Introduction

Macroeconomic uncertainty, energy poverty, and climate exposures are some major obstacles developing countries are facing towards achieving sustainable development (SD) (UNCTAD, 2022). Public debt (PD), the amount of money borrowed by governments from different sources, allows these countries to fund major development projects designed to boost economic growth, ensure social welfare, and climate resilience (Yusuf and Mohd, 2023). It is an important element that can have a big impact on a nation’s ability to develop sustainably. Strategic debt financing is directed toward policies that promote growth and reforms that enhance efficiency can contribute to long-term sustainable development. However, the soaring indebtedness is another concern for emerging nations that may impede their progress toward sustainability. Excessive levels of public debt may result in higher interest rates and budgetary costs, which could take funds away from social and productive expenditures that are crucial for sustainable development.

The PD in developing countries (excluding China) hit $11.5 trillion in 2021 and is escalating continually (Grynspan, 2023). Heightening costs of borrowing, inflation, depreciating currencies, and depleting reserves are intensifying the vulnerabilities further and making it even more difficult to repay this obligation. Along with this mounting financial stress, developing countries become trapped in a vicious loop. Moreover, the spiralling debt is impeding economic progress and jeopardizing developing countries’ chances of SD, which eventually pushes nations like Ghana, Pakistan and Sri Lanka into the debt crisis (Grynspan, 2023).

Economic theory also implies that developing nations, with investment potential and constrained capital during the early stages of development, may boost their economies by prudent borrowing; though, the ‘debt overhang’, i.e., excessive PD accompanied by macroeconomic uncertainty and adverse shocks, jeopardizes the transition by triggering default and demotivating investment (Pattillo et al., 2002).

However, to restrain these externalities and gear PD towards SD, governance can play a vital role. Weak governance entices over borrowing, economic inefficiency and underdevelopment (Tarek and Ahmed, 2017; Ngobo and Fouda, 2012; Ibrahim, 2020). Scholars theorize that governance leverages the attainment of development goals (North, 1990) by influencing economic and social outcomes (Adams, 1990). Governance prioritizes collective interest and curtails corruption (Stiglitz, 2005), which will help in managing PD efficiently (Jalles, 2011) and promote SD. Giavazzi and Pagano (1990) argue that effective governance reforms and fiscal retrenchment can enhance public debt sustainability by restoring market confidence and reducing debt burdens.

Despite this theoretical predicament and pragmatic relevance of Public Debt and governance in attaining Sustainable development, to date, no research has systematically investigated the causal relationship among them. While existing studies have examined the individual or paired associations between these variables, an integrated analysis that establishes the direction and magnitude of their interdependencies remains absent from the literature. Understanding this causality is crucial, as governance mechanisms can mediate the effects of public debt on sustainable development outcomes, influencing economic stability, social equity, and environmental sustainability. The lack of empirical evidence on this nexus highlights a significant gap in the academic discourse, warranting further investigation using robust methodological frameworks.

Considering these substantial gaps in the literature, this investigation intends to investigate the causality between PD, governance, and SD for developing countries using Generalized Method of Moments-Panel Vector Auto-regression (GMM-PVAR) and System Generalized Method of Moments (SGMM) model.

The contribution of this paper is manifold. First, it addresses a critical gap in empirical research by analyzing the causality between public debt and sustainable development using a comprehensive, multidimensional indicator of the latter. Previous studies have only examined public debt’s impact on isolated dimensions of Sustainable development (e.g., economic growth, social welfare, or environmental quality), despite Sustainable development being an inherently multidimensional concept (Dutta & Saha, 2023a, 2023b). Notably, research employing a holistic Sustainable development measure in the PD-SD nexus is rare. Even studies claiming to explore this relationship—such as Chien et al. (2022) and Georgescu (2014)—rely on GDP as a proxy for SD, despite its well-documented limitations. GDP fails to account for welfare disparities, social equity, and environmental degradation, rendering it an inadequate measure of true sustainable development.

Second, this study incorporates governance quality into the PD-SD framework—a dimension previously overlooked in the literature. To the best of our knowledge, no prior research has empirically examined the interactive effects of public debt, governance, and Sustainable development, particularly in the context of developing countries. This focus is especially policy-relevant, as many developing nations face high indebtedness alongside rapid but uneven economic growth, all while striving to meet Sustainable development targets.

Third, we advance the literature by investigating the nonlinear dynamics between public debt and Sustainable development, identifying the critical debt threshold beyond which public debt’s impact on Sustainable development shifts significantly. This provides policymakers with a clear benchmark for debt sustainability.

Fourth, methodologically, we employ advanced econometric techniques—including GMM-PVAR and SGMM—to rigorously assess causal relationships among the variables of interest. Moving beyond conventional estimators, these approaches account for endogeneity, cross-sectional dependence, and dynamic interactions, offering more robust empirical insights.

The rest of our manuscript is arranged as follows. Section 2 includes conceptual framework and review of relevant literature, Section 3 depicts the data and methodology used in this research, Section 4 explains the findings, and Section 5 concludes.

Literature review

Conceptual and theoretical framework

This study builds upon foundational economic theories and institutional perspectives to examine the complex, multidimensional relationships between public debt (PD), governance (GOV), and sustainable development (SD) (Fig. 1). Sustainable development—encompassing economic, social, and environmental dimensions—relies on effective institutional frameworks. Simultaneously, sustainable development can reinforce governance structures, indicating a bidirectional relationship. Strong institutions foster sustainability, while progress in sustainable development can enhance institutional capacity and policy coherence.

Fig. 1: Conceptual framework.
Fig. 1: Conceptual framework.
Full size image

The relationship among public debt (PD), governance (GOV), and sustainable development (SD).

The proposed framework also posits that public debt can either support or hinder sustainable development, depending on the quality of governance. Effective governance enables the productive use of public debt by promoting transparency, accountability, and strategic investment—particularly in critical sectors such as education, healthcare, and environmental protection. Conversely, poor governance may lead to fiscal inefficiencies, corruption, and adverse developmental outcomes due to the misallocation of debt resources.

This conceptual model is grounded in the following theoretical foundations:

Debt overhang theory

This theory posits that excessive public debt discourages private investment and structural reform, thereby impeding economic growth and sustainable development. In situations where debt becomes unsustainable, expectations of future fiscal instability or tax increases deter private investment, especially in key sectors such as green infrastructure, healthcare, and education. This effect is particularly severe in developing economies, where institutional weaknesses reduce the effectiveness of borrowed funds. The theory also suggests a nonlinear relationship: the impact of debt shifts from positive to negative beyond a certain threshold—a hypothesis this study tests using two advanced regression models.

Crowding-out effects

Public borrowing can also trigger crowding-out effects, where government demand for financial resources competes with the private sector. Elevated public debt levels may increase interest rates, restricting private access to credit—especially for sustainability-focused investments like renewable energy or sustainable agriculture. However, the extent of crowding-out depends largely on the allocation of borrowed funds. For example, infrastructure projects may stimulate short-term growth but undermine long-term sustainability if they prioritize immediate economic returns over environmental and social considerations. This tension highlights the critical role of governance in ensuring that public debt complements, rather than displaces, sustainable development goals—a dynamic explicitly explored in this study.

The role of governance

Governance quality is a decisive factor in shaping how public debt impacts sustainable development. Transparent institutions, efficient public administration, and sound fiscal management amplify the benefits of debt-financed investments. In contrast, weak governance—characterized by corruption, mismanagement, and lack of accountability—can lead to debt misallocation, fiscal crises, and unsustainable growth, thereby exacerbating inequality and environmental degradation.

Importantly, these relationships are not unidirectional. Sustainable development outcomes can, in turn, influence governance by enhancing institutional capacity, regulatory effectiveness, and policy focus. Reforms aimed at effective governance can improve debt sustainability by aligning borrowing with long-term development goals. Conversely, poorly managed debt can erode institutional credibility, weaken governance structures, and heighten financial vulnerabilities.

Theoretical and empirical implications

These theoretical perspectives underscore the inherent trade-offs in debt-financed growth. While public borrowing has the potential to support sustainable development—particularly in areas like climate resilience, education, and health—its effectiveness depends on both the level of indebtedness and the efficiency of resource deployment. These challenges are particularly pronounced in developing economies, where borrowing costs are high and development needs are urgent.

This study contributes to the theoretical discourse by adopting a multidimensional framework for sustainable development that incorporates economic, social, and environmental indicators—moving beyond conventional GDP-based evaluations. This comprehensive approach allows for a more nuanced analysis of the trade-offs embedded in public debt dynamics.

The proposed conceptual model illustrates the interdependent relationship among public debt, governance quality, and sustainable development outcomes. Within this framework, governance acts as a pivotal mediating mechanism—capable of transforming public debt into a catalyst for sustainable progress rather than a source of instability, inequality, or environmental harm.

To empirically validate these relationships, the study employs a wide array of indicators and applies robust econometric methodologies, ensuring the rigorous testing of its core hypotheses.

Empirical literature

Public debt and economic growth

Economic growth, social aspects, and environmental aspects are the three dimensions of sustainable development (ESCAP, 2015). The role of PD in influencing economic growth has been extensively studied across different contexts, including developing countries (Presbitero, 2012), advanced economies (Panizza and Presbitero, 2013), and OECD countries (Panizza and Presbitero, 2014). However, the relationship between PD and growth remains obscure and can be positive, negative, or nonlinear (Rahman et al., 2019). This inconsistency highlights a significant gap in understanding the nuanced effects of PD on economic growth, especially within the framework of sustainable development.

Several studies confirm a negative impact of PD on economic growth, particularly in South Asia (Mohsin et al., 2021) and other developing countries (Megersa and Cassimon, 2015; Ali and Mustafa, 2012; Akram, 2011). Conversely, some research indicates a positive impact of PD on growth, such as in Malaysia (Bakar and Hassan, 2008) and South Asia (Tung, 2022). Additionally, PD and economic growth are interdependent (Abdelkafi, 2018), with evidence of bidirectional Granger causality in South Africa (Ncanywa and Masoga, 2018) and Ghana (Owusu-Nantwi and Erickson, 2016). This relationship is often non-linear in developing countries (Pattillo et al., 2002).

Governance plays a vital role in guiding PD to accelerate growth. Higher transparency in PD enhances growth (Kim et al., 2017), and good governance reduces the negative effects of PD on growth (Abbas et al., 2021). In countries where PD negatively affects growth, improving governance can reverse this effect (Manasseh et al., 2022). However, many existing studies focus on aggregate governance indicators without sufficiently disentangling which specific governance mechanisms (e.g., corruption control, regulatory quality, political stability) are most effective in mitigating PD-related growth constraints. This study addresses this gap by investigating the differential roles of governance components in shaping the PD-growth relationship.

Public debt and social aspects

PD and social aspects are intricately linked. Research has shown that PD is a significant determinant of income inequality (Salti, 2015; Carrera and de la Vega, 2021; Miyashita, 2023). Even a 1% increase in PD can lead to a 0.17% increase in inequality, with the effect doubling when the PD ratio reaches 57.47% (Andoh et al., 2023). In developing countries, PD exhibits a U-shaped relationship with inequality, initially reducing inequality but increasing it at higher levels (Kilinc and Kilinc, 2023). High PD levels also increase inequality in OECD countries (Arawatari and Ono, 2017), and both external and internal debt contribute to inequality (Obiero and Topuz, 2021). Moreover, inequality itself can lead to increased debt levels (Jabłoński et al., 2015).

PD has broader social consequences, inhibiting human development (Wang et al., 2021; Sadiq et al., 2022) and decreasing the standard of living (Arshed et al., 2022). It reduces social expenditure (Lora and Olivera, 2007) and creates social anxiety over potential tax burdens (Zhang, 2006). Moreover, PD negatively affects public education expenditure, which is crucial for human capital development (Ono and Uchida, 2018).

While governance is acknowledged as pivotal in shaping PD’s social consequences (Ali and Yahya, 2019; Sadiq et al., 2022), existing studies often overlook the role of governance quality in moderating PD’s effects on inequality and social well-being.

Public debt and environmental aspects

The relationship between PD and environmental aspects is also ambiguous. Some studies suggest that PD decreases renewable energy consumption (Hashemizadeh et al., 2021) and that governments should reduce PD to improve environmental quality in the long run (Fodha and Seegmuller, 2014). Auteri et al. (2024) also argue that rising government debt can restrict investments in renewable energy infrastructure, whereas expanding renewable energy positively influences government debt trends.

On the other hand, Raouf (2022) demonstrated that public debt exerts a positive influence on renewable energy consumption in OECD countries. This finding aligns with a growing body of research affirming that renewable energy consumption is a critical driver of environmental sustainability (Ximei et al., 2025; Li et al., 2024; Jiatong et al., 2023; Sibt-e-Ali et al., 2024). In China, local government debt promotes urban emission reduction (Qi et al., 2022), while globally, PD can help achieve environmental targets (Clootens, 2017). Given these conflicting findings, Sadiq et al. (2022) suggest that governments should invest PD in green energy and cleaner production to meet environmental targets.

However, the governance dimension remains underexplored in this literature. While some studies hint at governance’s role in ensuring that PD is channelled toward sustainable environmental projects, comprehensive empirical analyses on this governance-debt-environment nexus are scarce. This study contributes by analyzing how governance influences the effectiveness of PD in achieving environmental sustainability goals.

Identifying the research gap and contribution

The existing literature provides valuable insights into the relationship between PD and different aspect of sustainable development. However, several gaps remain. While numerous studies discuss the economic, social, and environmental implications of PD, few explicitly examine in respect of comprehensive sustainable development and role of governance as a mediating factor. Furthermore, the specific governance mechanisms that shape PD’s impact on sustainable development remain insufficiently explored. Therefore, by integrating holistic measure of sustainable development and governance into the PD-SD framework, this study provides novel insights into how public debt can be managed more effectively to foster sustainable development.

Data and methods

Data

To investigate the PD, governance and SD relationship, we use unbalanced panel data of 35 countries over 1991–2020. Data constraints limit our sample selection. We have chosen the middle- and lower-income countries and timeframe of the studies based on data availability. The list of countries chosen is included in Appendix-3, Table A4. Other factors studied in this study, namely GDP per capita (GDPC), Consumer Price Index (CPI) and Trade openness (TO), are inextricably linked to Sustainable Development, governance and Public debt. Data of PD, SD, GDPC, CPI and TO are sourced from World Development Indicators (World Bank, 2023a) and data of six governance indicators are collected from World Governance Indicators (World Bank, 2023b). Variable descriptions and data sources are given in Appendix-1, Table A1.

Sustainable development

To assess sustainable development (SD), we employ Adjusted Net Saving (ANS) as a percentage of Gross National Income (GNI). ANS is computed by adjusting Gross National Savings (GNS) through multiple stages: first, by accounting for capital depreciation, depletion of natural resources, and environmental degradation; second, by incorporating public expenditure on education; and finally, by normalizing the resultant figure against GNI (United Nations, 2007). The adjustment process involves deriving Net National Savings from GNS, factoring in resource depletion—such as the exhaustion of oil, minerals, and forestry—integrating the economic impact of CO₂ emissions, and incorporating investments in human capital accumulation (United Nations, 2007).

Public debt

Public debt, typically measured as a percentage of Gross Domestic Product (GDP), reflects a country’s accumulated government borrowing and its potential implications for economic performance.

Governance

Governance is represented through the Worldwide Governance Indicators (WGI) framework developed by Kaufmann et al. (1999). This dataset consolidates extensive national governance information and classifies it into six distinct dimensions using an unobserved components model, which are: (i) voice and accountability (VA), (ii) political stability and absence of violence/terrorism (PS), (iii) regulatory quality (REQ), (iv) government effectiveness (GE), (v) rule of law (RL), and (vi) control of corruption (CC). However, given concerns related to multicollinearity and over-specification when incorporating these indicators separately into a single estimation model, we employ Principal Component Analysis (PCA) to synthesize them into a composite governance index (GOV) (Appendix-2). Furthermore, to assess which dimensions of governance matters more, we also use the six disaggregated governance indicators to supplement our analyses.

Additional variables

To capture variations in economic growth and broader macroeconomic conditions across countries, we incorporate the GDP growth rate (gGDP) and the consumer price index (CPI) (Samargandi et al., 2015, Saha and Dutta, 2021). Furthermore, to account for differences in international economic integration and openness across economies, we include trade openness (TO) as an additional explanatory variable (Samargandi et al., 2015).

Model

Generalized Method of Moments-Panel Vector Auto-regression

We examine the PD, governance and SD causality using GMM-PVAR because it is suitable for capturing unobserved heterogeneity and time-invariant observation by employing fixed effects, examining endogenous interactions, dynamic links and possible direction of causality among variables and deriving the model specification, when the theoretical links among variables are not adequately defined (Abrigo and Love, 2016).

For conducting the GMM-PVAR estimation, first we check the unit-root of our selected variables. Then, we determine the necessary lag order of the variables to build the following model by using Schwarz criterion (SC), Akaike information criterion (AIC) and Hannan and Quinn information criterion (HQIC).

$$\begin{array}{c}{Y}_{it}={A}_{0i}+A(\ell ){Y}_{it-1}+{\mu }_{it}\\ i\in \{1,2,\ldots ,\,N\};\,t\in \{1,2,\ldots ,\,{T}_{i}\}\end{array}$$
(1)

where \({Y}_{{it}}\) is a vector of k endogenous variables for i country and t time and can be summarized as \({Y}_{{it}}=\left[{{D}_{{it}},{{GOV}}_{{it}},{{GDPC}}_{{it}},{{CPI}}_{{it}},{{TO}}_{{it}},{SD}}_{{it}},\right]\). Here, D represents public debt (PD), measured as central government debt. \({{GOV}}_{{it}}\) is a composite index of six country governance indicators, aggregated using the Principal Component Analysis (PCA) technique (Appendix-2). SD, CPI, TO, and GDPC denote sustainable development, Consumer Price Index, trade openness, and GDP per capita, respectively. Sustainable development (SD) is measured using Adjusted Net Savings (ANS) as a percentage of Gross National Income, in line with Dutta and Saha (2023a, b). This approach integrates economic, environmental, and social dimensions of sustainability, drawing on the concepts of “wealth accounting” and “green national accounts” (Peeters, 2003). ANS captures the actual level of aggregate savings by accounting for human, manufactured, and natural capital (Huang, 2012), making it a robust measure for sustainable development. Additionally, \({A}_{0i}\) and \({\mu }_{{it}}\) represent 1×k vectors of country-specific intercepts (constant over time) and idiosyncratic disturbances, respectively, while A(ℓ) is a k×k polynomial matrix capturing lagged coefficients.

To prevent estimation errors resulting from the correlation between \({A}_{0i}\) and \({\mu }_{{it}}\), and to eliminate specific fixed effects, we employ generalized method of moments (GMM) method. This approach employs lagged variables as instruments and applies the forward orthogonal deviations transformation, as introduced by Arellano and Bover (1995), as described in Eq. (2)

$${y}_{{it}}^{* }=\left({y}_{{it}}-\bar{{y}_{{it}}}\right)\sqrt{\frac{{T}_{{it}}}{{T}_{{it}}+1}}$$
(2)

where \({{\rm{every}}{y}_{{it}}\in {Y}_{{it}},T}_{{it}}\) is the available observation for country i at time t, and \(\bar{{y}_{{it}}}\) represents its mean. Compared to the first difference transformation, this provides a few benefits. Forward orthogonal deviation minimises data loss and results distortion by variables gaps between observations, as is the situation with imbalanced panels, by calculating deviations from a mean instead of from another observation. Lastly, after verifying that the GMM-PVAR is stable, we estimate the forecast error variance decomposition (FEVD) and impulse response functions (IRF).

System Generalized Method of Moments method

To support our assertion of nonlinearity and identify the threshold level of public debt, we estimate the following equation:

$$\begin{array}{ll}{{SD}}_{{it}}={\beta }_{1}{{SD}}_{{it}-1}+{\beta }_{2}{{PD}}_{{it}-1}+{\beta }_{3}{\left({{PD}}_{{it}-1}\right)}^{2}+{\beta }_{5}{{GOV}}_{{it}-1}\\\qquad\qquad+\mathop{\sum }\limits_{j=1}^{k}{\gamma }_{j}{X}_{{it}-1,j}{+\nu }_{t}+{e}_{{it}}\end{array}$$
(3)

where subscripts i and t denote country and time, respectively. Here, SD represents sustainable development, and its first lag is included as a regressor to account for persistence over time. PD stands for public debt, while GOV represents governance. The term \({{\rm{X}}}_{{\rm{it}}-1,{\rm{j}}}\) encompasses a set of {k} control variables to account for country-specific heterogeneity. To mitigate endogeneity concerns, we use the lagged values of all explanatory variables. Additionally, we incorporate a quadratic term of PD to capture potential nonlinearity in the public debt–sustainable development relationship. The coefficients β’s represent the parameter vectors, \({{\rm{\nu }}}_{{\rm{t}}}\) denotes time-specific fixed effects, and \({\text{e}}_{\text{it}}\) is the error term.

To estimate the aforementioned equations, we account for endogeneity, a critical concern in cross-country analyses. Endogeneity may arise due to factors such as reverse causality, simultaneity bias, measurement error, and unobserved heterogeneity. Reverse causality, for instance, can occur if unsustainable development leads to natural disasters, socio-economic instability, and government inefficiency, thereby increasing public debt and undermining overall governance. Additionally, a bidirectional relationship between these variables may further exacerbate endogeneity concerns.

To address these challenges, we employ the System Generalized Method of Moments (System-GMM) estimator, which provides consistent and efficient estimates while effectively handling unobserved effects and endogeneity. System-GMM is particularly well-suited for panel studies with a limited number of time periods, as is the case in this study. This method treats the model as a system of equations—one for each time period—where the predetermined and endogenous variables in first-differences are instrumented using suitable lags of their own levels, thereby mitigating endogeneity issues (Blundell & Bond, 1998).

Empirical results and insights

Statistical summary

Table 1 presents the summary statistics for the key variables, highlighting substantial cross-country variations. Sustainable development (SD) exhibits a mean of 8.28 with a standard deviation of 12.68, indicating notable disparities in sustainability performance. The minimum value of −54.72 (observed in the Solomon Islands in 1991) reflects severe sustainability challenges, potentially driven by economic instability, environmental degradation, and weak institutional frameworks. Conversely, the maximum value of 43.69 (recorded in Belize in 1991) suggests relatively stronger sustainability, likely supported by effective resource management and policy interventions. The broad range of SD underscores the heterogeneous nature of sustainable development across nations, shaped by differences in economic growth, environmental policies, and social investments.

Table 1 Statistical summary.
Full size table

Public debt (PD) also displays considerable variation, with an average of 50.62 and a standard deviation of 32.88. The lowest recorded debt level of 3.67% of GDP (Thailand in 1996) suggests strong fiscal discipline or limited reliance on borrowing. In contrast, Zambia’s peak debt level of 277.53% of GDP (1991) highlights significant fiscal strain, potentially linked to economic crises or structural adjustments. These disparities reflect differing fiscal policies, economic structures, and external debt dependencies among nations.

Governance (GOV) exhibits relatively lower variability, with an average of 0.51 and a standard deviation of 0.19, suggesting that while governance quality differs across countries, it does not fluctuate as drastically as other macroeconomic indicators.

Other macroeconomic indicators, such as GDP per capita (GDPC), consumer price index (CPI), trade openness (TO), and Government Debt (GD), also show substantial dispersion. GDP per capita ranges from 110.46 to 15,941.45, reflecting the stark economic divide between low- and high-income countries. CPI varies widely, from 0 to 536.54, illustrating differences in inflationary pressures across economies. Trade openness (TO) spans from 0.086 to 1.538, indicating varying degrees of economic integration with global markets. Finally, Government Debt (GD) fluctuates significantly, from −93.78% to 72.82%, capturing different debt dependencies of economies.

Baseline GMM-PVAR estimations

The econometric literature emphasizes the importance of testing for stationarity in the presence of cross-sectional dependencies (Abrigo and Love, 2016). For the GMM-PVAR estimation, we begin by assessing the panel unit root of the variables using the widely adopted second-generation unit root test proposed by Pesaran (2007). The results of this analysis are presented in Table 2.

Table 2 Contemporary panel unit-root test (second-generation).
Full size table

To calculate unit-root, first-generation tests require the variables to be independent across countries, which is often difficult to ensure for co-movements of macroeconomic variables. Therefore, we use the second-generation test (Pesaran, 2007) that allows for cross-sectional dependency of the variables in a panel. This test was also used by Koengkan et al. (2020). Results show that all the variables at their first differences possess stationarity irrespective of trend specification of the test; therefore we use first differences of variables for further estimations.

Deciding the optimal lag-order is key for GMM-PVAR estimation. Inadequate lag-order may produce inaccuracy, while exaggerated lag-order decreases the degree of freedom and induces an over-parameterization problem (Boubtane et al., 2013). Based on the optimal lag outcomes presented in Appendix-3 (Table A3), we establish and estimate a second-order GMM-PVAR model using the GMM technique. Although the BIC, AIC, and HQIC criteria are least at lag 1, the first-order panel VAR model rejects Hansen’s over-identification restriction at the 5% confidence level. Consequently, we opt for the second-order panel VAR model.

The results in Table 3 indicate that all variables are interrelated. Public debt exhibits a dynamic relationship with sustainable development. Specifically, at the first lag, PD has a positive and statistically significant impact on SD, with a magnitude of 11.3% at the 10% significance level. However, this relationship reverses at the second lag, where PD exerts a strong negative effect of 35.3%, significant at the 1% level. On the other hand, a one standard deviation increase in SD leads to a 9.4% reduction in PD, an effect that is statistically significant only at the first lag. These findings suggest that while public debt may initially contribute to sustainable development, it ultimately undermines long-term sustainability. In contrast, sustainable development appears to consistently exert a moderating effect on public debt accumulation.

Table 3 GMM-PVAR estimation.
Full size table

Additionally, PD negatively influences governance. At the second lag, PD has a significant negative effect of 12.5%, despite a weak and insignificant positive impact at the first lag. This implies that higher public debt undermines governance over time, possibly due to increased corruption, mismanagement of funds, and reduced government effectiveness. In contrast, governance has a strong positive impact (40.3%) on SD at the first lag, reinforcing that effective governance is crucial for promoting sustainable development—a finding consistent with Dutta and Saha (2023a, 2023b).

Interestingly, GDP per capita (GDPC) has an insignificant positive impact on sustainable development (SD) at the first lag, but a statistically significant negative effect of 12.3% at the second lag. This finding aligns with the conclusions of Dutta and Saha (2023b), who empirically demonstrate that higher levels of economic development do not necessarily translate into greater sustainable development or inclusive growth. Instead, increased GDPC may come at the expense of environmental degradation and rising social inequality. Similarly, trade openness, while promoting economic activity, can also lead to environmental pollution and exacerbate income disparities.

Next, we approach determining the causality using the GMM-PVAR Granger causality test (Table 4).

Table 4 PD, GOV-SD Granger causality.
Full size table

The null hypothesis that “Public debt (PD) does not Granger-cause sustainable development (SD)” is rejected at the 99% confidence level. Similarly, the null hypothesis that “SD does not Granger-cause PD” is also rejected at the same confidence level, indicating a bidirectional causal relationship between public debt and sustainable development. These findings are consistent with previous studies by Ncanywa and Masoga (2018) and Owusu-Nantwi and Erickson, 2016.

Additionally, the hypothesis that “Governance does not Granger-cause SD” is rejected at the 99% confidence level, whereas “Governance does not Granger-cause PD” fails to be rejected, indicating that governance significantly influences sustainable development but has no discernible causal effect on public debt. Conversely, both hypotheses—“PD does not Granger-cause Governance” and “SD does not Granger-cause Governance”—are rejected at the 99% confidence level, suggesting that both public debt and sustainable development independently affect governance.

Overall, the GMM-PVAR Granger causality results reveal bidirectional causality between PD and SD, as well as between Governance and SD, while the relationship between Governance and PD appears to be unidirectional, flowing from PD toward Governance.

Without imposing restrictions on parameters estimates, we cannot explain the causal relationship (Abrigo and Love, 2016). Therefore, we use the Impulse Response Function (IRF). However, before estimating IRF, we first examine the stability condition of the GMM-PVAR model. The eigenvalues of the estimated coefficients are all less than one and fall within the unit circle (Fig. 2), confirming the stability of the GMM-PVAR estimation.

Fig. 2: Stability graph.
Fig. 2: Stability graph.
Full size image

This graph shows the eigenvalues of the estimated coefficients, and all are less than one and fall within the unit circle, which confirms the stability of the estimated GMM-PVAR results.

To further reinforce the PD-SD, Governance-PD and Governance-SD causality, we conduct the IRF analyses, since the reduced-form GMM-PVAR coefficients are not credible to infer causality until restricted parameter estimates are considered (Abrigo and Love, 2016). The IRF plot (Fig. 3) explains the responses of PD (1st column), Governance (2nd column) and ANS (4th column) to a shock in either ANS, Governance, GDPC, CPI, TO or PD, while assuming other stimulus equal to be zero. The solid lines in the plots represent the orthogonal IRF of the respective variable over a five-year period, while the shaded regions denote the 95% confidence intervals. These confidence intervals are calculated using 200 Monte Carlo simulations derived from the distribution of the fitted reduced-form GMM-PVAR model.

Fig. 3: IRF pots.
Fig. 3: IRF pots.
Full size image

The IRF plot explains the responses of PD (1st column), Governance (2nd column) and ANS (4th column) to a shock in either ANS, Governance, GDPC, CPI, TO or PD, while assuming other stimulus equal to be zero.

The resultant IRF plot in the fourth column, fourth row illustrates that PD and SD have a nonlinear interaction that starts from a positive impact with an episodic negative influence that declines subsequently and eventually disappears in the third year approximately. The result aligns with the previous researchers’ findings (Eberhardt and Presbitero, 2015; Ncanywa and Masoga, 2018; Tung, 2020). The intuition behind these results could be that PD is essential to support the large development activities of the developing nations, yet, when PD grows excessively large, payback becomes a running sore, and then instead of supporting development and welfare, PD begins eroding economic well-being and diverting resources from investing in social sectors (Bakar and Hassan, 2008; Phiri and Fotoyi, 2023). Moreover, higher PD may lead to higher GDP; however, these higher economic activities often result in higher emissions (Rai et al., 2019) and thereby hamper SD.

Similarly, the plot in the second column, second row shows that SD reduces PD, and its influence eventually fades away because SD requires a healthy economy (Pezzey, 1992) and a healthy economy reduces PD (Greiner and Fincke, 2009).

Besides, the plot in the fifth column, first row, evinces that strong governance upholds SD, which diverges to zero over time. This implication corroborates the findings of Dutta and Saha (2023b), who show that governance promotes SD in both developed and developing countries. Interestingly, in plot second column, first row, the impacts of governance on PD start with a negative intersect followed by a positive impact, which disappears in the third period. This can be attributed to the fact that good governance discourages PD, which is also suggested by Sundararajan et al. (1997). This initial result is also consistent with Neumayer (2002), who suggests that countries with good governance should receive a high portion of PD remission. However, countries with effective governance find it easier to access PD (Saha and Dutta, 2023), which leads to increasing PD subsequently.

Surprisingly, the relationship between PD and governance (in the plot third column, second row) exhibits a temporary positive effect followed by a negative impact in the long run, because countries that accumulate PD tend to initially maintain good governance but eventually struggle, resulting in PD harming governance. This can be attributed to the fact that countries with effective governance find it easier to access PD (Saha and Dutta, 2023), therefore, countries that require PD focus on improving governance initially. However, fiscal constraints, limited policy flexibility, reduced public trust resulting from high PD weakened institutions. SD, on the other hand, has an increasing effect on governance, followed by a decrease, and eventually disappears in the long run. This may occur due to the shifting priorities, lack of continuity and changing political government.

These results empirically confirm the exposure to debt overhang, which policymakers in emerging economies must closely monitor to avoid any interruptions in the path of attaining good governance and SD.

Among other variables, GDPC and CPI hamper SD, as said earlier, that higher GDPC does not necessarily mean higher inclusive development and higher CPI may lead to higher inequalities in society. The impact of TO on SD is not significant.

Lastly, we enhance the analysis by conducting a Forecast Error Variance Decomposition (FEVD) to assess the cumulative contribution of one variable in illustrating variations in others. The results for our main variables of interest—ANS, PD, and GOV—are presented in Table 5. The variance decomposition reveals that PD and GOV account for approximately 2.6% and 3.05% of the variability in SD, respectively, while SD elucidates 6.7% and 2.3% of the changes in PD and GOV, respectively.

Table 5 Variance decomposition analysis: Public Debt and Sustainable development.
Full size table

Nonlinearity of public debt and sustainable development nexus

Table 6 shows the SGMM estimation of Eq. 3. The positive and statistically significant coefficient (0.470) associated with the lagged dependent variable (ANS) indicates strong persistence in sustainable development, as measured by ANS. This suggests that past values of ANS have a substantial influence on its current values, reinforcing the notion of path dependency in sustainability outcomes.

Table 6 Explaining the nonlinearity of PD-SD nexus: SGMM.
Full size table

The variable representing governance (GOV) exhibits a positive and marginally significant effect, implying that improvements in governance contribute positively to sustainable development. Effective governance fosters financial stability, enhances economic performance, and promotes environmental sustainability by ensuring a stable macroeconomic environment. Strong institutional frameworks and policy implementations can support long-term sustainability by reducing uncertainties and facilitating efficient resource allocation.

The coefficient for lagged public debt is positive and statistically significant, suggesting that, within a certain range, public debt-financed policies can enhance sustainable development. However, the squared term of lagged public debt \({{({\rm{PD}}}^{2}}_{{\rm{t}}-1})\) exhibits a significant negative coefficient, indicating a nonlinear relationship. To improve interpretability, Fig. 4 presents a graphical representation of the relationship between public debt (PD) and sustainable development (SD).

Fig. 4: Nonlinearity of PD-SD nexus.
Fig. 4: Nonlinearity of PD-SD nexus.
Full size image

Figure 4 shows the nonlinear relation of Public Debt and Sustainable Development and suggests an inverted U-shaped effect.

To construct this visualization, we generated a range of public debt values from 0 to 1 at equal intervals. The corresponding predicted values of sustainable development were computed by substituting the value of β1 and β2 (the estimated regression coefficients) in Eq. (4):

$${ANS}={\beta }_{1}{PD}+{\beta }_{2}{{PD}}^{2}$$
(4)

These predicted values were then plotted, with public debt on the X-axis and predicted sustainable development on the Y-axis, illustrating the quadratic nature of the relationship.

Figure 4 suggests an inverted U-shaped effect, where moderate levels of public debt and associated policy interventions stimulate sustainable development. However, excessive accumulation of debt may lead to diminishing returns or even negative consequences.

To determine the threshold level of Public Debt from the provided regression results, we need to focus on the coefficients of \({{\rm{PD}}}_{{\rm{t}}-1}\) and \({{{\rm{PD}}}^{2}}_{{\rm{t}}-1}\) .The threshold of PD occurs where the marginal effect of PD on ANS is zero. This is found by taking the derivative of the equation with respect to PD and setting it to zero:

$$\frac{d({ANS})}{d({PD})}={\beta }_{1}+2{\beta }_{2}{PD}=0$$

Solving for PD:

$${PD}=-\frac{{\beta }_{1}}{2{\beta }_{2}}$$

Substituting the coefficients:

$${PD}=-\frac{0.611}{2(-1.109)}=\frac{0.611}{2.218}\approx 0.275$$

This calculation indicates that when public debt remains below 27.5%, its impact on sustainable development is positive. However, once public debt exceeds this threshold, further increases have an adverse effect on sustainable development.

While our study focuses exclusively on developing countries, which prevents direct comparisons with higher-income economies, the identified threshold effects likely reflect structural and institutional conditions prevalent in these settings—such as varying degrees of financial depth, regulatory capacity, and macroeconomic stability. Our findings suggest that the level at which public debt influences sustainable development could differ in more advanced economies, where stronger market institutions, better governance, and more stable policy frameworks might alter the relationship. Future research should examine these dynamics across a broader range of countries to determine how thresholds vary by income level, which would strengthen the policy relevance of these findings.

Notably, countries with more developed financial systems or higher institutional quality may exhibit higher thresholds, implying that such systems can withstand greater public debt exposure before significantly affecting sustainable development. Although our analysis is restricted to developing economies, the sample includes diverse regions—such as Sub-Saharan Africa, South Asia, and Latin America—which differ in economic structures and policy environments. While we do not estimate region-specific thresholds due to limited sample sizes within each sub-group, these regional variations highlight the need for context-specific policy considerations when interpreting public debt effects.

Among control variables, Inflation exhibits a positive coefficient, suggesting a mild positive impact on the dependent variable. This may reflect the idea that moderate inflation is indicative of robust economic activity, potentially boosting corporate profitability and investment. However, excessive inflation could have adverse effects, which may not be fully captured within the current model specification.

GDP per capita does not demonstrate a statistically significant effect on ANS, implying that its direct influence on sustainable development is not clearly established within this framework. This could be due to the complex interactions between GDP per capita and other macroeconomic variables or the possibility that GDP per capita serves as a proxy for broader economic conditions already accounted for by other factors in the model.

Trade openness also does not exhibit a significant effect on sustainable development, suggesting that external trade policies alone may not directly shape sustainability outcomes. The impact of trade openness could depend on factors such as the composition of trade, exposure to external shocks, and the level of financial development in different economies.

Diagnostic tests confirm the robustness of the model. The absence of second-order autocorrelation (AR(2) = 0.628) suggests that the model specification is appropriate and that the dynamic panel estimation is reliable. Additionally, the Hansen J-test (p-value = 0.890) indicates that the instruments used in the SGMM estimation are valid, implying that the model does not suffer from over-identification issues.

Identifying key governance drivers of sustainability

To identify which specific dimensions of governance matter for sustainability, we estimate equation-3 with different disaggregated Governance Indicators, instead of the composite governance index, using the same moment’s condition of the previous SGMM estimation. Results are reported in appendix – 4, Table 5A. Each column includes a different governance indicators lagged by one year—CC (Control of Corruption), PS (Political Stability), GE (Government Effectiveness), RQ (Regulatory Quality), RL (Rule of Law), and VA (Voice and Accountability). The results provide clear evidence that not all governance indicators contribute equally to promoting sustainability, as measured by ANS. Specifically, Control of Corruption (CC), Regulatory Quality (RQ), Rule of Law (RL), and Voice and Accountability (VA) all show statistically significant and positive coefficients, indicating that they strongly enhance sustainable development. Among them, Regulatory Quality (RQ) has the highest coefficient (0.664, significant at 5%), suggesting it may be the most influential mechanism, likely because effective regulations support resource-efficient, environmentally responsible, and growth-oriented policies. Rule of Law (0.510) and Voice and Accountability (0.547), both significant at the 1% level, also emerge as crucial. They ensure legal predictability, protection of rights, and public pressure for sustainability. Control of Corruption (0.283, significant at 10%) contributes by reducing inefficiencies and misallocation, though its effect appears somewhat smaller in magnitude and significance compared to RQ and RL.

On the other hand, Political Stability (PS) and Government Effectiveness (GE) are not statistically significant, suggesting that while important for general governance, they may not directly influence sustainable development outcomes in this dataset.

Robustness Test: Alternative measure of public debt

To validate the robustness of the baseline results, adopting similar approaches, we re-estimate the second-order PVAR model using Government Debt (GD)—defined as claims on central government (% of GDP)—as an alternative measure to Public Debt (PD). While PD reflects the total debt stock of the central government, GD emphasizes domestic financial institutions’ exposure to government liabilities. This distinction helps assess whether our findings are sensitive to how public debt is financed and held.

Since the coefficient of the reduced form PVAR estimate could not be regarded as a causal relationship without imposing limits, here we estimate and focus on the findings of IRF only. GD, Governance and SD IRF plots (Fig. 5) reinforce the previous findings, suggesting that initially GD contributes to the transition to sustainable development; though, the GD-SD interaction after a certain period becomes disadvantageous, exhibits a downward trend, and eventually discontinues being responsive. Besides, the governance contributes to strengthening sustainable developmen,t which corroborates our baseline findings. Other results are also consistent to the baseline findings.

Fig. 5: GD, Governance and SD IRF plots.
Fig. 5: GD, Governance and SD IRF plots.
Full size image

The IRF plot explains the responses of GD, an alternative measure of PD, (1st column), Governance (2nd column) and ANS (4th column) to a shock in either ANS, Governance, GDPC, CPI, TO or GD, while assuming other stimulus equal to be zero.

Conclusion

Public debt is crucial for fostering development, but without effective governance, it can hinder sustainability. Governments worldwide grapple with the challenge of balancing the necessity for public investment to drive sustainable development against the constraints of debt obligations. The interplay between public debt, governance, and sustainable development is intricate and multifaceted, demanding a thorough comprehension of its underlying dynamics.

Despite this complicated interaction, comprehensive studies to discuss the Public debt, governance, and sustainable development relationship are missing. To fill this gap, we estimate PD, governance, and SD nexus using GMM-PVAR and SGMM estimation, both are advanced econometric model. The results mainly exhibit bidirectional causality among Public debt, Governance and Sustainable development. Additionally, it reveals that Public debt demonstrates an inverted U-shaped relationship with sustainable development. At moderate levels, public debt can act as a catalyst for economic growth by financing critical infrastructure, social programs, and development initiatives. However, beyond a certain threshold, excessive debt accumulation leads to diminishing returns, crowding out private investment, and increasing the risk of economic instability, thus hampering sustainable development. Empirical evidence also suggests that the optimal threshold for public debt lies at approximately 27.5% of GDP. Exceeding this threshold tends to reverse the positive effects of debt, underscoring the importance of prudent debt management, strengthening governance and the adoption of sustainable fiscal policies to mitigate adverse economic consequences.

Furthermore, effective governance (GOV) plays a pivotal role in enhancing financial stability, economic performance and inclusive development. Strong institutional frameworks, transparent policy implementation, and efficient resource allocation are critical components of governance that ensure public debt is utilized productively. Governance mechanisms also help maintain investor confidence, reduce corruption, and foster long-term economic resilience. Thus, the interplay between public debt and governance is essential for achieving sustainable development.

Implications of the study

This study will contribute to government policy and help develop measures to avoid the viciousness of the debt overhang problem. High public debt has long been a concern for policymakers, particularly in emerging economies. The ability to finance fiscal deficits through borrowing is often seen as a necessary, if imperfect, solution. However, excessive public debt can have detrimental effects on long-term inclusive growth and development. Unbridled debt might turn PD from a boon to a bane. Therefore, Policymakers of developing countries should prioritize maintaining public debt levels below the estimated threshold of 27.5% of GDP to avoid adverse economic effects. This requires careful monitoring of debt accumulation and the implementation of counter-cyclical fiscal policies during periods of economic growth to build fiscal buffers. In cases where debt levels approach or exceed the threshold, proactive debt restructuring measures should be considered. This may include renegotiating terms with creditors, extending maturities, or diversifying funding sources to reduce repayment pressures.

In addition, institutional quality (RQ, RL) and citizen engagement (VA) are especially potent drivers of sustainability. Policymakers should focus on improving regulatory frameworks, strengthening legal institutions, enhancing democratic accountability and ensuring transparency in public spending, and promoting accountability to prevent mismanagement of resources as targeted governance reforms to promote long-term sustainable development. By integrating these policy measures, governments can harness the positive effects of public debt while minimizing risks, thereby fostering sustainable economic growth and development.

While this study makes important contributions to understanding the relationships between public debt, governance, and sustainable development in developing economies, several limitations warrant consideration. The most significant arises from our exclusive focus on developing countries. Although this focus is methodologically justified by data availability and research objectives, it inevitably constrains the generalizability of our findings. Specifically, the threshold effects identified for the impact of sovereign public debt on sustainable development may not hold in advanced economies, where more sophisticated financial systems, stronger institutional frameworks, and distinct policy environments could produce fundamentally different dynamics. For instance, the greater financial resilience and regulatory capacity of high-income countries may result in either higher threshold levels or entirely different patterns of association.

Moreover, even within the group of developing countries, substantial regional variation in economic structures and institutional quality likely introduces unobserved heterogeneity that our analysis cannot fully account for. This suggests that the estimated threshold may itself differ across subgroups within the developing world. Taken together, these limitations highlight two key directions for future research: (1) cross-country comparative studies assessing whether similar threshold effects emerge in high-income contexts, and (2) more disaggregated regional analyses within the developing world to uncover potential variation in threshold behaviour. Such extensions would enhance the external validity of the findings and support the formulation of more context-sensitive policy recommendations.