The principal–agent theory, originating from economics (Jensen & Meckling, 1976; Mitnick, 2013; Ross, 1973), explains the relationship between principals (those who delegate authority) and agents (those who exercise that authority on behalf of the principals). In the public sector, citizens (principals) entrust policymakers and government officials (agents) with the authority to manage public resources and advance collective welfare. However, this delegation of power creates information asymmetry, where agents possess more information about budgetary processes and fiscal decisions than citizens. When information asymmetry is high, agents may exploit their positions for personal or political gain, leading to corruption. This behavior diverts resources from developmental priorities such as education, health, and infrastructure, thereby constraining human development (Méon & Sekkat, 2005). According to the “sand the wheels” hypothesis, corruption impedes development by increasing transaction costs, distorting policy choices, and discouraging investment.

Within this theoretical framework, government budget transparency acts as a mechanism that mitigates the agency problem. By disclosing budgetary plans, revenues, and expenditures to the public, transparency reduces information asymmetry and strengthens accountability (Alt et al., 2006). When citizens, civil society, and oversight institutions can access fiscal information, they are better able to monitor public officials and demand justifications for spending decisions. Consequently, transparency helps align the interests of agents with those of principals, discourages rent-seeking, and enhances public service delivery, which collectively improve human development outcomes. Thus, the principal–agent framework supports the first two hypotheses of this study: that corruption negatively affects human development and that government budget transparency moderates this relationship by reducing the agency problem.

The institutional theory complements the principal–agent framework by emphasizing the role of formal and informal institutions (rules, norms, and enforcement mechanisms) that shape social and economic behavior (North, 1993; Scott, 2008). Institutions define the “rules of the game,” setting incentives and constraints that determine whether public officials act in self-interest or in the public interest. In environments characterized by weak institutions, the lack of transparency and accountability fosters corruption and inefficiency. Conversely, strong institutions anchored on openness, public participation, and effective oversight promote integrity and reduce opportunities for rent-seeking.

Government budget transparency represents a formal institutional mechanism that enhances accountability by institutionalizing fiscal openness and enabling public scrutiny (Craig & Kopits, 1998). Transparency reforms such as open budget initiatives, participatory budgeting, and legislative oversight build trust and reinforce institutional credibility. These mechanisms improve resource allocation and efficiency, thereby contributing to human development. Integrating both theories, this study posits that while corruption undermines human development by distorting incentives and misallocating resources, government budget transparency mitigates these effects through strengthened institutional quality and reduced information asymmetry. Figure 1 illustrates this conceptual linkage between corruption, human development, and government budget transparency.

Corruption is a pervasive global problem. However, considering the impact of corruption on national development, there are two opposing views (efficiency-reducing and efficiency-enhancing view)/(“sand the wheels” or “grease the wheels” hypotheses). Advocates of the efficiency-reducing impact emphasize the fact that corruption hampers industry and commerce, consequently obstructing economic progress and leading to distortions in resource allocation (Tanzi & Davoodi, 1998). Conversely, advocates of the efficiency-enhancing impact assert that corruption facilitates company operations and commerce, promoting economic growth and investment. Consequently, corruption fosters efficiency and enhances economic performance (Huntington, 1968). Given the diverse perspectives, it may be argued that conflicts exist over the role of corruption in the literature. However, empirical evidence overwhelmingly indicates that countries with lower levels of corruption exhibit higher levels of human development.

For instance, Ortega et al. (2016) show that corruption increases the cost of healthcare and education delivery and reduces the quality of these services. Again, Gupta et al. (2002) report that corruption derails human development as it promotes income inequality and poverty, while Iskandar and Saragih (2018) suggest that corruption reduces available government resources to provide social services. Similarly, Mironov and Zhuravskaya (2022) report that corruption impedes access to quality education and undermines human capital development, while Ahmed and Anifowose (2023) report that corruption adversely impacts sustainable development. Further, Urbina and Rodríguez (2022), in a study of Latin American and Nordic countries, show that corruption has a more noticeable negative impact on human development in all countries from both regions. Furthermore, Fagbemi et al. (2022) explored the role of corruption in human development in Nigeria. They report that corruption is harmful to human development in Nigeria. They again show a bidirectional causality between corruption and human capital development. Consequently, the pervasive corruption issue in the country appears to have foreshadowed enduring inadequate human development.

Undoubtedly, there is overwhelming empirical support for the harmful impact of corruption on human development. However, some studies assert the non-linearity of the effects of corruption on economic activity (see Sharma & Mitra, 2019). Theoretically, the marginal effect of corruption may vary when the level of corruption changes. In other words, the harmful impact of corruption may be more pronounced at higher levels of corruption. For instance, Fankem and Ndzana (2025) report a non-linear effect of corruption on financial development in Africa, while Sharma (2021), in a global study, provides evidence that corruption is more financial development-reducing when its level is very high. Hence, we examine the following hypothesis:

Corruption presents a major obstacle to human development, impacting economic, political, and social systems. The empirics have largely supported the notion that corruption is detrimental to human development (see Ahmed & Anifowose, 2023; Gupta et al., 2002; Iskandar & Saragih, 2018; Ortega et al., 2016). However, in recent times, the role of government budget transparency in mitigating the adverse effects of corruption and fostering sustainable development has come to the fore. Budgets serve as the primary documents through which governments execute their governance program. Therefore, the effective implementation of government budgets has a direct impact on the general well-being of society (Fubbs, 1999). Consequently, the budget and the budgetary process must be clear, transparent, participatory, and credible, establishing a financial oversight system that necessitates high transparency to promote effective governance. Therefore, it is cogent to forebode that government budget transparency should play a critical role in curbing the negative impact of corruption on human development. Budget transparency ensures the proper allocation of resources, enhances good governance and accountability and builds public trust, all of which are essential for curbing corruption and promoting human development.

Empirically, Cuadrado-Ballesteros and Bisogno (2021), in a study of 110 countries across the globe, provide evidence that transparency fosters accountability, leading to optimal resource management, which, in turn, drives developmental progress. Similarly, Prijaković (2022) argues that budget transparency significantly reduces corruption, leading to better political and economic outcomes vital for sustainable development. Furthermore, in a cross-country analysis of 95 countries, Chen and Neshkova (2019) underscore that fiscal transparency strengthens public sector functionality and budget discipline. This helps enhance accountability and anti-corruption efforts, which are directly linked to improved development outcomes. Again, the World Bank (2020), in a study of EU countries, reports that enhanced budget transparency can facilitate improved governance and elevated human development by assuring the optimal use of public resources.

Furthermore, extant literature portends that the detrimental impact of corruption on national development is dependent on certain country-specific indicators. For instance, Ahmed and Anifowose (2023), in a study of 42 African countries, provide evidence that the impact of corruption on sustainable development goals depends on the extent of corporate governance structures in a country. They further reported that the positive impact of corporate governance on sustainable development is stronger in countries with high corruption levels. Again, Fankem and Ndzana (2025) investigated 32 African countries and intimated that the effect of corruption on financial development depends on the quality of institutions. Undoubtedly, there is overwhelming empirical support for the role of budget transparency in mitigating corruption. Again, the existing literature is replete with studies on how country-specific factors determine how corruption impacts national development. However, the extant empirical literature has ignored a lacuna in how budget transparency moderates the relationship between corruption and human development.

Additionally, Lindstedt and Naurin (2010) argue that transparency must reach a critical level to be effective. Weak transparency (e.g., superficial data releases) may not curb elite capture. Transparency in itself is not enough. This suggests that the impact of government budget transparency on the effect of corruption on human development may be non-monotonic. When government budget transparency is weak, it may not have any significant influence on how corruption hurts human development. Hence, we examine the following hypotheses:

To test our hypotheses, the study employed panel data from 119 countries across the globe spanning the years 2010–2021. The Open Budget Index is available only from 2010 and for the countries in the sample. The choice of countries for this study was based on data availability. Panel data are employed because they allow the exploration of both cross-sectional and time-series variations. Panel data control for omitted variables and also help account for both long-run and short-run effects. (Wooldridge, 2012). The data were taken from the United Nations Development Programme (UNDP), Transparency International, Open Budget Survey (OBS)—International Budget Partnership, and World Development Indicators (WDI).

Our dependent variable, the Human Development Index (HDI), measures Human Development. The HDI is a geometric mean of three dimensions: a long and healthy life, knowledge (education), and a decent standard of living. The HDI is sourced from the UNDP. It is calculated on a scale from 0 to 1, with higher values indicating greater levels of human development. This paper will present HDI values on a scale from 0 to 100 for clarity in analysis. The HDI is widely recognized as the standard metric for assessing human development due to its quantitative nature and extensive use in research.

Corruption was measured using the Corruption Perception Index (CPI) from Transparency International. The index ranges from zero (0) to a hundred (100), where 100 is very clean and 0 is a highly corrupt country. However, for easy interpretation, we rescale the CPI using the formula –1* (CPI – 100) proposed by Agoba et al. (2019), where higher values denote corrupt countries while lower values denote clean countries. This index, which relies on expert assessments and opinion surveys, aggregates various data sources to provide a composite score indicating the extent of corruption (Olken, 2009).

Government budget transparency is an index created from a questionnaire of 122 questions about the availability, quality, and timeliness of budget documents. It is measured using three primary components: the public availability of budget information (transparency), opportunities for public participation in the budgeting process (public participation), and the effectiveness of formal oversight institutions, including legislative bodies and supreme audit institutions (strength of audit institutions) (International Budget Partnership, n.d.). The index assigns a score ranging from 0 (least transparent) to 100 (most transparent) based on the amount and punctuality of budget information that governments make publicly available.

We control for economic growth, infrastructure, inflation, employment levels, household savings, and trade openness. First, economic growth increases government revenues for essential services like health and education, enhancing human development (Barro, 2013). Again, increased economic output leads to employment generation, which is crucial in improving human development indicators. We measure economic growth by GDP per capita growth (annual %). Further, infrastructure development is another critical factor influencing human development. Calderón and Servén (2010) found that infrastructure improvements, such as better roads, electricity, and water supply, significantly enhance human development by improving access to essential services and economic opportunities. We measure infrastructure as the number of telephone lines plus mobile subscriptions per 100 people. Again, inflation is detrimental to human development as it affects the purchasing power of people, increases the cost of living, and promotes instability in the financial sector, among other things (Kumar & Jie, 2023). We use the consumer price index to measure inflation.

Furthermore, Employment levels directly influence human development by providing income and improving living standards. Studies by Dagohoy et al. (2025) suggest that higher employment rates are associated with better health and educational outcomes, contributing to higher HDI scores. We measure employment as a ratio of the employed to the working population. Again, household savings, reflecting financial stability and future security, also play a role in human development. Ribaj and Mexhuani (2021) emphasized the importance of savings in economic development as they contribute to increased investment, which accelerates economic growth. It is measured using gross savings as a ratio of GDP. Finally, access to international markets could promote human development by attracting foreign direct investment, technological transfers and knowledge, and increasing access to goods and services (Radmehr et al., 2022). Trade openness is measured as exports plus imports scaled by GDP. Further, we present the variable descriptions, measurement of variables and the sources of data in Table 1.

In this paper, we posit that corruption impacts human development across the globe. Therefore, we specify the empirical model as follows: HDit=β0+β1CORit+β2GBTit+β3(GBT*COR)it+Zit+μi+vt+εit(1)

Where subscripts i and t refer to country and time, respectively. HD_it represents Human Development. GBT_it represent Government Budget Transparency. COR_it represents the extent of corruption in country. Additionally, (GBT * COR)_it denotes the interaction between indicators of government budget transparency and corruption. Nonetheless, we calculate the net effect of corruption on human development by following the works of Baltagi et al. (2009) and Ofoeda et al. (2022). In light of this, we partially differentiate equation (1). The following is the resultant equation: ∂HDit∂CORit=β1+β3*GBTit(2)

Additionally, we include several control variables in our model, which is consistent with extant literature. Z_it is a vector of control variables such as economic growth, infrastructure, inflation, employment, household savings, and trade openness.

In this study, we also posit that the impact of corruption on human development is non-monotonic. That is, corruption has a non-linear effect on human development, with a more detrimental effect for higher levels of corruption. We further hypothesize that a more transparent government budget should mitigate the negative impact of corruption on human development. We employ the Hansen (2000) threshold regression estimation technique to explore the non-linear behavior of our hypothesized relationships. The conventional method for exploring the threshold effect is by introducing a quadratic term in the model (see Ibrahim & Alagidede, 2018). However, this technique is unable to identify the exact point where the relationship changes direction. Again, this technique may be exposed to the problem of multicollinearity, and also, the functional form of the non-linearity is imposed on the data. However, the Hansen (2000) panel threshold estimation technique is able to trace the turning point or the threshold for policy decisions and is also able to automatically correct multicollinearity problems in the dataset. The threshold regression model can be described as follows: HDi,t=φXi,t+{αi+β1CORi,t+θ1GBTi,t+μi,t GBTi,t<γαi+β2CORi,t+θ2GBTi,t+μi,t GBTi,t≥γ}(3) where subscripts i and t refer to country and time, respectively. HD_it represents human development, while COR_it is used for Corruption. GBT_it represents Government Budget Transparency. α_i denotes the country-specific fixed effects, while μ_it is a zero mean, finite variance, i.i.d. disturbance. X_it is the vector of variables hypothesized to affect human development. GBT_it is the threshold variable used to split the sample into groups, γ is the threshold value, while β₁ and θ₁ are the threshold coefficients when the threshold value is lower than γ, whereas β₂ and θ₂ are the threshold coefficients when the threshold value is higher than γ.

The study investigates the possibility of heteroskedasticity, autocorrelation/serial correlation, multicollinearity, and endogeneity in our dataset. This is because the presence of any of these problems in our datasets may lead to biased estimates and, therefore, incorrect conclusions drawn from the result of the study. The study employed the Breusch–Pagan/Cook–Weisberg and White tests to detect heteroskedasticity in our data. The presence of heteroskedasticity may lead to inefficient estimates, biased standard errors and, therefore, misleading conclusions. The results demonstrate the presence of heteroskedasticity in the dataset, as shown in Appendix B. Again, the study tested for autocorrelation by employing the Wooldridge test (Appendix C). The results show there is evidence of autocorrelation in our data. Furthermore, due to the possibility of endogeneity in models of this nature, we assessed the presence of endogeneity in our dataset using the Du-Wu Hausman test (Appendix D). However, our dataset has no evidence of endogeneity. As a result of the above tests, the study used the Prais–Winsten estimation technique, which corrects for heteroskedasticity and autocorrelation in the dataset.

Further, because Prais-Winsten is a fixed effect model, the study used a Hausman test (Appendix A) to test whether the model is a fixed effect or random effect model. The results show that our model is a fixed effect model, making Prais–Winsten an appropriate estimation technique for our model. Finally, because multicollinearity can result in skewed coefficients, the study tested for multicollinearity using the variance inflation factor (Appendix E) and the correlation matrix (Table 2). The results show that there is no multicollinearity problem in our dataset. This shows that the independent variables are not substantially correlated.

In this study, we employed the Prais–Winsten panels-corrected standard errors (PCSEs) (Prais & Winsten, 1954) in estimating the regression model. Unlike the fixed or random effect models, the Prais–Winsten estimation technique gives more efficient estimates when serial correlation and heteroskedasticity are present in the data. Alternatively, the feasible generalized least squares (FGLS) could produce similar results. However, according to Beck and Katz (1995), the Prais and Winsten (1954) method is preferred over the FGLS because the FGLS produces incorrect standard errors and autoregressive errors with shorter time dimensions. Therefore, with our dataset’s shorter time dimension (2010–2021), we employ the Prais–Winsten technique, which is better suited for datasets with shorter time dimensions. Further, unlike the Cochrane–Orcutt method, which loses the first observation in its iterative approach, the Prais–Winsten technique retains all observations, making it better suited for time-dependent data. Again, compared to the Cochrane–Orcutt method, the Prais–Winsten method corrects serial correlation closer to 1 and generates more reliable and efficient results when correcting heteroskedasticity.

In addition, we hypothesized that corruption adversely affects human development worldwide. Tables 4 and 5 show the Prais–Winsten regression results for the entire sample (models 1–3), developed (models 4–6), developing (models 7–9), and African countries (models 10–12). The study reveals a negative relationship between the CPI and human development across all models (models 1–12) and economies.

We also find that transparency, public participation, and oversight positively influence human development. Transparency significantly enhances human development in all economies, while public participation and oversight have stronger effects in developed countries. In developing and African contexts, their effects are weaker due to limited institutional capacity and elite capture. These results align with Cuadrado-Ballesteros and Bisogno (2021), who argue that transparency strengthens accountability and resource efficiency.

Further, we also explore government budget transparency’s influence on human development, hypothesizing a positive relationship. We explore the impact of the indicators of government budget transparency (transparency, public participation, and oversight) on human development across different economies. The results reported in models 1, 4, 7, and 10 from Tables 4 and 5 reveal that transparency positively impacts human development across economies. The study further reveals that public participation significantly impacts human development for our full sample and developed countries (models 2 and 5). Additionally, the results demonstrated that oversight institutions play a significant role in the budget process, positively impacting human development. Models 3, 6, 9, and 12 from Tables 4 and 5 present the results for full, developed, developing, and African country samples.

Further, we hypothesize that budget transparency dampens the adverse effect of corruption on human development. Therefore, we interact corruption (CPI) with the three dimensions of government budget transparency (transparency, public participation, and oversight institutions). We observe that the interactive variable and the total effect are positive and significant across all samples for transparency (models 1, 4, 7, and 10) while interacting public participation with corruption produced a positive coefficient for the interactive variable and the total effect for only our full sample and developed countries (models 2 and 5). Again, we observe that oversight institutions also moderate the negative impact of corruption for our full sample, developed and developing countries.

Given that this study aims to establish the non-linear impact of corruption on human development and the non-linear behavior of the role of government budget transparency on the connection between corruption and human development, we first establish whether or not the relationship is monotonic. Therefore, we hypothesize a threshold effect (i.e., nonlinear relationship) between corruption and human development as well as a threshold effect in the impact of government budget transparency on the corruption-human development nexus. Accordingly, we test the null hypothesis of linearity where θ₁ = θ₂ against the alternative hypothesis of a threshold model, t theta ends subscript sub 1 not equal theta sub 2. We employ 2000 bootstrap replications and a 15% trimming percentage. We test the threshold effect for each of the proxies of government budget transparency (transparency, public participation and oversight). In Tables 6 and 7, we present the results on the existence of the threshold test

As shown in Tables 6 and 7 above, the bootstrap p-values for the test of the existence of threshold effect for the relationship between corruption and human development and the impact of government budget transparency on the corruption-human development connection for the three proxies of government budget transparency (transparency, public participation and oversight) for only the full sample. The results presented show the existence of a threshold effect in the corruption-human development connection and the existence of a threshold in the impact of government transparency on the corruption-human development nexus across all proxies of government budget transparency. This is because the p-values are far less than 0.01, and the Lagrange Multiplier (LM)-test statistic is quite large. We, therefore, conclude that there is a threshold effect in the Anti-Money Laundering (AML) regulations-growth connection and a threshold effect in the influence of AML regulations on the Foreign Direct Investment (FDI)-growth relationship. Our findings, therefore, suggest that the level of corruption in a country determines its impact on human development. Likewise, the effect of corruption on human development is determined by the extent of government budget transparency. Consequently, we divide the sample into two regimes: one regime is above the threshold value, and the other is below the threshold value.

Having established the existence of the threshold effect in the impact of corruption on human development and the influence of government budget transparency on the corruption-human development link, we proceed with the panel threshold regression. We present the results of our hypothesized relationships in Tables 8 and 9 from models 13 to 30. For each of our government budget transparency proxies, we present the global ordinary least squares regression (OLS), regime below the threshold value (low), and regime above the threshold value (high). We find a threshold value of 89 across all proxies of government budget transparency (transparency, public participation and oversight). Further, we find that corruption hurts human development across all regimes and indicators of government budget transparency (transparency, public participation and oversight). However, the negative impact of corruption on human development is more pronounced at higher corruption regimes for all government budget transparency indicators. For instance, the study reported a coefficient of −0.384 below the corruption threshold and −2.268 above the corruption threshold.

Further, we also examine the influence of government budget transparency on the corruption-human development link. We use the three proxies of government budget transparency (transparency, public participation and oversight) as regime-switching triggers. We run our results for only the full sample. We find threshold values of 9, 22 and 46 for transparency, public participation and oversight, respectively. The study finds that corruption has a negative impact on human development below the threshold across all indicators of government budget transparency. However, it has a positive insignificant effect on human development above the threshold of government budget transparency.

The study emphasizes the importance of budget transparency, public participation, and oversight in enhancing human development globally, particularly in regions with governance challenges. Policy recommendations include strengthening institutional frameworks across countries, enhancing budget transparency, promoting public participation, and reinforcing oversight mechanisms to effectively combat corruption and foster sustainable human development. Developed countries are urged to sustain efforts to maintain high levels of transparency, accountability, and civic engagement, ensuring that governance frameworks remain resilient and responsive to evolving needs and challenges. Additionally, for African and developing countries with weaker institutional capacities, efforts should focus on capacity-building initiatives, strengthening governance frameworks, and promoting civic education and empowerment. Specifically, governments should improve information accessibility, strengthen civic engagement through public awareness initiatives, build trust through enhanced accountability mechanisms, and reinforce institutional independence of institutions, among others.