External Debts, Re-Election Possibilities and the Infrastructure Development Via Public Private Partnership: Evidence from Low- and Middle-Income Countries

Abstract

Public private partnership in infrastructure development can be used as an ‘off the balance' tool to help governments hide real fiscal problems and facilitate the transfer of debt-repayments to their succeeding administrations. Using negative binomial regressions on a panel dataset of 40 low-and middle-income countries from 1992 to 2021, the paper finds that governments which high short-term external debts in stock and flows measurements are more motivated to participate in public private partnership. The results are robust through sensitive tests on different sampled dataset. These findings reaffirm the existence of fiscally opportunistic behaviors of public sectors in public private partnership in low-and middle-income countries. Results also indicate that governments which face re-election limitations have stronger motivations to implement infrastructure projects via public private partnership. These impacts are found to be stronger in less-developed economies and in Sub-Saharan African countries.

KEYWORDS:

• Public private partnership
• short-term external debts
• re-election limitation
• low- and middle-income countries
• negative binomial regressions

JEL Classifications:

• H63
• H70
• J11
• P00

1. Introduction

Private participation is believed to be a solution for public sector’s failures in provision of public goods and infrastructure development (Engel et al., 2013; Saussier & de Brux, 2018; Xiong et al., 2020), especially in developing countries. It is estimated that low-and middle-income countries face an average annual gap of from 1.0 to 1.5 trillion USD for infrastructure development goals under non-Covid calculations (United Nations, 2015). These capital shortages have been aggravated under current global economy which is embodied by rising deglobalization (Chen et al., 2023), long lasting pandemic, geopolitical related contractions of financial flows (Feng et al., 2023; Tetteh & Ntsiful, 2023).

World Bank (2022) finds that budget constraints governments and heavily external indebted economies are leaning to public private partnerships to finance infrastructure projects, with highly probable solvent risks entailed such as debt-hiding motivations (Buso et al., 2017) or rent-seeking behaviors (Alesina & Passalacqua, 2015 and Buso et al., 2017) which can results in sovereign solvency problems (Blanchard, 2019; Kose et al., 2021; Lorenzoni & Werning, 2019). Moreover, the existence of political opportunism (Bernhardt et al., 2004; Raveh & Tsur, 2020, 2023) compounded with ill-planned public private partnership projects results in deteriorated insolvency in low-and middle-income countries (Fabre & Straub, 2023; VCCI, 2022; VietnamNews, 2023).

Hence, this paper aims to examine relationship between external debts as a proxy variable for presence of debt hiding motivation and the implementation of public private partnership for infrastructure development in low-and middle-income countries. Additionally, the paper will investigate the existence of politically opportunistic behaviors of governments via a proxy of re-election limitation in the implementation of public private partnership in low-and middle-income countries.

Theoretically, public private partnership mechanism can help reallocate risks between public sectors and corresponding private counterparts among countries, regardless levels of development (Yurdakul et al., 2022). Under the context of low-and middle-income countries, private sector can play more active roles in providing infrastructures services (Zhang, 2014) and cover the limitations of capital, technological capability and management of public sector.

Empirically, public private partnership projects are closely linked to the budget constraints of governments (Amos & Zanhouo, 2019; Buso et al., 2017; Wang et al., 2019). This linkage is vicious because budget constraint governments tend to accept low quality public private partnership (Xiao & Lam, 2022) with small, domestic private partners due to uncapability to partner with larger domestic and credited foreign counterparts. This exercise leads to presence of cream skimming (Bonnafous, 2012) or encourage opportunistic/fiscally irresponsible behaviors of public sectors (Alesina & Passalacqua, 2015; Klašnja & Titiunik, 2017). Furthermore, Brender and Drazen (2008) find that governments can use public private partnership to avoid punishment from voters for their budget deficits.

Our findings contribute to current literature by providing evidence on the debt hiding motivations of governments and the presence of political opportunism in the implementation of public private partnership by regressing a panel dataset of 40 low-and middle-income countries from 1992 to 2021. Additionally, our results reaffirm the impacts of short external debt liabilities on public private partnership. The findings again showcase the existence of ‘off the balance’ effects of public private partnership which is in line with the conclusion of Amedanou (2023). These are expected to provide more insightful guildlines on external debt sustainability for both the borrowers and creditors in current global context.

Apart from the introduction, the paper will be organized as follows: Part 2 will provide a critical literature review. Part 3 discusses the examined data, variables, and possible econometric issues with recommended solutions. Part 4 displays the empirical results and interpretations. Conclusions and policy implications will be covered in Part 5.

2. Literature Review

2.1. Theoretical Background

The paper is constructed on premises of interactively and recursively institutional theory which provides useful frameworks for economists, political scientists and sociologists (Scott, 2005) to examine national and global macroeconomic problems (Arranz & Arroyabe, 2023; Scott, 2005) because of its well-establishment on rationality – a fundamental pillar of economics (Djelic & Quack, 2003). The rationality in the form of ‘rational choice’ in institutional theory (Djelic & Quack, 2003) helps explain mechanism of rationale negotiations for interests among all actors within institutional order(s) through coordinations (Crouch & Streeck, 1997). The theory investigates interactive and recursive responses of actors subject to determined institutional structure and relative power comparison at both state levels and transnational level (Crouch & Streeck, 1997; Djelic & Quack, 2003). Accordingly, actors as state agencies possess the power of ‘control and sanction’ obtaining from exclusive policy making capability; while private actors have influences based on resource accessibility (Strange, 1996) such as technologies, knows-how and capital (Núñez-Serrano & Velázquez, 2017; Wang et al., 2019). The relative exertion of these power/authority will classify actors into dominant and fringe actors (Djelic & Quack, 2003). At state level, dominant actors will actively promote institutional changes through rationalized mechanism and take a driving role in the process to promote collaboration during crisis time (Djelic & Quack, 2003). At a transnational perspective, state-level dominant actors play as a catalyst for institutional changes by learning, appplying or introducing new rules and/or policies from other countries (Djelic & Quack, 2003; Mohammad Ebrahimi & Koh, 2021; Scott, 2005).

Public private partnership arrives when state agencies face budget constraints, debt burden and possibly executive problems; and private sectors suffers from abundance in management, technological capability, and capital resources (Hammami et al., 2006; Kresch & Schneider, 2020; Roehrich et al., 2014; Zhang, 2014). These forms of crises are typical in low-and middle-income countries (Hammami et al., 2006; Roehrich et al., 2014) which result in over-popularity of public private partnership implementation in economies as a form of transnational institutional impacts on state level.

2.2. Empirical Review

In general, the relationship between public private partnership and macro-economic performance is controversial (Collier et al., 2016; Fay et al., 2021; Percoco, 2014; Takano, 2017; VietnamNews, 2023; Wang et al., 2019; Yurdakul et al., 2022). On one hand, public private partnership contributes significantly to economic growth by providing additional capital/technological/managerial resources for public partners.

On the other hand, public private partnership embodies various macro risks (Ampratwum et al., 2022; Bastida et al., 2017; Bonnafous, 2012; Reinhart & Rogoff, 2010), especially in low-and middle-income countries due to lack of high-quality institutions and macroeconomic uncertainties. It is evident that public private partnership can be useful for governments to hide real local public debt (Buso et al., 2017; Liebe & Pollock, 2009) or as an ‘off the balance’ expense to ensure growth targets (Xiong et al., 2020). Reaching to long-term borrowing caps from external creditors indicates ongoing costlier and riskier short-term external loans. This scenario creates a vicious fiscal cycle. Aggregately, borrowers face huge pressures from debts repayments and more expensive borrowing (Avellán et al., 2021; Blanchard, 2019; Evans, 2020; Lian et al., 2020). Consequently, higher ratio of short-term debts to total external debt signalizes unstable and volatile macro environments which can lead to unexpected consequences of economic growth and governmental low creditability among voters.

Wu and Lin (2022) find out that public private partnership projects in China are mainly motivated by budget constraints and political imperatives of local governments. Fabre and Straub (2023) argue that average cost of an infrastructure project implemented in public private partnership is higher than that in on traditional ways. Therefore, public private partnership can be a preferred approach of governments with highly external indebtedness (Amedanou, 2023).

Public private partnership can be used by heavily deficit governments to transfer liabilities of debts (re)payments to their successors (Linder, 1999; Roehrich et al., 2014). These exercises will harm fiscal space of succeeding governments to achieve development goals (Chalk, 2004; Xiong et al., 2020). It would be more problematic as long as the private partners’ resources are financed by foreign currencies (Hammami et al., 2006) or governments over-guarantee debts of private sectors to attract their participation. Especially, corrupted and weak institutional countries attract more public private partnership projects (Banerjee et al., 2006; Buso et al., 2017; Takano, 2017) as this mechanism can facilitate rent-seeking behaviors of public and private actors (Takano, 2017).

Politically, Flinders and Buller (2006), and Klašnja and Titiunik (2017) find that public private partnership can be used for tactically purposes by dominant state agencies. It is evident in the absence of sufficient counterweights, politicians will behave in ‘personalistic and anti-institutional’ manners (Levitsky & Cameron, 2003; Raveh & Tsur, 2023; Takano, 2017). This politically opportunistic effect is stronger when state agencies face reelection limitation (Bernhardt et al., 2004; Raveh & Tsur, 2020, 2023).

Additionally, when a government faces limitations in re-election, it is motivated to behave opportunistically to remain their own positively political impacts among votes by sacrifying fiscal benefits of successors (Bernhardt et al., 2004). This argument can be read that being limited in re-election will push for more public private partnership implementation in low-institutional qualified economies as a form of ‘opportunistic behaviors’ (Raveh & Tsur, 2023). Previous studies find that higher possibility of re-election gives a rise in debt accumulation (Raveh and Tsur, 2020) in any forms, in and off the balance otherwise. In addition, Saiz (2006) concludes that on average, more roads are built under dictatorship governments than more democratic administrations.

Empirically, numerous public private partnership projects in developing and even developed countries resort into cancellation or renegotiation their public private partnership due to inefficiency, poor-planned (Engel et al., 2013) and unexpected economic – social consequences (Fabre & Straub, 2023; VCCI, 2022; VietnamNews, 2023). Specifically, public private partnership in the Philippines and Vietnam is considered inefficient and time-consuming (AsianInvestor, 2019). Liebe and Pollock (2009) conclude that public private partnership does not unconditionally benefit to governments in terms of fiscal spaces due to embedded higher risks and costlier returns for private investors.

As public private partnership is under-explored theoretically and empirically, authors construct two hypotheses as follows:

H1: Highly external indebtedness will encourage governments to coordinate with private actors via public private partnership mechanism as evidence of debt hiding.

H2: Facing reelection limitations will push governments to join public private partnership mechanism as a proof of political opportunism.

To address these hypotheses, authors employ different econometric methods and robustness checks to test the hypothesis which will be presented in next sub-section.

2.3. Econometric Methodology

The paper selects number of public private partnership projects as main dependent variable of interest as each project can provide useful implications about external fiscal liabilities and economic performances (Munoz-Jofre et al., 2023; Opara et al., 2017; Sheppard & Beck, 2023; Xiao & Lam, 2022). Statistically, after removing China, India, Argentina and Brazil for outlier controlling, the paper’s final dataset shows that number of these projects ranges from 1 to 25, with the average value of 3.1 projects annually in low- and middle-income countries. These numerical descriptions reflect that challenges, fiscally and institutionally, for low-and middle-income countries to manage a well-planned and ultimately successful public private partnership projects.

Consequently, against theoretical and empirical backgrounds, the paper utilizes Poisson and Negative Binomial regressions which are is evidently appropriated to a count dependent variable (Luo et al., 2022; Suryadi et al., 2023). The selection of Poisson regression is resulted from countable nature of the dependent variable, which helps avoid negativity-related biasedness of linear regressions (Martin, 2024).

The Poisson approach employs an assumption of ‘equi-dispersion’ where mean and variance of the variable are equal (Martin, 2024; Suryadi et al., 2023). Martin (2024) argues that this assumption is too strong and hardly satisfied in social sciences. Thus, regressions results could be biased toward underestimated standard errors (in presence of ‘over-dispersion’ or overestimated standard errors) in the existence of ‘under-dispersion’ (Suryadi et al., 2023). Hence, ‘overdispersion’ and ‘under-dispersion’ tests are performed to ensure appropriateness of methodology applied and unbiased regression results. If the initial assumption of ‘equi-dispersion’ cannot hold, Negative Binomial Regressions are more appropriate than Poisson regression(Da Silva & Rodrigues, 2014; Martin, 2024; Suryadi et al., 2023).

Other econometric issues of count variable regressions are related to absence or presence of zeros (Da Silva & De Sousa, 2023; Narukawa & Nohara, 2018). Failures to control zeros’ impacts can cause inconsistency of standard parametric estimators (Narukawa & Nohara, 2018). Such absence or presence of zeros can be controlled by employing ‘zero-truncated’ or ‘zero-inflated’ count variable – related econometric technics (Luo et al., 2022; Martin, 2024). Subject to the absence of zeros in our final dataset, the paper employs Zero-Truncated Negative Binomial to deal with this possible econometric biasedness cause.

2.4. Econometric Models

The paper borrows econometric approach of Martin (2024) and Suryadi et al. (2023). As the nature of dependent variable is non-negative count. The Poisson probability distribution can be written as: $p(y)=\frac{{\mu }^y{ϵ}^{-\mu }}{y!},y=0,1,2,3,$where $p(y)$is the probability of observing the event y; $\mu$ is the predicted mean of $y$. The $ϵ$ will take value of 2.718. The Poisson distribution restrict to the equidispersion condition.

Then, the Log-likelihood function of Poisson distribution can be expressed as follows: $\mathcal{L}(\mu ;y)=\sum _{ı=1}^n\{y_i\ln ({\mu }_i)-{\mu }_i-\ln (y_i!)\}$Take ${\mu }_i=\text{exp}(X_i^{\mathrm{\prime }}\beta )$ as the inverse link of Poisson defined link of $\ln ({\mu }_i)$; and $y_i!$ can be expressed in the log gamma function $\ln \mathrm{\Gamma }(y_i+1)$ (Suryadi et al., 2023). After substitution ${\mu }_i$ and $y_i!$ in previous formula, then taking derivative of the Log-Likelihood Poisson function to zero, the Poisson model with can be rewritten as follows: $g({\mu }_i)={\eta }_i=y_i={\beta }_0+{\beta }_1{X}_1+{\beta }_2{X}_2+\dots +{\beta }_k{X}_k$When the equidispersion requirement is violated. Then, Poisson model will lead to biasedness. Then, Negative Binomial regression will provide more unbiased results (Da Silva & De Sousa, 2023; Da Silva & Rodrigues, 2014; Suryadi et al., 2023).

Define that $y_i$ follows Negative Binomial distribution with mean $E(y_i)={\mu }_i$ and Variance $\mathrm{Var}(Y_i)={\mu }_i(1+\alpha {\mu }_i).$

Assume $E(y_i)={\mu }_i={ϵ}_i\text{exp}(X_i^T\beta )$, Then, the Log Likelihood Negative Binomial functionscan be expressed as follows: $\begin{array}{rl}\ell (\beta ;\alpha )& =\sum \log (\mathrm{\Gamma }(y_i+{\mu }_i{\alpha }^{-1}))-\log (\mathrm{\Gamma }({\mu }_i{\alpha }^{-1}))-\log y_i!-{\mu }_i{\alpha }^{-1}\text{log}(\alpha )\\ & -(y_i+{\mu }_i{\alpha }^{-1})\text{log}(1+{\alpha }^{-1})\end{array}$Then, the maximum likelihood estimation can be performed by maximizing the previous function $\ell (\beta ;\alpha )$ with respect to $\alpha$ and $\beta$ (Suryadi et al., 2023).

2.5. Construction of Variables

Table 1 shows the construction of dependent variable, independent variables and alternative variables for our robustness check. Selections of dependent variables and independent variables are grounded on theoretical review mention aboved where external debt position is an important indicator for public private partnership (Amedanou, 2023). Especially, the ‘election’ binary varible is constructed based on a 5 level classifications of term limits. Other than employing categorical varibles which may create confusions among types of term limitation which are detailed in Table 1, the paper’s binary variable provides a more consistent, less biased, and comprehensible understanding on possible impacts of re-election limitation at all formats (Lami et al., 2021) on the implementation of public private partnership. The inclusion of other independent variables is closely referred to the work of Hammami et al. (2006) who state that GDP per capita, national reserve, inflation, population growth, and resource rents are good indicators for public private partnership studies.

Table 1. Construction of variables.

Category	Name	Measurement	Data source
Dependent variable of interest	Number	Aggregate number of projects implemented through public private partnership in country i in year t	PPI (2023)
Independent variable of interest	Shortdebt	Ratio of short external debt stock to total external debt stock, both in DOD, current USD	World Bank (2021)
Independent variable of interest for robust check	Shortdebt(ln)	Logarithm value of short external debt stock (in DOD, current USD) to gross capital formation (current USD)	World Bank (2021)
Independent variable of interest	Election	Binary variable takes value of ‘unit' if the existing government face no limitation at all, either in the absence of any provision to the contrary or with a specific provision allowing for unlimited re-election', otherwise. Otherwise stands for cases include limitation on consecutive terms (with no maximum number); limitation through fixed number (two) of possible terms'; limitation through fixed number (two) of possible consecutive terms, and absolute ban on re-election.	The Database of Political Institutions of the Inter-American Development Bank (2020)
Independent variable of interest for robust check	Years in office	Years has the chief executive been in office	The Database of Political Institutions of the Inter-American Development Bank (2020)
Control variable	GDP per capita	GDP per capita, PPP (current international USD)	World Bank (2021)
Control variable	Reserve	Total reserves in months of imports	World Bank (2021)
Control variable	Inflation	Consumer price (annual %)	World Bank (2021)
Control variable	Population	Population growth (annual %)	World Bank (2021)
Control variable	Resource	Total natural resources rents (% of GDP)	World Bank (2021)

Robustness checks are implemented with alternatives of independent variables of interests on different econometric methodology over samples. These approaches help avoid external debt heterogeneity subject to scale and scope of each economy, which can reduce biasedness in regression’s results. Similarly, ‘years in office’ variable can be a good candidate to replace ‘election’ variable for robustness checks more years in office indicate a further deviation of a national institution from international standards, which is a sign of politically corrupted existence and execution (Oluseye, 2024). Details are shown in Table 1.

2.6. Data Sources

Data is from Private Participation in Infrastructure Database of the World Bank at https://ppi.worldbank.org/en/ppi, accessed on March 12, 2023. The second independent variable are also constructed from this databased by aggregation method in R program, then normalized with total external debts. The total external debts data is extracted from the World Development Indicator Dataset provided by World Bank at https://datacatalog.worldbank.org/search/dataset/0037712/World-Development-Indicators, accessed December, 2021. Re-election limitations are from The Database of Political Institutions 2020 of the Inter-American Development Bank, which is available at https://publications.iadb.org/en/database-political-institutions-2020-dpi2020, accessed on February 23, 2023.

3. Empirical Results

3.1. Statistics Analysis

Descriptive statistics in Table 2 show the number of public private partnership projects implemented annually in low-and middle-income countries ranges from 01 to 25. These figures imply complexity and resource investments from public sectors and private sectors to materialize each project. The figures also reaffirm previous findings that each public private partnership project embodies insightful messages on economic and institutional positions (Opara et al., 2017; Xiao & Lam, 2022). Additionally, the maximum value of ‘shortdebt’ variable is 40.9 per cent. This high ratio showcases a heavily external indebtedness in format of short-term loans. This will increase risks towards sovereign solvency as short-term debts are more expensive than other types of debts.

Table 2. Descriptive statistics.

Variable	Obs	Mean	Std. dev.	Min	Max
Number	267	3.108.614	3.540.236	1	25
Shortdebt	235	.1413383	.078333	0	.4096361
Shortdebt(ln)	230	−1.939.071	2.130.128	−18.002	.2827075
GDP per capita	244	11017.42	7.188.073	1.901.717	29564.01
Reserve	234	5.764.138	3.569.553	.8823208	3.379.144
Inflation	241	5.685.705	4.354.177	−1.403.608	2.950.661
Resource	226	4.694.488	4.700.687	.0689198	2.453.419
Population	244	1.316.835	.8283626	-.802164	3.755.687
Election	267	.329588	.4709463	0	1
Years in office	244	5.536.885	6.190.617	1	34

Finally, measurement of ‘years in office’ ranges from 1 to 34 years. This reflects a diverse temporal dimension of power grasping in these countries which can play as a good proxy for investigation institutional impacts on implementation of public private partnership.

To ensure regressions’ efficiency, the authors construct matrix of correlation among independent variables. Table 3 shows that there is not serious multicollinearity among variables.

Table 3. Correlation matrix of independent variables.

	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
Shortdebt	1.0000
Shortdebt(ln)	0.4994	1.0000
GDP per capita	0.3434	0.2674	1.0000
Reserve	0.0631	−0.1017	0.2223	1.0000
Inflation	−0.0311	−0.1160	−0.1210	−0.1600	1.0000
Resource	−0.2299	−0.3664	0.0447	0.2998	0.1633	1.0000
Population	−0.3151	−0.3509	−0.3948	−0.1321	0.2324	0.1552	1.0000
Election	0.0777	−0.0145	−0.1761	−0.0758	0.0610	−0.0121	−0.0836	1.0000
Years in office	−0.1531	−0.0630	−0.0880	−0.0185	0.0499	0.2963	0.3403	0.2711	1.0000

3.2. Baseline Specifications

Authors first employ Poisson regressions on baseline specifications. Then, overdispersion test is performed. The highly significant results and significant difference from zero of ‘alpha value’ which is shown in Appendices indicate the existence of overdispersion of dependent variable. This allows to utilize Negative Binomial regressions. Additionally, as the main dependent variable is non-negative and non-zero, Zero-Inflated Negative Binomial is further employed to ensure regressions’ efficiency. Results show positive and significant impacts of being external indebted on public private partnership implementation. These results are hold across various robustness check with Poisson regressions, Negative Binomial and Zero Truncated Negative Binomial regressions, respectively. When overdispersion is controlled, effects’ sizes increase. These increases are more drastically with all variables when zero truncates are controlled. The results are robust as Log-Likelihood values of Zero-Truncated Negative Binomial Regressions are highest compared to other regression methods.

Among other control variables, the population growth rate has sizable and positive influences on the number of projects implemented via public private partnership. This impact is highly consistent and significant across specifications. The paper finds higher gross domestic products per capita will lead to higher number of public private partnership projects implemented. Detailed results are shown in Table 4.

Table 4. Poisson and negative binomial regression results.

VARIABLES	Annual number of Public Private Partnership Projects
	Poisson	Negative Binomial	Zero-Truncated Negative Binomial
Shortdebt	3.686***	3.925**	6.010**
	(0.970)	(1.540)	(2.402)
Election	−0.564***	−0.568***	−0.947***
	(0.130)	(0.198)	(0.312)
GDP per capita	4.22e05***	4.98e05***	6.97e05***
	(1.16e05)	(1.82e05)	(2.69e05)
Reserve	0.0550**	0.0661**	0.111**
	(0.0221)	(0.0333)	(0.0512)
Inflation	0.0391***	0.0351**	0.0423*
	(0.0104)	(0.0173)	(0.0252)
Resource	−0.0369**	−0.0498*	−0.0881**
	(0.0174)	(0.0264)	(0.0426)
Population	0.439***	0.413***	0.571**
	(0.106)	(0.155)	(0.243)
Constant	−0.0308	0.0634	−0.582
	(0.600)	(0.868)	−1.292
Observations	103	103	103
r2_p	0.221	0.103	0.0964
alpha		0.286	0.715
ll	−260.0	−230.6	−209.4
ll_0	−333.7	−257.0	−231.7
chi2	147.3	52.75	44.69
p	0	8.27e08	2.47e06

Standard errors in parentheses.

*** p < 0.01, ** p < 0.05, * p < 0.1.

3.3. Robustness Check with Alternative Variables of Interest

To test our models’ validity, first-round robustness test is implemented by alternating short-term external debt variable with the logarithm of ratio of short-term external debt stock to gross domestic capital formation. Then, ‘election’ dummy variable is replaced by ‘years in office’. We find that impacts’ sizes increase from Possion regressions to Negative Binomial approachs. In all speficications, the value of Log-Likelyhood of Zero-Truncated Negative Binomial is largest. It confirms better goodness of fit of this econometric application.

The impact sizes of two main independent variables are consistent with the basic specifications. The results prove that higher short-term external debts will lead to higher number of public private partnership projects implemented. Similarly, the longer a government is in office, the fewer numbers of public private partnership projects implemented. Impacts of other control variables are remained in terms of sign, size, and significant levels. Details are displayed in Table 5.

Table 5. Robustness check with alternative main independent variables.

VARIABLES	Annual Number of Public Private Partnership Projects
	Poisson	Negative Binomial	Zero Truncated Negative Binomial	Poisson	Negative Binomial	Zero Truncated Negative Binomial
Shortdebt(ln)	0.255***	0.152*	0.534**
	(0.0967)	(0.0920)	(0.230)
Shortdebt				3.055***	2.955**	4.257*
				(0.985)	(1.475)	(2.267)
Election	−0.335***	−0.350*	−0.496*
	(0.129)	(0.193)	(0.294)
Years in Office				−0.0356***	−0.0423**	−0.0783**
				(0.0132)	(0.0209)	(0.0358)
GDP per capita	4.67e05***	5.76e05***	6.50e05**	3.00e05***	4.52e05**	6.79e05**
	(1.15e05)	(1.80e05)	(2.76e05)	(1.13e05)	(1.88e05)	(2.95e05)
Reserve	0.0756***	0.0906***	0.133***	0.0426*	0.0530	0.0937*
	(0.0210)	(0.0328)	(0.0512)	(0.0228)	(0.0345)	(0.0546)
Inflation	0.0392***	0.0385**	0.0393	0.0444***	0.0382**	0.0478*
	(0.0105)	(0.0175)	(0.0257)	(0.0104)	(0.0176)	(0.0268)
Resource	−0.0361**	−0.0550**	−0.0727	−0.0110	−0.0277	−0.0580
	(0.0176)	(0.0271)	(0.0443)	(0.0182)	(0.0280)	(0.0462)
Population	0.494***	0.462***	0.691***	0.479***	0.514***	0.800***
	(0.108)	(0.160)	(0.252)	(0.104)	(0.159)	(0.262)
Constant	0.501	0.199	0.395	−0.655	−0.637	−1.874
	(0.655)	(0.897)	(1417)	(0.578)	(0.829)	(1274)
Observations	103	103	103	103	103	103
r2_p	0.217	0.0970	0.0975	0.203	0.0951	0.0884
alpha		0.296	0.701		0.315	0.895
ll	−261.2	−232.0	−209.1	−266.0	−232.5	−211.2
ll_c		−261.2	−248.1		−266.0	−259.4
chi2	144.9	49.85	45.20	135.4	48.88	40.95
p	0	2.85e07	2.00e06	0	4.28e07	1.15e05

Standard errors in parentheses.

*** p < 0.01, ** p < 0.05, * p < 0.1.

3.4. Robustness Check with Different Geographical Regions

To control for heterogeneity among countries, further sensitive tests are performed by rerunning basic specifications with different samples, including non-Sub-Saharan African countries, and low-income countries, respectively.

Regarding to external debt variable, our results pass sensitive tests with different samples. Firstly, on the sample excluding Sub-Saharan African countries, effects short-term external debts on public private partnership implementation remain unchanged in terms of sign. However, the size of effect, reported in coefficients, decreases slightly from 3.686 to 3.020 for Poisson regressions; and from 3.925 to 2.980 for Negative Binomial regressions; and from 6.010 to 4.257 for Zero Truncated Negative Binomial regressions. All significance levels are remained. Similarly, when regressing on all middle-income countries, the effects of short-term external debts remain sign. The size change is more moderate. For Poisson regressions, coefficient values increase slightly from 3.686 to 3.707 while those of Negative Binomial regressions decrease slightly from 3.925 to 3.909. For Zero-Truncated Negative Binomial regression, the coefficients decrease slightly from 6.010 to 5.098.

With regards ‘election’ dummy, sensitive tests indicate consistent results. With the non Sub-Saharan African countries, the (absolute value) magnitude of impact of election dummy reduces from 0.564 to 0.459 for Poisson regressions, from 0.568 to 0.463 for Negative Binomial regressions, and from 0.947 to 0.738 for Zero-Truncated Negative Binomial Regressions. These reflect a more heavily dependence on public private partnership in relation to short-term external debt liabilities, and a more nuanced existence of opportunistic behaviors of governments of the Sub-Saharan African countries, compared to its peers in other regions. Additionally, with middle income countries, the impacts of election dummy are consistent in terms of sign. Again, the (absolute value) magnitude of impact reduces from 0.564 to 0.537 for Poisson regressions, from 0.568 to 0.542 for Negative Binomial regressions and from 0.947 to 0.890 for Zero-Truncated Negative Binomial Regressions.

Impacts of other macro-control variables almost changes slightly, which indicate the robustness of variable selections. Details are display in Table 6.

Table 6. Regressions on different sampled datasets.

VARIABLES	Annual number of Public Private Partnership Projects
	Non SSA Sample			Middle Income Sample
	Poisson	Negative Binomial	Zero Truncated Negative Binomial	Poisson	Negative Binomial	Zero Truncated Negative Binomial
Shortdebt	3.020***	2.980**	4.200**	3.708***	3.909**	5.898**
	(1026)	(1434)	(2010)	(0.973)	(1537)	(2370)
Election	−0.459***	−0.463**	−0.738***	−0.537***	−0.542***	−0.890***
	(0.136)	(0.188)	(0.274)	(0.133)	(0.203)	(0.316)
GDP per capita	5.02e05***	5.60e05***	7.49e05***	4.10e05***	4.86e05***	6.74e05**
	(1.27e05)	(1.78e05)	(2.45e05)	(1.17e05)	(1.84e05)	(2.68e05)
Reserve	0.0722***	0.0804**	0.121***	0.0529**	0.0644*	0.107**
	(0.0231)	(0.0319)	(0.0461)	(0.0222)	(0.0335)	(0.0510)
Inflation	0.0377***	0.0342**	0.0391*	0.0393***	0.0353**	0.0425*
	(0.0106)	(0.0158)	(0.0214)	(0.0104)	(0.0174)	(0.0252)
Resource	−0.0397**	−0.0526*	−0.0851**	−0.0334*	−0.0470*	−0.0818*
	(0.0197)	(0.0270)	(0.0410)	(0.0176)	(0.0269)	(0.0428)
Population	0.504***	0.476***	0.641***	0.461***	0.434***	0.612**
	(0.118)	(0.155)	(0.226)	(0.109)	(0.160)	(0.249)
Constant	0.159	0.214	−0.266	−0.102	−0.00173	−0.714
	(0.622)	(0.824)	(1157)	(0.605)	(0.877)	(1298)
Observations	95	95	95	102	102	102
r2_p	0.239	0.119	0.113	0.219	0.101	0.0942
ll	−220.0	−205.5	−189.1	−258.5	−229.3	−208.9
ll_0	−289.2	−233.2	−213.2	−330.9	−255.0	−230.6
chi2	138.3	55.42	48.22	144.8	51.48	43.45
p	0	2.64e08	5.65e07	0	1.42e07	4.14e06

Standard errors in parentheses: *** p < 0.01, ** p < 0.05, * p < 0.1;

Note: SSA = Sub-Saharan African countries.

3.5. Discussions

3.5.1. External Debts and Public Private Partnerhip

Our regressions show positive and significant impacts of external indebtedness on public private partnership implementation. Effects are robust and consistent in sign and significance across variable sampling of low-and middle-income countries with different economic and institutional conditions. As these countries are highly external indebted (Lorenzoni & Werning, 2019; Marcos & Jonathan, 2021; Mitchener & Trebesch, 2021), the findings reveal popular dependency on public private partnership in relation to highly external debts in low-and middle income countries. It is a concrete reflection of a transnational impacts of institutions which explain top-down applications of public private partnership to upgrade and build infrastructures in low-and middle-income countries. This trend would retain its popularity in case such capital shortages still doom these economies.

The findings reveal solvent vulnerabilities and capital-related problems of infrastructure developments in these countries. Highly short-term external indebtedness degrades growth engines, which leads to shortage of financial sources for infrastructure developments. More detrimentally, the high ratio indicates low sovereign credit rates and limited accessibility into international capital markets. In efficient and mature domestic capital markets, the duo impacts can be made up for by domestic creditors. Yet, that is not the case of low- and middle-income countries. Hence, opting for more public private partnership as the last resort will be preferred which is proved in Section 3.3 and Section 3.4.

Additionally, the impacts are more profound when Sub-Saharan countries are included in the research sample. Similarly, removal of low-income countries out of our regressed sample will reduce size of effects of external debts on public private partnership implementation. Heterogeneity in effects indicates higher dependency public private partnership in infrastructure developments in low-income countries and Sub-Saharan African countries which are classified as highly indebted (Coulibaly et al., 2019; Woldu & Szakálné Kanó, 2023).

3.5.2. Re-Election Possibilities and Public Private Partnership

Our results find negative relationships between ‘election’ binary variable on implementation of public private partnership in low-and middle-income countries. Negative sizes indicate the existence of politically opportunistic behaviors of state agencies in these countries. It can be read that when executive term is ending without possibility of reelection, governments are opted for more public private partnership. The paper’s findings are well in line with related conclusions of (Linder, 1999; Roehrich et al., 2014).

In absolute values, sizes of impacts decreases when Sub-Saharan African countries, and low-income countries are respectively removed. The variations showcase high level of political opportunism in these countries which are grouped as under-developed institutionally (Binder et al., 2024; Chigeto et al., 2024; Percoco, 2014). Again, institutional quality plays an fundamental role in healthy public private partnership implementation, and other macro-economic indicators (Arranz & Arroyabe, 2023; Chigeto et al., 2024).

3.5.3. Macro Fundamental Variables and Public Private Partnership

The paper finds positive and significant impacts of gross domestic products per capita, national reserve on implementation of public private partnership. These findings reaffirm supportive roles of positive macro-economic fundamentals to provide more fiscal spaces for governments in search for infrastructure development.

Importantly, our results find highly consistent and significant impacts of population growth rates on public private partnership implementation in low-and middle-income countries. Sensitivity tests show slightly changes in impacts’ sizes across samples which indicate that a current high demand for infrastructure provisions in these countries to meet rapid growth rate of population. It also projects a higher demand for public private partnerhip in these countries in coming years.

4. Conclusions, Theoretical Suggestions, and Policy Recommendations

Public private partnership, a product of inter and transnationally institutional interactions has recently become a widely used policy toolkit in low-and middle-income countries for infrastructure development under budget constraint scenarios. Under low-quality institutional structures which are typical in low-and middle-income countries, public private partnership can be exploited as ways of debt hiding and/or debt transferring by politically opportunistic governments.

The current context of global economy embodied with macroeconomic uncertainties in advanced countries, geo-political tensions in less developed regions, and the retreat of globalization has put sovereign solvency of low-and middle-income countries in danger. Thus, it is advisable for international creditors and local policymakers to follow more considerate approach in infrastructure development plans.

The paper stands as one of pioneering studies to provide evidence of debt hiding behaviors at aggregate national level, rather than provincial level in previous studies. Our paper confirms evident presence of political opportunism of governments in low-and middle-income countries. The presence is stronger in low-income and in Sub-Saharan African countries. This phenomenon provides explanation for these economies’ insolvency and highly indebtedness, and low economic growth rate in a broader spectrum.

Additionally, the paper finds increasing demand and higher dependency on public private partnership in infrastructure development to keep up with population growth in low-and middle-income countries. Hence, it projects a long-lasting popularity of public private partnership in these economies in coming years.

Therefore, the paper’s findings are informative and policy-useful for all related parties in their rationale coordination under public private partnership infrastructure. A thorough reconsideration of the approach in relation with institutional quality and accurate (re)election evaluation is fundamental for all stakeholders to avoid failures in public private partnership implementation in low-income and Sub-Saharan African countries which are projected to remain its popularity in coming years.

The analysis does not examine the relevance of two independent variables of interests on classified types of public private partnership projects. Additionally, the paper does not look at impacts on different types of public private partnership contracts. Thus, future studies can enhance our work by further investigating these topics.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This study is partially funded by the University of Economics Ho Chi Minh City, Vietnam.

Notes on contributors

Tung Nguyen-Thanh

Tung Nguyen-Thanh is postgraduate student of Joint Doctorate Programme (JDP), a joint program between the International Institute of Social Studies (ISS), Erasmus University Rotterdam (EUR) and University of Economics Ho Chi Minh City (UEH). His research focuses on topics such as smart city development and economic development. His research fields are macroeconomics, macro fiscal disciplines and trade at global level and national levels of Vietnam and other ASEAN countries.

Syed Mansoob Murshed

Syed Mansoob Murshed is Professor of the Economics of Peace and Conflict at the International Institute of Social Studies (ISS), Erasmus University in the Netherlands. His research interests are in the economics of conflict, resource abundance, aid conditionality, political economy, macroeconomics and international economics. He was the first holder of the rotating Prince Claus Chair in Development and Equity in 2003. He was a Research Fellow at UNU/WIDER in Helsinki where he ran projects on Globalization and Vulnerable Economies and Why Some Countries Avoid Conflict, While Others Fail. He also ran a project on The Two Economies of Ireland, financed by the International Fund for Ireland at the Northern Ireland Economic Research Centre (NIERC), Belfast. He is the author of ten books and over 170 refereed journal papers and book chapters. His latest monograph, published in 2010, is Explaining Civil War (Edward Elgar). He is on the Editorial boards of Peace Economics, Peace Science and Public Policy (PEPS)Opens external, as well as Civil WarsOpens external. He is an associate Editor of the Journal of Peace Research (JPR).

Hoai Nguyen-Trong

Hoai Nguyen-Trong is Professor at University of Economics Ho Chi Minh City (UEH), and a member of the Editorial Advisory Board of the Journal of Asian Economic and Economic Studies (JABES) of UEH. His research focuses on topics such as sustainable development, educational internationalization, green growth, green behavior, public health, and smart city and regional development. Currently, he is the Vice Rector of UEH, in charge of international relations and academic research. At the same time, he was also nominated by the Vietnam National University Ho Chi Minh City (VNU-HCMC) to be a member of the Education Accreditation Center. In addition, he is the Director of the Vietnam – Netherlands Master of Applied Economics, a joint program between UEH and the International Institute for Social Studies at Erasmus University Rotterdam.

Appendices

Figure A1. Histogram of number of projects variable.

Table A1. List of sampled countries.

No.	Country Name	ISO	Low-income	Lower middle income	Upper middle income
1	Albania	ALB	0	0	1
2	Algeria	DZA	0	1	0
3	Armenia	ARM	0	0	1
4	Bangladesh	BGD	0	1	0
5	Bulgaria	BGR	0	0	1
6	Cambodia	KHM	0	1	0
7	Colombia	COL	0	0	1
8	Costa Rica	CRI	0	0	1
9	Côte d'Ivoire	CIV	0	1	0
10	Dominican Republic	DOM	0	0	1
11	Ecuador	ECU	0	0	1
12	Egypt, Arab Rep.	EGY	0	1	0
13	El Salvador	SLV	0	1	0
14	Georgia	GEO	0	0	1
15	Ghana	GHA	0	1	0
16	Guatemala	GTM	0	0	1
17	Honduras	HND	0	1	0
18	Indonesia	IDN	0	1	0
19	Jordan	JOR	0	0	1
20	Kazakhstan	KAZ	0	0	1
21	Kenya	KEN	0	1	0
22	Lao PDR	LAO	0	1	0
23	Malaysia	MYS	0	0	1
24	Mexico	MEX	0	0	1
25	Morocco	MAR	0	1	0
26	Nepal	NPL	0	1	0
27	Nigeria	NGA	0	1	0
28	Pakistan	PAK	0	1	0
29	Peru	PER	0	0	1
30	Philippines	PHL	0	1	0
31	Russian Federation	RUS	0	0	1
32	Senegal	SEN	0	1	0
33	South Africa	ZAF	0	0	1
34	Sri Lanka	LKA	0	1	0
35	Tanzania	TZA	0	1	0
36	Thailand	THA	0	0	1
37	Turkey	TUR	0	0	1
38	Uganda	UGA	1	0	0
39	Ukraine	UKR	0	1	0
40	Vietnam	VNM	0	1	0
	Total		1	21	18

Table A2. Classification of reelection limitation.

Code	Type of Limitations for the chief executive to have their ability of being reelected?
1	No limitation at all, either in the absence of any provision to the contrary or with a specific provision allowing for unlimited re-election
2	Limitation on consecutive terms (with no maximum number)
3	Limitation through fixed number (two) of possible terms
4	Limitation through fixed number (two) of possible consecutive terms
5	Absolute ban on re-election
6*	Limitation through fixed number (three) of possible consecutive terms
7*	Re-election possible after sitting out two consecutive terms
8*	Limitation on consecutive terms (with fixed number = 2)

Note: * there is only one country under this category.

Source: The Inter-American Development Bank (2020).