) private-to-private, and (d) technical–biological topics. This conceptual organization allows us to better track how corruption manifests in different contexts and extract relevant insights to support our argument.

3.1. State Corruption

State corruption represents the most prevalent dimension in our analysis.5 A paradigmatic example is the systematic diversion of public resources by officials for personal or political gain, undermining institutional integrity and civic trust. This dimension addresses the traditional and more extended view on corruption concerning how public power is used for private gains and benefits through institutional structures and governance mechanisms. Accounting for 55% of all analyzed papers, it examines how state actors (such as government officials, politicians, and authorities) use power for personal gain or for the benefit of specific groups. The dimension encompasses both grand corruption at high levels of government and petty corruption in daily administrative operations.

The Governance and Institutional Quality cluster (17.90%) emerges as an elemental theoretical framework. Research on institutional quality and corruption (t_06, 4.0%) has established quality of government as fundamentally dependent on institutional impartiality [34], providing a framework for understanding how institutions either enable or constrain corrupt practices [34,35,36,37]. This theoretical understanding informs anti-corruption research (t_20, 2.78%), where studies have consistently shown that isolated measures like transparency are insufficient without broader institutional reforms [1,2,37,38]. The interconnected topics on institutional reforms (t_23, 2.72%) and partisan corruption (t_27, 2.48%) reveal how political dynamics shape reform effectiveness [34,36,39]. The studies in this cluster are regionally distributed, with analyses presented in Europe, Asia, Latin America, the United States, and Africa.

The second largest cluster of research (17.15%) is called Regional Variations and contains studies in Eastern Europe (t_17, 3.09%), Latin America (t_46, 2.54%), post-Soviet countries (t_30, t_36, t_37, cumulative 5.77%), Africa (t_07, 1.97%), and East Asia (t_43, 1.91%). Research in these contexts has documented how corruption exposure affects political legitimacy and social trust. These regions serve as natural laboratories for understanding how different institutional contexts shape corruption patterns. Studies of post-Soviet states have been particularly influential in developing concepts like “predatory policing” to explain institutionalized corruption [40], while revealing distinctive patterns of state capture and informal networks. Research on East Asian and African contexts highlights how cultural and historical factors influence corruption manifestations [41,42,43].

The Measurement and Effects cluster (11.47%) represents the empirical backbone of corruption studies. Effects of corruption (t_13, 4.67%) emerges as the most frequent individual topic, with studies quantifying both direct costs and indirect impacts on economic growth and social development. This connects with studies on corruption perceptions (t_12, 2.74%), revealing how public trust and institutional legitimacy are shaped by corruption experiences. These measurement challenges, further explored in t_21 (2.56%), underscore the need for more sophisticated methodological approaches [44,45,46].

The Public Administration cluster (8.76%) examines organizational manifestations of corruption. Studies of public officials (t_18, 2.73%) reveal how corruption becomes normalized through organizational processes, with a particular focus on bribery patterns and behavioral aspects. Research on public procurement (t_45, 2.28%) and local government (t_15, 1.98%) demonstrates corruption’s impact on administrative processes and public resource management [47,48,49].

These findings reveal three key patterns in state corruption research: (a) well-developed conceptual and theoretical institutional–normative analysis understanding the corruption mechanism as the use public power for personal or group benefits, thus focusing on reform design to control this behavior; (b) the empirical–descriptive study of corruption impacts; (c) the comparative analysis of regional variations, which demonstrates that Europe and Latin America show political legitimacy problems because of corruption, informality and institutional corruption rein in post-Soviet countries, and corruption practices are culturally embedded in Africa and East Asian countries. The significant presence of anti-corruption and reform-related topics particularly in European and Latin American countries indicates a strong orientation toward practical solutions and policy recommendations. However, the prevalence of regional studies suggests that universal anti-corruption solutions may be ineffective without careful adaptation to local contexts. This multi-faceted nature explains both the diversity of research approaches and the persistent challenges in developing effective anti-corruption measures.

3.2. Private-to-Public

Accounting for approximately 21% of the corpus, private-to-public corruption represents the second most prevalent dimension in our analysis. A paradigmatic example of this is a private business company that offers bribes or undue benefits to public officials to secure government contracts or obtain favorable regulatory decisions, thereby distorting competition and public interest. This dimension is characterized by the complex interplay between business interests and state functions, revealing how private entities interact with public institutions in ways that undermine governance effectiveness and social welfare [50,51,52].

The Social Norms and Institutions cluster (7.27%) emerges as the most prevalent research focus. Studies on corruption and social capital (t_39, 3.10%) examine how social networks and institutional practices enable or constrain corrupt interactions between private and public actors. The studies demonstrate how these interactions affect organizational legitimacy and strategic decision-making [50,53,54]. Further studies show how corruption emerges as situational action within specific institutional contexts [55,56] while other research reveals patterns of collective corruption in organizational settings [57,58]. Research on corruption in religious institutions (t_41, 2.31%) and media networks (t_33, 1.86%) highlights how private institutions can either legitimize corrupt practices or serve as accountability mechanisms of public institutions [11,59].

The cluster Economic Development and Resources (5.44%) represents the second largest group. Research on corruption and investment (t_22, 2.04%) has documented how institutional distance (i.e., differences in normative, regulatory, and cultural frameworks) and natural resources influence foreign direct investment patterns [52,60]. Studies of corruption and inequality (t_14, 1.90%) demonstrate how resource exploitation and corrupt practices contribute to economic disparities [61,62,63], while research on rent-seeking behavior (t_02, 1.50%) examines how private actors extract benefits through political connections [64,65,66].

The Regulatory Enforcement cluster (4.57%) investigates the intersection between private interests and public regulatory frameworks. Research on tax evasion (t_16, 1.72%) reveals how corruption undermines fiscal systems, exemplified by Cerqueti and Coppier’s [67] analysis of tax revenues and fiscal corruption. Studies of corruption in aid (t_26, 1.67%) demonstrate how informal institutions shape entrepreneurial behavior in economic development [51,68], while research on real estate projects (t_44, 1.18%) examines corruption in infrastructure development.

The Public Service Delivery cluster (3.27%) analyzes the impact of the private-sector corruption on public welfare. Studies of environmental corruption (t_10, 1.70%) document how private interests compromise environmental protection, particularly in resource-rich regions [69,70]. Research on corruption in public health systems (t_25, 1.57%) reveals how informal payments and systemic corruption affect healthcare access and quality [71,72].

All these clusters are based on the model of state corruption, in which public office is viewed as a source of corruption. This may make sense in some cases, such as the bribery of public officials to obtain private benefits. However, for private-to-public corruption to occur, the first barrier that must be overcome is the one established by the private organization itself through compliance norms or codes of ethics. In other words, the problem does not lie solely in the deviation from public standards but also from those that private actors impose upon themselves [73,74,75]. This concept can be analyzed through the idea of an operational deviation from private compliance norms or publicly codified ethical standards.

These findings reveal three fundamental patterns in private-to-public corruption research: (a) the critical role of informal networks in facilitating corrupt exchanges, (b) the systematic exploitation of regulatory gaps by private actors in the public sector, and (c) the detrimental impact of private-to-public corruption on public service delivery and social welfare. The significant presence of social norms and institutional research suggests that private-to-public corruption operates through complex relationships that override public–private boundaries.

In this regard, private-to-public research requires some level of conceptualization. The focus of private-to-public corruption lies not only in the use of public power for personal or group benefits but also in the need to undermine the company’s own internal regulatory expectations, whether ethical or compliance-related. Both situations motivate the deviation from established procedural standards that are known or expected by relevant publics (stakeholders, subcontractors, competing organizations, the general public), whose expectations are ultimately disappointed. Social networks provide access to public agents, economic incentives drive corrupt behavior, regulatory weaknesses create opportunities, and public service deterioration as well as loss of trust in business environments represent societal cost. This comprehensive understanding suggests that effective anti-corruption strategies must address both formal institutional frameworks and informal social mechanisms simultaneously.

3.3. Private-to-Private

Private-to-private corruption represents the third dimension in our analysis, comprising around 11% of the corpus. It includes corrupt practices that occur between private-sector actors without direct state involvement, such as collusion between competing companies to fix prices or manipulate private bids, as well as commercial bribery, where an employee accepts payments from another company to gain an undue advantage (e.g., securing a contract or leaking confidential information). This dimension reveals how corruption manifests within and between private organizations, affecting market competition, corporate governance, and business ethics.

The Business Corruption cluster (6.37%) emerges as the most studied area. Research on corruption in business (t_29, 2.76%) examines how firms engage in corrupt practices with other private actors. However, the studies show how political connections with corporate networks facilitate preferential treatment in private institutions [76,77,78], thus revealing the adoption of the public-sector model to understand private corruption. The literature highlights how corruption influences economies with institutional voids [51,68], which may work as a strategic variable for low transaction costs regarding entrepreneurship.

The analyses also show how organizational structures enable corrupt practices, with Gao [79] demonstrating how mimetic isomorphism influences bribery decisions in transitional economies in China. Studies of compliance in the USA (t_49, 1.81%) examine the effectiveness of regulatory frameworks like the FCPA, with Magnuson [80] analyzing challenges in international enforcement. Research on corporate corruption (t_11, 1.81%) highlights how financial-sector governance shapes corrupt behavior, particularly in banking relationships [47]. The accounting and auditing literature contributes to the understanding of private corruption [47,78,81], showing how accounting practices can either facilitate or prevent it. Auditing serves as a detection mechanism, while accounting fraud represents a specific form of corruption based on the concealment of relevant information. This cluster emphasizes the role of institutional pressures and organizational responses in shaping corrupt practices.

The Crime and Finance cluster (5.02%) examines the intersection between private-sector corruption and illegal activities. Research on corruption and organized crime (t_08, 1.92%) reveals how legitimate businesses can become entangled with criminal networks [82]. Studies on money laundering (t_34, 1.70%) show how financial institutions become complicit in corruption through complex transaction networks, while research on corruption in sporting events (t_04, 1.40%) provides evidence of how betting markets and sports management create opportunities for match-fixing [83,84].

These findings unveil three critical patterns in private-to-private corruption research: (a) the influence of industry-specific contexts on corruption opportunities, (b) the concealment of information to gain advantages over competitors or clients and to cover up the traces of the operational deviation, and (c) the increasing integration of criminal–business networks. The significant presence of business corruption research suggests that social contexts, organizational culture and governance mechanisms are crucial determinants of corrupt behavior. Meanwhile, the substantial focus on criminal activities indicates growing concern about the criminalization of business practices. In these cases, the conceptualization of state corruption falls short in capturing the diverse incentives and motivations for corruption that emerge in private settings. At times, corruption is aimed at circumventing national or transnational rules of free competition; in others, it targets the exploitation of customer ignorance and the capture of clients; and in others, it involves negotiations with criminal organizations that control a territory in which a private company has interests. In all these cases, an operational deviation from regular procedures takes place, which must be concealed through further operational deviations related to the management of information—towards competitors, the public, and often the upper levels of multinational corporations.

The research in this dimension demonstrates how private-to-private corruption operates through multiple organizational levels: individual decision-making, corporate governance structures, and inter-organizational networks including political actors. The studies show how corruption becomes normalized within organizations through institutionalization, rationalization, and socialization processes, revealing the complex nature of private-sector corruption [56,57]. The interconnections between clusters suggest that private-to-private corruption requires understanding the mechanism that shapes corruption practices combining formal organizational structures and informal networks. This dual nature explains both its resilience and the challenges in detection and prevention. Research indicates that effective anti-corruption measures must address both the organizational conditions that enable corruption and the criminal networks that exploit these conditions [58,82].

3.4. Technical–Biological Corruption

Technical–biological corruption emerges as a distinct dimension, representing approximately 13% of the literature.6 This category illustrates a use of the term that extends to systemic degradation or malfunction in technical or biological fields. For example, the concept of “data corruption” in computer systems describes how alterations in transmission or storage compromise the integrity of information. Similarly, in biology, “cellular corruption” refers to cases where pathogens alter the normal functions of a cell, diverting its standard operational processes. Unlike other dimensions focused on human behavior, this dimension conceptualizes corruption as an inherent challenge in complex systems design and maintenance. While the technical–biological corruption is of a different nature than human-related issues, the meaning of the concept remains: there is an operational deviation from regular functioning standards.

The Data and System Corruption cluster (10.55%) dominates this dimension, representing the largest body of research. Studies of facial recognition and processing (t_28, 2.04%) examine how visual data can become corrupted, with [85] developing frameworks for robust facial recognition that can withstand data corruption. Research on image corruption (t_24, 2.00%) investigates how visual information can be degraded and methods for maintaining image integrity.

Data integrity research (t_38, 1.74%) investigates how information systems can be compromised by external interventions, while studies of network and security corruption (t_32, 1.67%) examine vulnerabilities in communication systems. Research on computer systems (t_48, 1.54%) and signal corruption (t_05, 1.54%) complete this cluster, revealing the multiple technical levels at which corruption can occur. For example, some studies demonstrate how input uncertainty and data corruption affect modeling outcomes [86], highlighting the importance of robust methodological approaches.

The Biological Corruption cluster (2.25%), though smaller, reveals important insights into how corruption manifests in living systems. Research on cellular corruption (t_35, 1.13%) examines how biological processes can be compromised by pathological agents, particularly in the context of disease development. Studies of biological processes (t_50, 1.12%) investigate how system integrity can be maintained despite corrupting external influences.

These findings highlight two main dynamics in the study of technical–biological corruption: (a) the need to safeguard system integrity against corruption and (b) the development of robust alternatives to maintain functionality in the presence of corruption. The term “corruption” here is not just a linguistic curiosity; it provides a valuable framework for dealing with organizational corruption. This perspective views corruption as deterioration, degradation, or outright destruction, which aligns with classical notions—specifically the Aristotelian view, where corruption (phthora) refers to a substance losing its essential form. This understanding involves deviations from established procedures that compromise the system’s functional integrity. Technical systems and ecological systems research have developed specific strategies to cope with corruption, such as early warning signals to avoid critical thresholds, redundancy measures to control local errors, and modular designs to prevent domino or spillover effects. These mechanisms offer useful models that can be applied in organizational contexts [85,87,88,89], thus reinforcing our central argument about corruption as operational deviation with implications for systemic integrity.

The research in this dimension demonstrates how technical–biological corruption operates across multiple scales: from individual data points to entire systems and from cellular processes to organism-level responses. The interconnections between clusters suggest that understanding corruption in technical and biological systems requires both preventive measures and resilience strategies. This dimension is unique in treating corruption not as a moral or social phenomenon but as a technical challenge requiring engineering and scientific solutions. The prevalence of technical solutions research indicates that while corruption may be inevitable in complex systems, it can be managed through proper design and robust methodologies.

The identification and characterization of these four distinct dimensions through LDA topic modeling provides an empirical foundation for the subsequent discussion. Having mapped the thematic landscape of corruption research, we now turn to a conceptual analysis of these findings, focusing on theoretical development across dimensions and specific limitations concerning non-state corruption.

4. Discussion: The Place of Non-State Corruption in the Literature

Building on the empirical differentiation revealed by topic modeling, we now examine how our results contribute to corruption studies in three ways: (a) by revealing the relative theoretical development across different dimensions of corruption research, (b) by identifying conceptual limitations in non-state corruption studies, and (c) by suggesting new theoretical directions for understanding corruption across institutional contexts. We argue that while state-centric frameworks have dominated corruption research, contemporary organizational complexity demands a more integrated concept of corruption.

4.1. Common Elements and Differences Across Dimensions

Our topic modeling analysis reveals four distinct dimensions with varying levels of research development: state corruption (55.28%), private-to-public corruption (20.55%), private-to-private corruption (11.39%), and technical–biological corruption (12.79%). This distribution not only confirms the predominance of state-centered approaches but also highlights important conceptual patterns. While state corruption research has developed robust and consistent conceptual frameworks for analyzing the phenomenon in modern society, and private-to-public corruption research shows similar conceptual consistency due to its relation to state corruption but fails to consider the private barriers to corruption, the research on corruption involving non-state actors remains underanalyzed (even less than technical–biological corruption) and often uncritically adopts terminology from state corruption studies.

State corruption research (Dimension 1) has developed a comprehensive conceptual framework, reflecting its dominance in corruption studies. It has identified a mechanism of corruption, namely, the use of public power for personal or group benefits, which refers to the more fundamental mechanism identified in this article, namely, the operational deviation from regular procedures, in this case involving the deviation from public administrative procedures to obtain particular gains. As Heidenheimer & Johnston [1] point out in their assessment at the beginning of the 21st century, this framework encompasses both public office approaches examining how officials misdirect their actions and the analysis of the rational decisions of agents to engage in corrupt practices. While public interest extends to any act seeking to place private interest over general interest and remains useful as a basis for private-to-public corruption, this conceptualization needs to be expanded, particularly considering the increasing complexity of public–private interactions in service delivery systems, concession-based services, and corporate authority over public matters.

Private-to-public corruption (Dimension 2) reveals complex patterns of interaction between corporate interests and state institutions. Probably one of the first works to assume that corruption also takes place in this sphere was Rose-Ackerman’s [90] seminal text. Rose-Ackerman devotes a special chapter to corruption in the private sector (e.g., companies, firms, and non-profit organizations), arguing that linking corruption so strongly to the government has ‘blurred’ the vision of academics who study this phenomenon, as the private sector can generate the same perverse incentives as the public sector. Rose-Ackermans’ analysis suggests that some particularities are still to be discovered regarding corruption in the private sector. For example, unlike the state corruption dimension, where private actors appear merely as corrupting agents, private-to-public corruption research examines the bilateral nature of corrupt exchanges, where the operational deviation appears simultaneously in both state and private organizations disturbing public and private procedures, ethic codes, or compliance agreements. Our analysis shows that this dimension attempts to transcend traditional state-centric approaches by examining how corruption emerges from the interaction between public and private spheres, rather than viewing it solely as public resource misappropriation.

Private-to-private corruption research (Dimension 3) remains notably underdeveloped in comparison to state and private-to-public corruption. As Argandoña [7] pointed out, corrupt practices that are a common preoccupation for governments should also be of business concern. While business corruption studies show a weak theoretical cohesion around corporate governance and firm performance (t_29, 2.64%), other topics remain largely descriptive or industry-specific. Interestingly, while these studies reference works like Jain’s [91] corruption models and Tanzi’s [2] cross-national studies, they do not make use of frameworks of canonical authors of state and private-to-public corruption such as Nye [92], Leff [93], Rose-Ackerman [94], La Porta [39], and Treisman [3]. This suggests that private-sector corruption is treated as a phenomenon distinct from public-sector corruption, thus revealing that something in private-to-private corruption is different compared to state corruption and cannot be captured by its framework. In this case, the problem does not lie in the violation of public–administrative regulations, but rather the operational deviation occurs in relation to the oversight of standards of regulatory agencies, ethical standards, or compliance norms developed by the very corporations involved in corrupt relations. In any case, the result is also the disappointment of expectations regarding the regular functioning of private agents. The motivation in this case may lie in personal financial gain, but it generally relates to institutional positioning in competitive environments, exploiting regulatory loopholes to reduce transaction costs, or direct involvement with illegal groups to gain or maintain control over a geographically strategic area with high returns. In such situations, private-to-private corruption also entails building a parallel architecture to conceal illicit activities, which may involve public agents such as police forces or supervisory agencies and also private actors such as the press or subcontractors who contribute to the profitability of a business. In other words, a corruption regime is constructed that operates in parallel to the public regime of legality and, in some cases, even replaces it in its capacity for territorial governance [95,96].

The technical–biological dimension (Dimension 4) conceptualizes corruption as system deterioration rather than a normative or moral transgression. This dimension examines corruption as a process by which systems—digital, physical, or biological—experience degradation and dysfunction. For example, in computer systems (t_48), corruption appears as deviations from normal operations that compromise data integrity and functionality. Similarly, in biological systems (t_35), corruption manifests as cellular dysfunction that disrupts organizational processes. In particular, the technical dimension of the analysis offers various mechanisms used in organizational analysis of complex systems to prevent critical situations such as cases of corruption. One such mechanism is redundancy, that is, the review of organizational procedures or actions by multiple independent actors simultaneously, allowing errors or deviations to be identified before they escalate across the organization as a whole [88,97]. Modularity, another type of preventive mechanism also applied in technical systems, has been proposed to prevent the spread of risky behaviors: the compartmentalization of operational areas can contain the generalization of corrupt practices throughout the organizational environment [98,99,100]. From the domain of biological corruption, the detection of early warning signs or weak signals constitutes another way of identifying risks (in corruption, overproduction, climate) in the management of modern organizations [89,101,102], which can anticipate critical transitions. This type of analysis originates in ecological and physical systems and has since expanded into other fields of research.

These parallel patterns of deterioration offer relevant insights to conceptualize corruption at an abstract level useful for all forms of corruption: corruption can be seen as a deviation from established procedures. Just as a corrupted computer or biological system may compromise their own integrity, corrupted organizations may appear functional while systematically deviating from legal or moral standards. This perspective broadens corruption theory across sectors by suggesting a general process of organizational dysfunction in public, private, or hybrid contexts, potentially bridging various dimensions of corruption research.

4.2. A New Conceptual Framework: Corruption as Operational Deviation

Our analysis identifies a fundamental commonality across all four dimensions that offers an abstract definition of corruption. We propose understanding corruption as “operational deviation”, that is, when an organization’s actual practices systematically depart from its established procedures, either legal, publicly informed, or custom-based but regularly known procedures. In other terms, an operational deviation occurs when a company violates its own regulatory frameworks or ethical codes, when a public official breaches administrative standards, when second-level agents trade information in exchange for money or favors, when companies collude to manipulate prices, or when private actors receive money in exchange for favors or future returns or benefits. This deviation results in what we call “disappointed procedural expectations”, which arise when relevant publics (citizens, stakeholders, customers, users) realize that the behavior according to the expected rules or procedures are not being applied in practice. The socially generalized expectation is that individuals and organizations behave in accordance with rules and procedures that are previously established and known by the relevant publics. When this does not happen, the expectation is disappointed and situations as those described above are perceived as a deviation from the regular operation of the organization or its actors. In this sense, corruption is an operational deviation from the procedures an organization or system has set to fulfill its intended function. Corruption is thus a disappointed procedural expectation [103].

This perspective broadens the conventional understanding of corruption, which has traditionally focused on the misuse of public office, public funds, or moral transgressions for personal or group benefits. Our comprehensive analysis suggests that, regardless of the institutional context—public, private, or technical–biological—corruption consistently manifests as a systematic departure from prescribed and accepted organizational procedures, creating risks and uncertainty for stakeholders and undermining institutional integrity. This reconceptualization allows us to view corruption not merely as a moral failing or abuse of power but as a broader pattern of organizational dysfunction that can arise in any environment.

This understanding becomes particularly relevant when examining organizations that hold significant regulatory or governance powers despite being non-governmental. Our topic modeling identifies several illuminating examples, particularly in international sports organizations (t_04). The cases of the International Olympic Committee (IOC) and FIFA provide compelling illustrations of how corruption can flourish in private organizations that hold governance powers over their respective domains. As documented by Simson & Jennings [104] and Mason et al. [105], the IOC’s interest-peddling networks with the World Anti-Doping Agency (WADA) demonstrate how Committee members can arbitrarily deviate from established procedures to benefit certain nations by not subjecting them to WADA’s regular protocols. The FIFA case goes even further, as it is identified as the central node of a corruption network [106] including politicians, businessmen, footballers, bankers and others, illustrating how operational deviations can create far-reaching networks of corruption that transcend traditional institutional boundaries.

Similar patterns emerge in non-profit organizations, where the mission orientation might suggest lower incentives for personal gain. Our topic modeling identifies a significant cluster of studies (t_26) examining corrupt practices in such organizations, particularly through the deviation of international aid funds. During the 1990s, a series of irregularities in UN departments came to public attention, including documented cases of corruption within UNICEF and UNDP [107]. These cases provide compelling evidence that operational deviations occur even in mission-driven organizations, suggesting that the potential for corruption exists wherever there are established procedures that can be subverted for private gain.

When these operational deviations occur, they create risks for third parties by violating their expectations about organizational functioning. We contend that violations affect public interest—not merely in moral terms (i.e., equity, transparency, equality), but also in procedural terms, as the expectation that institutions adhere to established procedures become disappointed.

A second key insight from our analysis is that corruption involves a more general mechanism than the use of public power for personal or group benefits as understood in state corruption research. Transnational private organizations, such as corporations and foundations, operate in a regulatory environment that national and international laws cannot cover fully. To fill these gaps, many private organizations voluntarily adhere to self-regulation standards, such as compliance norms. However, these standards lack the institutional enforcement that national or international legal treaties have [47,73,108]. The literature on corporate self-regulation highlights the challenges of credibility and compliance in private regimes [73,108], giving rise to “organizational self-deception”, i.e., the formal adoption of regulatory and ethical standards without effective implementation mechanisms. Business ethics studies indicate that this dissociation may reflect processes of “collective rationalization”—that is, the way in which individuals within an organization create and share justifications so that the gap between the written rules (what they say they carry out) and their actual behavior (violating those rules) appears acceptable or normal. Pragmatically, it involves finding collective justifications for not adhering to their own ethical or compliance standards [56,109,110]. As a result, the consequences of violating these self-imposed standards are often seen as less severe, mainly involving reputational or operational costs. This leads agents in the field to weigh these costs against the potential financial or influential gains from misconduct. In this vein, private corruption also consists of an operational deviation that disappoints procedural expectations. However, since these standards are self-imposed by the organizations themselves, private-to-private corruption involves an element of self-deceit that is absent in cases of state corruption or private-to-state corruption.

A third insight from our research is the practice of concealing traces of corruption, which is especially effective in private contexts where there is no independent authority to oversee procedural compliance. When corruption occurs, organizations no longer follow their stated procedures, but these deviations remain hidden until exposed as scandals. The FIFA case illustrates this point: the organization appeared to function normally for years until payment disputes and competing claims revealed systematic violations in World Cup bidding processes in 2015. What surfaced was not isolated misconduct but an entire shadow decision-making system operating parallel to FIFA’s official procedures [111,112,113]. Such scandals typically emerge when the corrupt network breaks down—when someone is denied payment or feels excluded or new competitors demand inclusion. At this point, operational deviations become visible, often revealing the depth of corruption within the organization.

This reconceptualization of corruption as operational deviation triggering disappointments of procedural expectations offers several advantages over traditional approaches. First, it provides a framework that can be applied across all institutional contexts, from state agencies to private corporations to technical systems. Second, it helps explain why corruption tends to emerge in clusters or networks rather than as isolated incidents, since operational deviations often require multiple participants to maintain the façade of normal operations while concealing the underlying violations. Finally, it suggests that preventing corruption requires attention not just to individual behavior or institutional incentives but to the integrity of organizational procedures and the mechanisms for detecting and responding to deviations from these procedures.

4.3. Implications for Non-State Corruption Research

While state-centric frameworks have dominated corruption research, they may also obscure unique aspects of non-state corruption that require distinct theoretical treatment. This is particularly crucial as the boundaries between public and private sectors become increasingly blurred and new forms of organizational authority emerge.

Private-to-public corruption research could benefit from greater attention to how operational deviations in hybrid spaces create distinct risks. The increasing complexity of public–private interactions, exemplified by public–private service delivery, concession-based services, and corporate authority over public matters, demands theoretical frameworks that can capture these unique dynamics. Our analysis shows that topics like regulatory enforcement (t_10) and public service delivery (t_25) involve complex operational expectations that do not fit neatly onto traditional state corruption frameworks. For instance, in healthcare delivery (t_25), corruption manifests not just as misuse of public resources but as systematic deviations from established protocols that put patient care at risk [71,114]. These hybrid spaces create unique opportunities for corruption that traditional state-centric frameworks may fail to capture adequately.

Private-to-private corruption research particularly needs theoretical development beyond state-centric models. While topics like business corruption (t_29) show some theoretical cohesion around corporate governance and firm performance, other areas remain undertheorized. Most corruption research and investigations (t_45) tend to overemphasize corporations as predators of state resources, reflecting the continued influence of state-centric frameworks [1,2,92,94]. However, this focus obscures important forms of corruption that occur purely within the private sphere. For instance, our analysis reveals cases where international private organizations’ funds are misappropriated by local private organizations, representing a form of private-to-private corruption that operates independently of state involvement. Similarly, corruption within and between private parties (t_29 and t_11) requires distinct theoretical treatment that current frameworks fail to provide.

The case of international sports organizations (t_04) offers a revealing example of how complex private-to-private corruption can work as a multilevel system. Organizations like FIFA and the IOC operate as global regulatory bodies, establishing hierarchical systems where corruption can flow both vertically (i.e., between international headquarters and national associations) and horizontally (i.e., between national associations or different sports bodies). This multilevel structure creates unique opportunities for corruption where international private funds can be misappropriated at various points in the organizational network, flowing through different jurisdictions and regulatory frameworks. These organizations interact with other private entities (e.g., sponsors, media companies, equipment manufacturers) across different jurisdictions, each with its own regulatory requirements and oversight mechanisms [105,106,115,116]. These cases demonstrate how private organizations can establish self-regulatory regimes parallel to state systems, creating opportunities for corruption that mirror but are distinct from state corruption. The emphasis on private-to-public corruption has led researchers to overlook these significant patterns of corruption between private entities operating at different scales and jurisdictions.

Moreover, our analysis reveals that corruption in non-profit and mission-driven organizations is undertheorized. Despite the normative assumption that mission orientation might reduce incentives for corruption, evidence suggests otherwise. The topic modeling identifies significant clusters (t_26) examining corrupt practices in such organizations, particularly through the deviation of international aid funds [117,118]. These cases highlight how traditional corruption frameworks, focused on profit motives and public office abuse, may miss important dynamics in organizations where power and influence, rather than direct financial gain, might be a primary motivation for corruption. The technical–biological dimension’s emphasis on system deterioration might offer valuable insights here, suggesting ways to understand corruption as organizational dysfunction regardless of the profit motive, and additionally proposing forms of dealing with corruption processes through mechanisms of redundancy, modularity, and early warning. These mechanisms, well known in complex technical systems and in the analysis of resilience in fragile socio-ecological environments (such as populated watersheds, coastal communities, populations near polluting industries, or forest plantations), can operate in conjunction. Modularity works by separating environments (e.g., firebreaks, tsunami containment barriers, the dispersion of concentrated industries), where redundancy in oversight instances serves to monitor the monitors. Meanwhile, the early detection of warning signals or weak signals contributes to the anticipation of potential future risks, for which the preparedness of the organization or community makes the difference in preventing the proliferation of critical events such as corruption, which, once unleashed, are difficult to contain and even harder to eradicate [88,89,119,120].

Our analysis suggests several key directions for future research. First, there is a need to develop conceptual frameworks that better capture how operational deviations and third-party risks manifest in non-state contexts, rather than simply applying state corruption models. Second, research should pay more attention to how corruption emerges and operates in hybrid organizational spaces where public and private interests intersect. Third, the technical–biological perspective of corruption as system deterioration provides valuable insights for developing integrated theoretical frameworks. Specifically, concepts such as entropy, resilience, self-organization, and self-regulation could help explain how corrupt networks emerge, persist, and eventually break down in organizational settings. Just as biological systems have evolved mechanisms to detect and repair cellular dysfunction, the preparedness of organizations and communities develop capabilities for identifying and correcting operational deviations before they become systemic. Finally, there is a need for more empirical research on corruption in private and non-profit organizations that do not rely on state-centric conceptual tools. These developments could help identify currently obscured forms of corruption and better understand how corruption functions across institutional boundaries in our increasingly complex transnational organizational landscape.

5. Conclusions

In this article, we have revealed patterns in the academic conceptualization of corruption, highlighting both advances and limitations in investigation of the phenomenon. Systematic analysis of the literature demonstrates that while sophisticated frameworks for understanding state corruption have been developed, increasing contemporary organizational complexity demands non-state-centered conceptual approaches. Corruption is an uneven field of research. The weak density in the study of non-state corruption indicates the need to develop more empirical research contributing to developing conceptual innovations to surpass the limitations in the field.

This is evident in the prevalence of regional studies in the corpus, which reveals two critical aspects: first, the conceptualization of corruption varies according to context, with each region developing its own approaches to the phenomenon, and second, these cultural variations impact how the boundaries between public and private are determined and understood. Europe places emphasis on the consequences for institutional legitimacy and anti-corruption reforms; Latin America on political corruption and the challenges to democratic legitimacy; East Asia on corruption practices culturally embedded in governance systems; post-Soviet countries on informality, predatory policing, and institutionalized corruption; and Africa on corruption practices integrated within cultural and historical frameworks. While this conceptual heterogeneity points to advances in the field, it also shows the need for theoretical frameworks that can integrate these perspectives.

The analysis reveals that private-to-public corruption, the second most studied dimension, tends to be analyzed through frameworks derived from state corruption, which translates better to this context than to private-to-private settings. However, even these adapted frameworks struggle to capture the complexity of modern institutional arrangements. Additionally, the study of private-private corruption represents a significantly smaller proportion of the corpus despite its growing importance in modern society. This is even more notable regarding non-governmental organizations and entities with strong governance powers, where corruption dynamics can differ substantially from traditional models. The technical–biological dimension, although less studied, offers valuable conceptual tools for understanding corruption as a phenomenon of systemic deterioration, suggesting new directions for future research, and offers techniques for dealing with corruption processes.

Based on these findings, we propose a reconceptualization of corruption beyond its traditional definition centered on the use public power for personal or group benefits. A common thread in states, businesses, and organizations is that corruption represents an operational deviation triggering disappointments of procedural expectations established for organizational functioning. This deviation generates a parallel set of operations that diverge from publicly recognized procedures while maintaining the appearance of normality through concealment and control of information. The perspective of corruption as systemic deterioration, evidenced in the technical–biological dimension, aligns with this understanding: corruption represents a deviation from the expected operations of the system that creates discrepancies between regular procedures and actual practices, potentially compromising the stability of the system.

Regarding the implications for policy and practice, our reconceptualization suggests that the concept of public interest should be understood not only in traditional terms—such as fairness, equity, transparency, and equality—but also as an expectation regarding how systems actually operate. In cases of state-related corruption, this expectation refers to the legal standards set by third parties, including legislators, politicians, and the government. In the case of private corruption, it involves violating self-created expectations, which results in a form of operation characterized by self-deceit. In both scenarios, this operational deviation leads to disappointment and has a negative impact on the public connected to these social milieus.

In the case of corruption in private organizations, the notion of public interest takes on a different meaning. It is not only about the interest safeguarded by the state but also about the interests of specific publics who depend on or are affected by the outputs of private organizations and who suffer the consequences of their corrupt actions. This is not limited to the local or national level but may extend to transnational publics, as in the case of corruption in football or sports in general or in foundations providing aid to highly vulnerable populations (refugees, migrants, chronically ill patients, children, women in oppressive regimes). Since it is highly complex for national courts to prosecute transnational crimes, regional legal policy agreements are a viable option for combating transnational corruption, for example, against organized crime networks and money laundering [121,122]. Recent developments toward the self-constitutionalization of social spheres (in trade, construction, finance, aviation, artificial intelligence, and medical research) may also build barriers that discourage corrupt acts or establish sanctions through transnational arbitral tribunals when corruption occurs [123,124,125]. Finally, compliance practices must focus not only on preventing acts of corruption but also on uncovering the hidden organizational deviations that are constructed to conceal misconduct. Just as innovative organizations separate their exploration units from the rest of the organization to provide them with creative freedom, compliance units in private organizations should be granted a high degree of operational autonomy from the organizational hierarchy so that they are not captured by internal interests or organizational problems and can implement corruption controls independently.

The methodological and theoretical implications of our findings are substantial for the development of the field. Topic modeling has proven to be a valuable tool for revealing patterns and limitations in the literature, especially when combined with deep interpretative reading. However, the limitations of our study, including the reliance on English-language texts and the non-inclusion of a substantial part of the literature on corruption found in working papers, books, and other formats not indexed in traditional databases, suggest the need to broaden the scope of future research. The development of more integrated theoretical frameworks that can address corruption in all its systemic complexity represents a crucial priority for the advancement of the field, particularly in the context of increasing interconnectedness between the public and private spheres.